Perspectives in Human Growth

PERSPECTIVES IN HUMAN GROWTH,
DEVELOPMENT AND MATURATION
A tribute to Professor Sudhir Ranjan Das

PERSPECTIVES IN HUMAN
GROWTH, DEVELOPMENT
AND MATURATION
edited by
Parasmani Dasgupta
Anthroplogy and Human Genetics Unit, Indian Statistical Institute,
Calcutta, India
and
Roland Hauspie
Laboratory of Anthropogenetics, Free University Brussels, Belgium
Springer-Science+Business Media, B.Y.

A C.I.P. Catalogue record for this book is available from the Library of Congress.
ISBN 978-90-481-5820-1 ISBN 978-94-015-9801-9 (eBook)

DOl 10.1007/978-94-015-9801-9
Printed on acid-free paper
All Rights Reserved

© 2001 Springer Science+Business Media Dordrecht
Originally published by Kluwer Academic Publishers in 2001.
Softcover reprint of the hardcover 1st edition 2001

No part of the material protected by this copyright notice may be reproduced or
utilized in any form or by any means, electronic or mechanical,
including photocopying, recording, or by any information storage and
retrieval system, without written permission from the copyright owner.
TABLE OF CONTENTS
List of contributors .................................................................................. .ix
Foreword .................................................................................. xiii
Editors note ........................................................................................... xv
Methodological Aspects of Growth Studies ................................................. 1
The Measurement of Human Growth: A Historical Review

L.D. Voss ........................................................................................... 3
2 Individual Physical Growth Models and Biological Parameters of

Japanese
T. Shohoji, T. Sumiya ..................................................................... 17
3 Smoothing Centile Curves of Height of Basque Boys and Girls by

the Application of the LMS-Method
J. Rosique, L. San Martin, J.R. Fernandez-L6pez, I. Salces, E.
Rebato, A. Vinagre, C. Susanne ......................................................... 33
4 A New International Growth Reference for Young Children

M. de Onis, C.G. Victora, C. Garza, E.A.J.R. Frongillo, T. Cole ........... .45
5 Fels and Tanner-Whitehouse Skeletal Ages of School Children 7-13

Years in Oaxaca, Mexico
M.E. Pella Reyes, R.M. Malina ......................................................... 55
6 Assessment of Childhood and Adolescents Body Composition: A

Practical Guide
A. Pietrobelli, M. Heo, M.S. Faith ..................................................... 67
Genetic and Environmental Factors ......................................................... 77
7 Genetics of Complex Traits with Particular Attention to Fat

Patterning
D.C. Rao, P.P. Majumder ................................................................. 79
8 Subcutaeneous Adipose Tissue Distribution in 7- to 16-Year Old

Boys of Calcutta in relation to socio-economic level
E. Rebato, J. Rosique, A. Pietrobelli, M. Chatterjee, S. Chatterjee,
R. Saha, P. Dasgupta ....................................................................... 91
v
vi
9 Dentofacial Morphology, Growth and Genetics: A Study of

Australian Aborigines
T. Brown, G. Townsend ................................................................. 109
10 Riddles in Human Development Patterns: a Few Family Case Study

R.D. Singh ................................................................................... 123
11 Nutrition in Venezuela at the End of the Millennium

M. L6pez-Blanco ........................................................................... 129
12 Thoughts on Secular Trends in Growth and Development

P.B. Eveleth ................................................................................. 137
13 Secular Trends and Longterm Serial Growth Studies

A.F. Roche, B. Towne ................................................................... 147
14 Secular Trends in Body Height - Indicator of General Improvement

in Living Conditions or of a Change in Specific Factors?
M. Henneberg ............................................................................... 159
15 Patterns of Growth and Scholastic Performance - A Case Study of

the Associated Factors
K.V. Rao, N. Balakrishna, S. Veena ................................................. 169
16 Mental Ability and Cognitive Thinking in Relation to Sex,

Pubertal Stage and Socio-Economic Background
G. Tanner-Lindgren ........................................................................ 179
Population Differences in Growth ...................................................... ... 203
17 How Genetic Are Human Body Proportions?

B. Bogin, M. Kapell, M.1. Varela Silva, B. Orden, P.K. Smith, J.
Loucky ........................................................................................ 205
18 The Sarsuna-Barisha Longitudinal Growth Study: Dynamics of

Growth
R.C. Hauspie, P. Dasgupta .............................................................. 223
19 Linear Components of Growth Among Rural Indian Children

S. Rao ......................................................................................... 237
20 Growth Stunting among Children, Aged Birth to 5 Years, in Peri-Urban

Kathmandu, Nepal
T. Moffat ...................................................................................... 251
Vll
21 Ethnic and Sex Differences in the Skelic Index among Fijian and
Samoan Children
T. Satake, K. Hattori, E. Kanazawa ................................................... 269
22 Skeletal Maturity in Children of Mixed American and Japanese

Parentage as Assessed by the TW2-Method
K. Kimura .................................................................................... 281
Biological Aspects of Growth ................................................................ 299
23 Somatotypes of Budapest Children

O.G. Eiben, A. Nemeth .................................................................. 301
24 Differential Rate of Growth of the Human Body Parts

M. Prokopec ................................................................................. 313
25 Short-term Growth
M. Hermanussen 321
26 Relations between Head Growth and Psychomotor Development in

Low-Birth-Weight Children
W. Furmaga-Jablonska, H. Chrzastek-Spruch, M. Kozlowska, A.
Orzechowski ................................................................................. 333
27 Growth in Length and Weight of Thalassemic Children in West

Bengal, India: A Mixed-Longitudinal Study
J. Banerjee, S. Saha, I. Dasgupta, D.K. Bhattacharyya, P. Dasgupta
................................................................................................... 341
Appendices ......................................................................................... 351
Life sketch of Sudhir Ranjan Das ...................................................... 353
List of publications of Sudhir Ranjan Das 357
Index ................................................................................................. 363

LIST OF CONTRIBUTORS
Balakrishna N., National Institute of Nutrition, Indian Council of Medical Research,

Hyderabad, India.
Banerjee 1., Vivekananda Institute of Medical Sciences, Ramakrishna Mission Seva
Pratishthan, Calcutta, India.
Bhattacharyya D.K., Vivekananda Institute of Medical Sciences, Ramakrishna
Mission Seva Pratisthan, Calcutta, India.
Bogin B., Department of Behavioral Sciences, University of Michigan-Dearborn,
Dearborn, USA.
Brown T., Dental School, The University of Adelaide, Adelaide, Australia.
Chatterjee M., Anthropological Survey of India, Calcutta, India.
Chatterjee S., Anthropological Survey of India, Calcutta, India.
Chrzastek-Spruch H., Department of Paediatrics Propedeutics, Medical University
School of Lublin, Lublin, Poland.
Cole T., Paediatric Epidemiology & Biostatistics, Institute of Child Health,
London, United Kingdom.
Dasgupta I., Thalassemia Unit, Vivekananda Institute of Medical Sciences,
Ramakrishna Mission Seva Pratisthan, Calcutta, India.
Dasgupta P., Indian Statistical Institute, Anthropology & Human Genetics Unit,
Calcutta, India.
de Onis M., Department of Nutrition for Health & Development, World Health
Organisation, Geneva, Switzerland.
Eiben 0., Department of Physical Anthropology, Eotvos Lonind University,
Budapest, Hungary.
Eveleth P.B., GSE Associates, Inc., Newburg, Maryland, USA.
Faith M.S., Obesity Research Center, St. Luke's/Roosevelt Hospital, Columbia
University, College of Physicians and Surgeons, New York, USA.
Fernandez-Lopez J.R., Facultad de Ciencias, Dpto. Biologia Animal y Genetica,
Universidad del Pais Vasco, Bilbao, Spain.
Frongillo E., Division of Nutritional Sciences, B 17 Savage Hall, Cornell
University, Ithaca, New York, USA.
Furmaga-Jablonska W., Department of Paediatrics Propedeutics, Medical University
Garza C., Division of Nutritional Sciences, B 17 Savage Hall, Cornell University,
Ithaca, New York, USA.
Hattori K., Department of Health and Physical Education, Ibaraki University, Mito
Ibaraki, Japan.
Hauspie R.C. Laboratory Anthropogenetics, Free University of Brussels, Brussels,
Belgium.
Henneberg M., Department of Anatomical Sciences, University of Adelaide Medical
School, Adelaide, Australia.
Heo M., Obesity Research Center, St. Luke's/Roosevelt Hospital, Columbia
University, College of Physicians and Surgeons, New York, USA.
Hermanussen M., Altenhof, Germany.
ix
x
Kanazawa E., Department of Anatomy, Nihon University School of Dentistry at

Matsudo, Matsudo, Chiba, Japan.
Kapell M., Department of Behavioral Sciences, University of Michigan-Dearborn,
Dearborn, USA.
Kimura K., Kimura Auxological Institute, Sayama, Japan.
Kozlowska M., Department of Paediatrics Propedeutics, Medical University School
of Lublin, Lublin, Poland.
Lopez-Blanco M., "Jose Maria Bengoa", Fundacion para la Alimentacion y
Nutricion, Caracas, Venezuela.
Loucky J., Department of Anthropology, Western Washington University,
Bellingham, USA.
Majumder P.P., Anthropology and Human Genetics Unit, Indian Statistical
Institute, Calcutta, India.
Malina R.M., Michigan State University, Department of Kinesiology, East
Lansing, USA.
Moffat T., Department of Anthropology, McMaster University, Hamilton, Ontario,
Canada.
Nemeth A.N., National Centre of Public Health, Budapest, Hungary.
Orden B., Centro de Investigaciones en Genetica Basica y Aplicada (CIGEBA),
Universidad Nacional de La Plata, La Plata, Argentina.
Orzechowski A., Department of Paediatrics Propedeutics, Medical University
Pella Reyes M.E., escuela Nacional de Antropologia e History, Mexico City,
Mexico.
Pietrobelli A., Pediatric Unit, Policlinico "GB. Rossi", Verona University
Hospital, Verona, Italy.
Prokopec M., National Institute of Public Health, Centre for Health and Living
Conditions, Praha, Czech Republic.
Rao D.C., Division of Biostatistics, Washington University School of Medicine,
St. Louis, USA.
Rao K.V., National Institute of Nutrition, Indian Council of Medical Research,
Hyderabad, India.
Rao S., Incharge Biometry & Nutrition Group, Agharkar Research Institute, Pune,
India.
Rebato E., Facultad de Ciencias, Dpto. Biologia Animal y Genetica, Universidad
del Pais Vasco, Bilbao, Spain.
Roche A.F., Lifespan Health Research Center, Department of Community Health,
Wright State University School of Medicine, Kettering, Ohio, USA.
Rosique J., Facultad de Ciencias, Dpto. Biologia Animal y Genetica, Universidad
Saha R., Indian Statistical Institute, Anthropology & Human Genetics Unit,
Calcutta, India.
Saha S., Indian Statistical Institute, Applied Statistics Unit, Calcutta, India.
Salces I., Facultad de Ciencias, Dpto. Biologia Animal y Genetica, Universidad del
Pais Vasco, Bilbao, Spain.
San Martin L., Facultad de Ciencias, Dpto. Biologia Animal y Genetica,
Universidad del Pais Vasco, Bilbao, Spain.
xi
Satake T., Department of Anatomy, Nihon University School of Dentistry at

Matsudo, Matsudo, Chiba, Japan.
Shohoji T., Faculty of Human Culture, Fukuyama University, Fukuyama, Japan.
Singh R.D., Department of Sociology and Anthropology, University of Windsor,
Windsor Ontario, Canada.
Smith P.K., Department of Behavioral Sciences, University of Michigan-Dearborn,
Dearborn, USA.
Sumiya T., Fac. of Integrated Arts & Sciences, Hiroshima University, Higashi-
Hiroshima, Japan.
Susanne C., Laboratory Anthropogenetics, Free University of Brussels, Brussels,
Belgium.
Tanner J.M., Stentwood Auxological Consultants, Stenwood Coach House,
Honiton, Devon, United Kingdom.
Tanner-Lindgren G., Stentwood Auxological Consultants, Stenwood Coach House,
Honiton, Devon, United Kingdom.
Towne B., Lifespan Health Research Center, Department of Community Health,
Wright State University School of Medicine, Kettering, Ohio, USA.
Townsend G., Dental School, The University of Adelaide, Adelaide, Australia.
Varela Silva M.I., Department of Behavioral Sciences, University of Michigan-
Dearborn, Dearborn, USA.
Veena S., National Institute of Nutrition, Indian Council of Medical Research,
Hyderabad, India.
Victora C.G., Universidade Federal de Pelotas, Pelotas, Brasil.
Vinagre A., Facultad de Ciencias, Dpto. Biologia Animal y Genetica, Universidad
Voss L., Early Bird Research Centre, Derriford Hospital, Plymouth, United
Kingdom.
FOREWORD
One morning in 1969, out of the blue, I received a letter which both distressed and
astonished me. It was from a Prof. S.R. Das in Calcutta, who requested me to
accept, for eventual analysis, a mountain of anthropometric data he had accumulated,
as he was ill and did not expect to survive to analyse it himself. The data provided
the astonishment; twenty-two anthropometric characters recorded every six months
or a year, over a period of 14 years, in a mixed longitudinal study of some 560
children, aged six months to twenty years. Most were in families with siblings also
in the study, and every child was measured every time by S.R. Das himself. The
archive was unique, combining the personal anthropometry of R.H. Whitehouse in
the Harpenden Growth Study and the family approach of the Fels Growth Study.
This was a study of which neither I, nor anyone of my acquaintance, had heard.
Even in India, Prof. Das' work was scarcely known. It turned out Das was a
scholarly man, quiet and unassuming, absolutely committed to his Sarsuna-Barisha
Growth Study,just the obverse of the professional showman.
Clearly this was not a request I could refuse, although I already had in hand
enough projects to occupy Siva himself. So, when the packing cases of data sheets
(not even punched cards) arrived, I put them, not in a storage cupboard, but under
the long table in my small office, where I saw them every time I looked up from my
desk, and where visitors barked their knees on them when they ate their sandwiches.
I felt more and more guilty at their neglect, but reality was reality, the impossible
remained impossible. Finally, a young human biologist from Brussels obtained a
Fellowship to work with me for a year. He brought no particular project of his own,
and the association of Roland Hauspie and Sudhir Ranjan Das began, for in the
meantime Das had recovered, and was working in official retirement, at the Indian
Statistical Institute. He would live to be nearly 90.
Thus history: a story of near-universal neglect of the work of a truly dedicated
man until late in life, a story perhaps more familiar in the Arts than in Sciences.
Now, however, we have this volume dedicated to Professor Das' memory, edited by
Dr. Parasmani Dasgupta, S.R. Das' former postdoctoral student, and Dr. Roland
Hauspie. It contains no less than 27 contributions, covering the whole of Human
Auxology. It is highly international in the provenance of both authors and subjects;
from Kathmandu to Caracas, Oaxaca to Alice Springs. There are papers on the
history of Auxology, on the modelling of the individual growth curve, the
construction of population growth references, growth as a measure of population
well-being, secular trend, and the much neglected subject of the relation between
mental and physical development. And, as the advertisements always say, much
more.
I salute the two editors on the successful completion what was clearly a labour of
love, and I commend this book most warmly to what I hope will be a wide circle of
readers amongst, Anthropologists, Educationists, Human Biologists and Paedia-
tricians around the world.
I.M. Tanner
xiii
EDITORS NOTE
Weare delighted to introduce this Volume on Auxology as a mark of homage to

Prof. Sudhir Ranjan Das, the pioneer of Auxology in India. Both of us, have
enjoyed associating ourselves with him for the last twenty years as his student and
his collaborator. During this time, through our academic interaction with him, we
have greatly benefited both in our careers and in our understanding.
S.R. Das may unquestionably be regarded as the father of Indian Auxology. He
graduated with Physics major (M.Sc. in Physics) and he contributed immensely to
most of the major branches of Physical Anthropology. Such as Franz Boas he
strongly believed in publishing raw data so others could use them to further
knowledge: a practice in line with Tanner's (1959) comments "with Boas'
unbounded regards for scientific integrity and the ethics of research". Further, S.R.
Das contributed significantly in the field of Human Genetics and Human Biology of
Indian populations.
The Sarsuna-Barisha Growth Study, the first Indian Longitudinal Growth Study
from 6 months to 20 years (22 measures, initiated in 1952), which brought fame,
recognition and appreciation, has been described as one of the rarest studies of the
world, because it was a family-based longitudinal study and the measurements were
taken by a single person (S.R. Das), continued for fourteen years, sometimes only
achieved previously by Paul Godin in France and R.H. Whitehouse in England. We
are extremely fortunate that S.R. Das gave us access to this precious data set which
has resulted in a considerable number of publications. However the data set remains
to be fully explored. Although this pioneering work is not as well known in the
auxological and paediatric worlds as we think it should be, it is indeed a bright star
in the Annals ofIndian Auxology.
Unfortunately, information about his diverse research activities remain unknown
to the Anthropologists, during his lifetime, as he had mentioned in a letter to a
publisher in Canada, who were compiling a directory of the Indian Anthropologists.
Such a serious lacuna may have resulted from his unassuming nature and
personality. Throughout his life he gave more importance to the dedication of his
work than to publicity and recognition. Unusually, just two years before his death,
he agreed to accept the felicitation offered to him by the Indian Academy of
Paediatrics, West Bengal chapter, for his seminal contribution in Indian Auxology.
The event was held in the Nilratan Sircar Medical College Hospital, Calcutta during
January 1997.
After his death in 1998 we decided to extend our tributes to him by drawing the
attention of the international scientific community to his significant contributions in
the field of Indian Auxology. This Volume is the result.
We extend our gratefulness to all authors for their overwhelming response to
contribute. We are grateful to J.M. Tanner, for writing the Foreword. Our special
thanks are due to M. Henneberg, without whose support in all stages of the work,
the Volume could not have been released.
xv
xvi
Our heartfelt thanks go also to P. De Liefde, the Publishing Director, Peter

Butler and Manja Fredriksz of the Biomedical Unit of Kluwer Academic Publishers
B.V. for showing keen interest in publishing this Volume.
We thank K. Ashizawa, T. Bielicki, R. Martorell, M.-F. Cachera and M.
Takaishi for providing support, and reviewing the submitted manuscripts.
The following Colleagues have extended their full support and co-operation in
various stages of the work: R. Bardhan, L. Bose, S. Chatterjee, M.K. Das, S.K.
Das, A.B. Daschaudhuri, S. Dasgupta, N. Ganguly, R. Gupta, K.B. Hossain, B.
Karmakar, P.P. Majumder, K.C. Malhotra, B.N. Mukherjee, M. Roy, R. Saha, D.
Sengupta, and J. Verghese. We are thankful to them.
Finally we wish to express our deep regards to Mrs. Sushama Das and B.R.
Das, widow and son respectively of late S.R. Das, for their enthusiasm and co-
operation in all stages of work to make this project a success.
Weare sure that the book will be of interest to the students, researchers and the
practitioners in the field of Physical Anthropology, Auxology, Public Health,
Human Nutrition, Paediatrics and Body Composition.
P. Dasgupta and R.C. Hauspie

METHODOLOGICAL ASPECTS OF
GROWTH STUDIES
CHAPTER 1
THE MEASUREMENT OF HUMAN GROWTH: A

HISTORICAL REVIEW
L.D. voss
Early Bird Research Centre, Derriford Hospital, Plymouth, United Kingdom
1. THE ANCIENT WORLD
There has long been an interest in anthropometry - the science of human

measurement. It is an interest that has come from many sources and its origins can
be traced back to the earliest times. Whether the ancient Egyptians routinely
measured body height is not known, but they do appear to have measured kings of
the later dynasties, if only for practical reasons concerning burial arrangements (Boyd
1981). Sadly, the Greek historian Diodorus who came across some of these
measurements, failed to record the details, being of the mistaken opinion that they
would be of little interest to posterity. Absolute dimensions would, however, have
been difficult to interpret today, in view of the anatomical units of measurement
employed: the cubit, hand and fingers. These were also used as a measure by the
ancient Hindus, as was the 'tala', or handspan, roughly equivalent to the length of
the face. The well-proportioned Hindu male measured eight such units, and their
sculptors were expected to adhere to these very strict guidelines when representing
the human body. Extremes of any kind in the living body were undesirable, whether
of weight, hair, skin colour or stature. Good health was also just a mean between
excesses and deficiencies (Boyd 1981).
Galen, an authority on medical matters in both Greece and Rome, was clearly of
the same opinion, ' ..if the exact mean of all the extremes were in all parts of the
body, this would be the best to observe as being the symmetry most suitable for all
activities' (Boyd 1981). Tiberius, it seems, was of above average stature, well
proportioned, and indeed 'symmetrical from head to foot' (Suetonius trans Rolfe
1928). It is of no surprise to learn that he enjoyed excellent health. In contrast,
CaJigula was very tall, but evidently regarded as disproportional, having a 'huge
body, but very thin neck and legs'. Predictably, he was 'sound neither of body nor
mind'.
The earliest surviving statement about the stages of growth is attributed to the
Greek statesman, Solon: he divided the human lifespan into seven parts or
hebdomads, puberty being the third. His fellow countryman, Hippocrates, warned
that puberty would be delayed in areas where water was hard and the wind cold, and
that final adult stature would also depend on geographical location (Tanner 1981).
3
P. Dasgupta and R. Hauspie (eds.), Perspectives in Human Growth, Development and Maturation, 3-15.
© 2001 Kluwer Academic Publishers.
4 CHAPTER 1
Aristotle was interested in the mechanism of growth and offered two possible
reasons why growth might be stunted - violent exercise or intercourse before puberty
(Tanner 1981). Likewise, the tall muscular youths described in the Gallic wars were,
in Caesar's opinion, the result of a chaste adolescence! Around this time, the Chinese
philosopher Wang Ch'ung produced an intriguing explanation for height differences
between individuals. It was, he said, due to variation in vital fluid - only when he
had the full amount could man reach his proper height of ten feet and live to one
hundred years. More often than not, some of this fluid was lacking, resulting in an
average height of only six to seven feet (presumably Chinese feet), and a much
reduced lifespan (Boyd 1981).
In contrast, the Romans, and the Greeks before them, must have had at least
some basic knowledge about factors affecting growth. They would deliberately stunt
the growth of slave children so as to fetch a higher price in the market (Tietze-Conrat
1957). There are no records of the measuring instruments used, but the historian
Suetonius, gives detailed descriptions of the shape and height of several historical
figures (Suetonius trans. Rolfe 1928). Augustus was five feet nine inches (a Roman
measure - a little under five feet seven inches in our terms), yet is judged to have
been 'short of stature'. Indeed, his shoes were 'somewhat high-soled to make him
look taller than he really was'. His short stature, however, was noticeable only by
comparison with some taller person standing beside him' Caesar, on the other hand,
was 'tall of stature' with 'shapely limbs'. There was a persistent belief in those days
that certain races had once been much taller and stronger. Tacitus described the
Germanic tribes of old as tall, red-blond and blue-eyed,· and in his account of the
Gallic war, Caesar still found much to admire in the 'noble savage' existence of the
German youth.
2. THE MIDDLE AGES

In medieval times, there was little, if any, advancement in the scientific
understanding of growth. Children were of no medical interest - they just lived or
died, their mothers having recourse only to home spun remedies (Rhodes 1985).
Physicians were chiefly interested in the age of menarche, having obvious
implications for a young girl's marriage prospects (Tanner 1981). As to the
development of the unborn child, illustrations of the foetus at various stages clearly
owed more to the imagination than to observation, and some bizarre illustrations of
homunculi or 'little men' appeared (Boyd 1981). Only in the middle of the 17th
century did the physician William Harvey demonstrate that the embryo is not a
preformed adult, but has to pass through various stages in its development.
3. THE RENAISSANCE
This period did much for art but little for scientific progress in the study of growth
and development. Scholars rediscovered the ancient Greek texts, and the invention of
printing merely served to spread their largely inaccurate ideas. The preoccupation
with the seven ages of man continued, likewise the four elements: earth, air, fire and
water, thought to be present in all matter. Nevertheless, by Shakespeare's time, it
was known that certain plants had the effect of stunting growth, thus satisfying the
THE MEASUREMENT OF HUMAN GROWTH 5
demand for dwarfs in the royal courts and travelling fairs of Renaissance Europe
(Kelnar 1990). Leonardo da Vinci produced the first accurate drawings of the foetus
and he, together with other artists, began to represent infants and children in a more
sympathetic and realistic manner. In Velasquez' painting of Las Meninas, one can
even distinguish, on the same canvas, a normal child, a hypopituitary dwarf and an
achondroplasic dwarf. Sir Francis Bacon, the 16th century philosopher, also
attempted a more scientific approach to the subject of growth, maintaining that
satisfactory growth required three things: adequate but not excessive nutrition, the
right kind of food (that is, not too dry) and natural heat. This natural heat could be
stimulated by exercise, and therefore 'much Going to Schoole', he wrote, 'where
they sit so much, hindered the Growth of Children' (Tanner 1981).
4. EARLY MEASURING INSTRUMENTS

The earliest record of a specially designed measuring instrument appears in the 15th
century (Tanner 1981). The 'finitorium', as the device was called, was based on the
astrolab but its function, it would seem, was to measure proportion rather than
absolute height, still considered largely irrelevant. The Renaissance had brought with
it a renewed interest in Classical sculpture and the instrument was essentially an aid
to statue making. Some two centuries later, the German Elscholz designed a similar
kind of ruler, but the sole purpose of this 'anthropometron', as it was known, was to
detect disproportion. As a medical man, Elscholz saw it as a useful diagnostic tool
that would differentiate between the sick and the healthy body - the healthy body
alone having perfect symmetry. 'Anthropometria' was the title of his thesis, and he
is almost certainly the first to have used the term in its present-day meaning (Boyd
1981). The need for around eighty different bodily measurements, however, as
outlined in his dissertation, must have rendered his protocol quite impractical, and
one doubts whether it was ever used.
5. THE EIGHTEENTH CENTURY

The mechanism of growth was still frequently misunderstood in the eighteenth
century. The assumption that illness was a prerequisite for growth, arose from
observation of the rapid 'catch-up' growth that can follow a period of poor health
(Prader, Tanner and von Harnack 1963). According to one physician, it was due to
' .. the force of the fever impelling the humours and by that means stretching out and
lengthening all the vessels, whilst the bones at the same time, not having yet
acquired their full hardness, easily yield to the same impressions' (Tanner 1981). As
to why growth was faster in some children than others, the Frenchman Virey felt
nutrition to be the key. 'Nourish a man ... with dry and hard foods, smoked, salted,
spiced or astringent ..it is quite clear that such a person will become thin, short,
compact in all his organs. In contrast, stuff him from childhood with soggy foods ...
milk and gruel and slimy drinks like beer, mead, whey and greasy chocolate ... then
he can become colossal and gigantic in stature .. .' (Virey 1816).
6 CHAPTER 1
5.1 Military data
The real impetus behind the measurement of absolute height, came from the
military, and the need for a powerful fighting force. A contemporary illustration of
recruits being measured for the Duke of Sachsen-Weimar's army in 1779, shows,
interestingly, a remarkably modem looking stadiometer together with faultless
measuring technique (Tanner 1981). Data from recruits allows the modem researcher
both to study secular changes in height, and to look for environmental factors
associated with those changes. Records show clearly that the young, mainly working
class boys, measured on entering the navy from 1770 to 1870 (solely for purposes of
identification, should they desert) were extremely small by today's standards, and
nearly six inches shorter than upper class recruits attending a military college (Floud,
Wachter and Gregory 1990).
As yet, the idea that growth data might be used for the purposes of social reform
had not arisen. The 18th century, known as the century of the Enlightenment, was
characterised, however, by a real attempt to replace mere speculation about the
physical world with observation and measurement, albeit with varying degrees of
scientific rigour. One such case concerned some soldiers who had been discharged for
being too short. A clergymen, curious as to why they had apparently shrunk,
suspected the cause to be diurnal variation in height. The ultimate proof was
presented to a meeting of the Royal Society, ' I tried myself.. and found it in like
manner... at Eleven in the Morning I sat down, and fixed an Iron Pin so as to touch
it.... After that, I fatigued myself for half an Hour with a Garden-Roller, and the
Consequence was, that at 12 Ho. 30 Min. I could not reach the Nail sitting, by
about 5 Tenths of an Inch ... '. He also confessed that he had once measured his horse
before and after riding 20 Miles, but' could not perceive the least difference in her
height.' (Wasse 1724).
5.2 Early growth studies

Interest in early life events, especially birth weight, began to burgeon. In 1749,
Buffon published the results of measurements (in 'pouces') that he had made on
aborted foetuses and neonates (Tanner 1981). At around the same period, a German
obstetrician was making meticulous measurements of lengths and weights of a
number of neonates, with the aim of producing an index of immaturity (Tanner
1981). A century later, however, Chaussier was measuring new-borns to assess their
viability, concerned not primarily for the child, but to establish the CUlpability of
mothers who had committed infanticide (Tanner 1981). When, in 1754, Jampert
published a cross-sectional table of measurements on a series of young orphans, it
was the first of its kind. At each age, however, only one 'typical' subject was
measured. Conscious of these limitations, Jampert did, nevertheless, collect his data
with great care as they formed the basis of an academic thesis (Tanner 1981).
The earliest, and indeed most famous, longitudinal record of growth was kept
between 1759 and 1777, when in the spirit of pure scientific enquiry, the French
aristocrat, the Compte de Montbeillard determined to measure his own son from
birth to maturity. Although recorded in 'pieds', 'pouces', and 'lignes', the data, if
plotted, is indistinguishable from any modern day growth chart (Scammon 1927).
Seasonal variations in growth were noted as well as diurnal variation in height. After
one particular all night party, the young man was observed to have shrunk (Boyd
1929). In view of the imprecision inherent in the measurement of height, however
(Voss, et at. 1990, Voss, et at. 1991, Voss and Bailey 1997), one suspects that the
observation was quite fortuitous.
6. THE NINETEENTH CENTURY

Doctors who applied any kind of measurement to the sick were once regarded as
'quacks' by their colleagues, but by the beginning of the 19th century, the taking of
pulse and temperature, were routine. Indeed, the 19th century was characterised by a
rapidly growing trend, sweeping all human sciences, to measure and to quantify. The
consequences were far reaching. As soon as it became possible to apply physical
measurements to them, diseases became real (Illich 1976). 'When you can measure
what you are speaking about', said Lord Kelvin,' and express it in numbers, you
know something about it; but when you cannot measure it, when you cannot
express it in numbers, your knowledge of it is of a meagre and unsatisfactory kind'
(Kelvin 1891).
6.1 Statistics and the quantification of society

In order to sort and interpret large amounts of numerical data, a new branch of
mathematics was required, and it was the early statisticians who brought a new
dimension to the study of growth. Quetelet recognised the importance of measuring
samples, rather than individuals, to look at human variation and was the first to use
the bell shaped curve to describe the distribution of height measurements (Bogin
1988). He went on to postulate the notion of 'l'homme moyen,' or the 'average
man', to include not only physical, but moral and intellectual attributes. 'L'homme
moyen moral' was to be derived from records of various criminal acts, acts of
courage and kindness and so on. Dividing these by the population would allow a
'penchant' for each attribute to be assigned to the 'average' man. 'L'homme moyen
physique' was derived by more conventional means, from a series of anthropometric
measures such as height, weight and dynamometric force, thus marking a turning
point in the history of anthropometry (Bennett et at. 1994). Galton took the
statistical treatment of man still further, producing a scattergram of the heights of
men - a graphic illustration of the principle of deviation from an average. He was
also the first to use centiles, another simple means of relating an individual to the
population (Tanner 1981). It is a measure of Galton's obsession that he set up
anthropometric booths in public places to collect data from passers-by (Hacking
1990). His guiding maxim was, 'Whenever you can, measure or count' (Gregory
1987).
Alphonse Bertillon, chief of Police in Paris in the late 19th century, put
anthropometry to a most original use. He developed a unique system of establishing
the identity of suspects, based on the supposed unchanging character of the human
frame. This method, known as 'Bertillonage', was adopted first by the Parisian
police in their fight against rising crime, and soon by most civilised countries.
8 CHAPTER 1
(Encyclopedia Britannica 1926). Photographs were unreliable - people could change

their appearance - but the skeleton was fixed. Bertillon therefore chose to measure
various bony parts, including length of the head, left foot, middle and little finger
and forearm as well as height and eye colour. Statistically, the chances of two
individuals having identical dimensions was very slight and eventually, thousands of
prisoners had their own identity card or 'portrait parle' on file. The system was
costly and labour intensive and was soon superseded by the simpler and more reliable
method of recording fingerprints, developed in colonial India, but further refined by
Galton (Bennett et al. 1994). The one great advantage of fingerprints is, of course,
that they might be left by the villain at the scene of the crime.
6.2 Racial classification
From the 18th century there had been a renewed interest in the classification of races
and attempts were made to substantiate reports of the early explorers by actual
measurements. Considerable ethnic differences in height were reported (Boyd 1981).
Lapps were the shortest, with the average male, allegedly, only four and a half feet,
and the Patagonians the tallest, with the largest men apparently over six and a half
feet tall. By the end of the 19th century, anthropometry had become the tool of all
new schools of physical anthropology, and for a long while, efforts to 'map' the
human body occupied those academics who were interested in the origins and
evolution of man. A more sinister development came when imperialist nations
began to collect anthropometric data in order to rank every race according to
presumed differences in capacities. With some distortion of data regarding brain sizes,
scientists were able to 'prove' what they had always believed in so vehemently,
namely, white supremacy. Paul Broca stands accused of using numbers, not to
generate new theories, but to illustrate a priori conclusions (Gould 1981).
6.3 Standard sizes and mass production

Anthropometry has been put to more mundane but practical purposes. The physical
examination of a large number of US conscripts led directly to the mass-production
of army uniforms: coats in four sizes, shirts in five and trousers in six. The large
scale commercial production of all types of outfits in standard sizes soon followed,
and made-to-measure clothing became the preserve only of the rich and influential -
only they could afford to have bodies of a non-standard size. At the same time, the
so-called 'science' of phrenology, which had positively celebrated individual
differences, by looking for (and finding) unique configurations of bumps on the head,
began to decline. (Bennett et al. 1994).
6.4 New techniques and instruments

From the 1850's German scientists, in particular, had begun to develop a range of
apparatus to measure physiological functions such as blood flow, pulse and strength
(Bennett et al. 1994). These functions were recorded graphically, thus capturing on
paper an easily accessible record of individual performance. Indeed, a cynic would
view the growth charts that were to follow, some 100 years later, as yet another
record of a child's achievement. The new breed of anthropometrists required precision

instruments. A typical anthropometric kit included a hand dynamometer, a
stadiometer and a head calliper to measure cephalic index (the ratio of breadth to
length) (Bennet et al. 1994). Aware of the wide variety of instruments and techniques
in use, Galton and others demanded that they be standardised and in 1882,
anthropometrists from around the world, met in Frankfurt for this purpose.
Consistency of technique was of prime importance and detailed instructions for the
measurement of dozens of different bodily parts were soon available (Hrdlicka 1939).
6.5 Growth measurement and social reform
Nineteenth century Europe now turned its attention to the health and welfare of
children, when physicians realised how appalling working and living conditions,
rather than climate, were stunting their growth. Villerme showed that the mortality
rate from all causes was greater in the poor districts of Paris than in the richer parts
of the country, and went so far as to suggest that government policy was directly
responsible for the stature of its people (Villerme 1828). In 1876, Pagliani also
noted that children from poorer homes were smaller and lighter than those from the
higher social classes, and was able to demonstrate, for the first time, that the growth
of destitute children improved after they were taken into care (Bogin 1988). Rapidly
accumulating social data in England was seized upon by political reformers and
opponents of the government, realising that short stature could be used as an index,
not yet of organic disease, but of social disadvantage. In particular, the plight of
young children in factory work was noted. There was, at that time, no comparative
data to demonstrate just how small these children were, but we do know that the
mean height of English working class children then was less than that in most
developing countries today (Tanner 1981).
6.6 Nature or nurture?
The 19th century also heralded the start of the eternal nature versus nurture debate,
between those who believe the major determinant of growth and development to be
genetic, and those who consider that it is largely shaped by the environment. Broca,
the French anthropologist was a staunch hereditarian. Whilst he conceded that
poverty and nutrition might have some temporary effect, he claimed that they merely
slowed down the rate of growth, delaying the acquisition of final adult height, which
was fixed by the genes (Tanner 1981). To add to this debate, in 1889, Galton
published Natural Inheritance, his seminal work, in which he sought to demonstrate
the heritability of stature and other physical traits.
The environmentalist cause was taken up by Bowditch, a prominent American
physiologist. The physical superiority of the non-labouring classes seemed, in his
view, to depend more on the greater average comfort in which such children live and
grow up (Boyd 1981). Allowing immigration from southern and eastern Europe was
not, therefore, likely to lead to the gradual physical degeneration of the Anglo-Saxon
race in America, feared by some. To support his argument, Boas showed significant
differences in physical characteristics between adult immigrants who had moved to
the US and their children, born in the new country. The children were always taller
10 CHAPTER 1
and heavier, which he ascribed to better health care and nutrition. The hereditarians
and eugenicists remained unconvinced, however, and in 1921, they scored a major
victory with the passage of the Immigration Restriction Acts (Bogin 1988).
7. GROWTH MEASUREMENT AND THE SCHOOLCHILD

Concern for the welfare of schoolchildren was the next impetus to the study of
growth. The idea that too much cerebral activity might have an adverse effect on
growth and development, by diverting the flow of blood from the organs to the
brain, rapidly began to gain ground during the latter part of the 19th century. The
aims of the many large-scale school surveys beginning to emerge in the US as well
as Europe, were not dissimilar to those of the factory reformers. Their purpose was
to ensure, not only that children were fit for school, but that conditions in school did
not impede their growth and development (Tanner 1981). Indeed, physical exercise
was recommended so as to build up sufficient strength to prevent the physique from
breaking down while the brain was being trained and educated. The insanitary
conditions in which many children lived and were taught were often overlooked.
7.1 North American studies

William Porter, professor of physiology at Harvard, was convinced that the physical
strength of the child and his power to do school work went hand in hand (Boyd
1981). It was recommended that no pupil be allowed to enter school without first
undergoing a medical examination, to ensure that he or she was physically strong
enough to withstand the mental strain. At first, Porter was primarily interested in
using weight as a measure of health, and he urged school doctors to make sure that
weight and other physical dimensions were 'normal' in relation to height. As far as
he was concerned, 'The absolute height of an individual is ... of very secondary
interest ... because it is not necessarily a state of ill health'. He also advised that a
short or tall child who was no further above or below the mean after one year was
probably growing normally (Porter 1893), a good rule of thumb, even today.
Increasing interest in child development, during the first half of the 20th century,
led to a number of large-scale longitudinal studies in the US, producing vast
amounts of growth data, though still of more interest to educationists. It was Porter
who initiated the first longitudinal programme of school measurements in the US
and, by 1923, his centile charts, based on monthly measurements of 4800 children
were published. He emphasised, however, that these charts merely described what
was, not what should be, and simply noted how many children moved across centile
bands from one year to the next (Porter 1923). The first standards for height and
weight in North American children were produced by Boas, but without any
guidelines as to their use (Tanner 1981). Although best known for his studies on
immigrants, Boas was also interested in the relation of physical growth to education.
It was imperative, in his view, for teachers to have a sound understanding of the
growth of both the body and brain. Aware of the fact that 'tempo of growth' will
differ from child to child, Boas was responsible for introducing the concept of
'developmental age'. He suggested that the condition of the bones would give a
better insight into physiological development than actual chronological age and
would thus be useful for the regulation of child labour and school entrance ... ' (Boas
1912).
The start of the modem era in terms of medical interest in growth surveillance
can be traced back to Bowditch, a pioneer in the field. In 1875, he started to collect
the heights and weights of many thousands of Boston children. 'It seems probable',
he said, 'that the accurate determination of the normal rate of growth in children will
not only throw light upon the nature of the diseases to which childhood is subject,
but will also guide us in the application of therapeutic measures.' (Bowditch 1881).
The Harvard Growth Study, combined both educatiom.l and medical interests.
Trainee teachers were taught anthropometric techniques and made aware of the
possible effect that illness might have on growth. So persistent was this belief,
among the medical profession too, that in 1934, as many as 61 % of schoolchildren
in New York had their tonsils removed by the age of eleven (Illich 1976).
7.2 European Studies

Growth studies in Europe, in contrast to those in North America, reflected a greater
concern for public health. Those responsible for most large scale school surveys
were interested primarily in making generalisations about and comparisons between
groups; for example, that children from overcrowded homes have the slowest growth.
In theory, if not in practice, social or political measures could then be put in place to
ameliorate the environment. The value of serial growth measurements to monitor
individual growth was also realised. 'If parents would once or twice a year have their
children weighed and measured by a competent person', wrote the English physician,
Boulton, ' ... they would detect at an early stage, irregularities of development, which
would act as danger signals to give warning of approaching mischief; for arrest of
growth ... is one of the earliest appreciable signs of disease.' (Boulton 1876). It was
not enough, however, simply to show the range of children's heights and weights,
for, as Boulton rightly pointed out, 'These school statistics ... do not indicate what
they ought to be.' He therefore attempted to draw up rules for judging whether
velocity was within normal limits. These rules, he realised, had to be conditional on
the height of the child, and he quite correctly allowed around three inches per annum
for a tall child, two for a short (Boulton 1880, Voss and Mulligan 1998).
The Italian, Pagliani, was one of the first to appreciate the importance of the
'individualising' (longitudinal) method, as opposed to the 'generalising' (cross-
sectional) method, as a means of understanding and describing growth (Tanner 1981).
In the latter, subjects are measured once only and though such studies may comprise
a wide age range and thus provide a general description of age-related changes, details
of individual development are lost. Longitudinal studies are necessarily very
demanding, requiring considerable dedication by the researcher, but the data can be
analysed for individual rates of growth and the timing of specific developmental
events.
One enthusiastic proponent of individual growth surveillance was the French
army doctor Godin. In 1919, he introduced the term 'auxology' (Greek: auxein: to
increase) in an article, La Methode Auxologique, although Tanner claims the
responsibility for its current widespread use (Tanner 1981). Godin's definition of the
term, however, was very specific: 'The study of growth by the method of following
the same subjects during numerous successive six-monthly periods (semestres) with
12 CHAPTER 1
a great number of measurements.' As Tanner points out, this must make Godin the
only true auxologist! The system of growth surveillance published by Godin was far
from simple, and involved indices of psychological, motor, growth, energy and
developmental power, that could be compared to normative values. Where a child's
index was more than (an apparently arbitrary) seven semesters from the standard for
his age, he was deemed 'abnormal', though Godin did rightly stress that changes in
position were more important than absolute position on a chart.
7.3 Population growth studies

Although longitudinal standards for growth are to be preferred for individual growth
surveillance, cross-sectional standards are required for population comparisons, or to
monitor the health and nutritional status of, for example, a developing country
(Waterlow 1988). Their differential use has been most succinctly described by
Tanner: 'The neglect that the population cross-sectional type of survey reveals is
neglect by the state: neglect of education, neglect of amenity, neglect of work,
neglect of money. Neglect by the parents is revealed by individual monitoring ... '
(Tanner 1981). Care of the child depends critically on the economic and social
circumstances into which he or she is born. Four major population growth studies
have appeared in the UK in the last fifty years or so, all largely concerned with an
examination of the social and environmental influences on growth and development.
One confirmed the continuing presence of social class differences in height,
(Douglas, Ross and Simpson 1965). Another set out to find the link between short
stature and poverty, supposing that the answer lay in the greater prevalence of
disease, but concluded that the common factor, likely to account for both poor
growth and poor health, was inadequate nutrition (Miller et al. 1974). A third nation-
wide survey of a birth cohort showed regional differences in growth as well as the
effects of socio-economic class, birth weight, family size and smoking during
pregnancy (Pringle, Butler and Davie 1966; Goldstein 1971). A fourth study showed
similar data, with the additional finding that children of unemployed fathers were
shorter by some two cm than those of employed fathers, regardless of class (Rona,
Swan and Altman 1978). While of academic interest, one might well question the
value of costly population surveys while major inequalities in health, many of
which can be attributed to neglect by the state, are shown to persist (Power 1991;
Reading, Raybould and Jarvis 1993, Voss, Mulligan and Betts 1998).
8. JAMES TANNER
Physical anthropology and clinical paediatrics had once been regarded as separate
disciplines, but it was James Tanner who brought the two together. Very much in
the Galtonian spirit, Tanner felt that one of the chief problems in the field of growth
was to determine 'in what ways do men consistently differ from one another' (Tanner
1953). These differences can be observed or measured, and the monumental Atlas of
Child Growth does both (Tanner and Whitehouse 1982). It contains a wealth of
conventional auxological data, but includes, for the first time, 'photogrammetric
anthropometry', a unique visual record of the growth and development of a large
series of children (healthy and otherwise), thus successfully combining Tanner's
earlier preoccupation with somatotypes, or body shape, (Tanner 1947, 1964) with
his emerging interest in child health.
The Harpenden Growth Study, from which the 1966 British standards for height
and weight were mostly derived, is familiar to all paediatricians (Tanner, Whitehouse
and Takaishi 1966). It was a remarkable team effort, led by Tanner, though the idea
for the study came originally from a nutritionist in the Ministry of Health. An
orphanage in Harpenden had taken part in a study concerning the provision of food
during war time rationing, but in 1948, the value of continuing to monitor the
growth and development of these children was realised. Tanner, to quote his own
words, was 'the obvious, and indeed, only candidate' for the post (Tanner 1981). A
recent graduate in medicine from London, he was at the time an anatomy lecturer in
Oxford. He had both the credentials and, in particular, the enthusiasm to pursue such
a study, but first took the opportunity, while in the US, to familiarise himself with
the methodological problems of longitudinal studies. On returning to London, he
secured funding for a full-time assistant. A former army officer, Reg Whitehouse,
took up the post, and thus began a partnership that was to last some 28 years.
Whitehouse's contribution to The Harpenden Study was remarkable. As well as
being a meticulous worker, he single handedly measured every child for the entire
duration of the study, that is, from 1949 to 1970. It has been estimated that he took
15 measurements on approximately 9,000 child occasions (Tanner 1981). In spite of
its imperfections, The Harpenden Study remains a landmark study in the history of
growth measurement. Out of it came, not only the first cross-sectional British
Standards for height and weight, in an easily accessible form, but also, for the first
time, Porter's long awaited longitudinal standards that would take account of the
variation in the 'tempo' of growth (Tanner and Whitehouse 1976). The first usable
standards for pubertal development were also derived from the study. These have
never been superseded, and are widely known as Tanner Stages. Not least, the study
produced a new range of anthropometric instruments. Galton had designed one of the
first 'modern' stadiometers - a fixed vertical rule with sliding horizontal headpiece
(Fergus and Rodwell 1874). Whitehouse perfected the design, creating the Harpenden
stadiometer, still very much in use today.
Two key figures this century had a considerable influence on Tanner. The first
was the British scientist, D'Arcy Thompson, whom Tanner recalls having heard
lecture as a schoolboy (Tanner 1981). Biologist, mathematician, philosopher, and by
all accounts an intellectual giant (Bogin 1988), he devoted a lot of thought to the
nature of the growth process and visualised growth as an essentially dynamic process
involving movement through time. Tanner felt that he was the first really to
understand the velocity curve, that is, the continuous nature of growth. The other
great influence on Tanner has been the Swiss paediatrician, Prader. Unlike Tanner, he
is first and foremost a paediatrician, his interest lying in the clinical application of
growth research. Tanner confesses that it was largely due to Prader that he was
persuaded to renew his own interest in clinical matters, eventually becoming
Professor of Child Health and Growth at the Institute of Child Health in London
(Tanner 1981).
14 CHAPTER 1
9. AUXOLOGY, CHILD HEALTH, AND THE NEW GROWTH CLINICS

Until very recently, the study of growth remained a relatively minor discipline,
tucked away in the syllabus for physical or biological anthropology. At the start of
the child welfare movement in the US, there was great enthusiasm for a 'whole
child' approach that would bring together scientists, clinicians and psychologists
(Tanner 1981). Inevitably, over the decades, each discipline went its own way,
publishing in increasingly specialised journals, such as Annals of Human Biology -
a vehicle for the more theoretical aspects of human growth. Tanner was very much
involved in these new developments from their inception, and writes with some pride
that, .. .'human auxology, over the last twenty or thirty years, has come to have its
feet planted firmly where it is reasonable they should stand: one in human biology
and the other in paediatrics.' (Tanner 1981). Growth, as an index of good health, has
been accepted for some time in the field of preventive medicine. The use of auxology
in the diagnosis and treatment of growth disorders is relatively new. In 1963, just
five years after Raben injected human growth hormone into a child with growth
hormone deficiency (Raben 1958), Tanner opened the first Growth Disorder Clinic in
London. The unlimited supply of genetically engineered growth hormone now
available, has led to a surge in the number of referrals, world-wide, for both real and
perceived disorders of growth. While the current interest in the diagnosis of growth
related pathology is to be welcomed, one has to consider whether the potential to
increase stature, in the many short, but otherwise normal healthy children, is
necessarily in their best interests (Voss 1999,2000).
10. REFERENCES
Bennett, J., Brain, R.. Schaffer, S., Sibum, H.O., and Staley, R., 1994, 1900: The New Age (Whipple
Museum of the History of Science: Cambridge).
Boas, F., 1912, The growth of children. Science, 6,815-818.
Bogin, B., 1988, Patterns of Human Growth (Cambridge University Press: Cambridge).
Boulton, P., 1876, Some anthropometrical observations. British Medical Journal, I, 280-282.
Boulton, P., 1880, On the physical development of children. Lancet, ii, 610-612.
Bowditch, H.P., 1881, The relation between growth and disease. Transaction of the American Medical
Association, 32, 371-377.
Boyd, E., 1929, The experimental error inherent in measuring the growing human body. American
Journal of Physical Anthropology, 13, 389-432.
Douglas, J.W.B., Ross, J.M., and Simpson, H.R., 1965, The relation between height and measured
educational ability in school children of the same social class, family size and stage of sexual
development. Human Biology, 37,178-186.
Encyclopedia Britannica, 1926 (London).
Fergus, W., and Rodwell, G.F., 1874, On a series of measurements for statistical purposes, recently
made at Marlborough College. Journal of the Anthropological Institute, 4, 126-130.
Floud, R., Wachter, K., and Gregory, A., 1990, Height, health and history, nutritional status in the
United Kingdom 1750-1980 (Cambridge University Press: Cambridge).
Goldstein, H., 1971, Factors influencing the height of seven year old children - results from the national
child development study. Human Biology, 43,91-111.
Gould, S.J., 1981, The Mismeasure of Man (Penguin Books Ltd. England).
Gregory, R.L.,1987, The Oxford Companion to the Mind (Oxford University Press: Oxford).
Hacking, 1., 1990, The Taming of Chance (Cambridge University Press: Cambridge).
Hrdlicka, A.,1939, Practical Anthropometry (The Wistar Institute, Philadelphia).
Illich, 1., 1976, Limits to Medicine (Penguin Books Ltd. England).
Kelnar, C.1.N., 1990, Pride and prejudice - stature in perspective. Acta Paediatrica Scandinavia (Suppl)
370,5-15.
Kelvin, W., 1891, Popular Lectures and Addresses, Vol 1. (Macmillan Press) p.80.
Miller, F.J.W., Knox, E.G., Court, S.D.M., Brandon, S., 1974, The School Years in Newcastle upon
Tyne 1952-1962, Being a further contribution to the study of a thousand families (Oxford
University Press: Oxford).
Porter, W., 1893, On the application to individual school children of the mean values derived from
anthropological measurement by the generalising method. Quarterly Journal of the American
Statistical Association, 1, 576-587.
Porter, W.T., 1923, Percentile charts of the height and weight of Boston schoolchildren. Boston Medical
and Surgical Journal., 188,639-644.
Power, c., 1991, Social and economic background and class inequalities in health among young adults.
Social Science Medicine, 32, 411-417.
Prader, A., Tanner, J.M., von Harnack, G.A ., 1963, Catch-up growth following illness or starvation.
Journal of Pediatrics, 62, 646-659.
Pringle, M.L.K., Butler, N.R, and Davie, R.,1966, 11,000 Seven-year-olds (Longman: London).
Raben, M.S.,1958, Treatment of a pituitary dwarf with human growth hormone. Journal of Clinical
Endocrinology and Metabolism, 18,901-903.
Reading, R., Raybould, S., and Jarvis, S.,1993, Deprivation, low birth weight, and children's height, a
comparison between rural and urban areas. British Medical Journal, 307, 1458-1462.
Rhodes, P.,1985, An Outline History of Medicine (Butterworths: London).
Rona, R.J., Swan, A.V., Altman, D.G., 1978, Social factors and height of primary school children in
England and Scotland. Journal of Epidemiology and Community Health, 32, 147-154.
Scammon, RE., 1927, The first seriatim study of human growth. American Journal of Physical
Anthropology,.10,.329-336.
Suetonius Translated by J.C. Rolfe, 1928, Loeb Classical Library (Heineman: New York) ..
Tanner, J.M., 1947, The morphological level of personality. Proceedings of the Royal Society of
Medicine, XL, 301.
Tanner, J.M., 1953, Growth and constitution. In Anthropology Today, edited by A.L. Kroeber
(University of Chicago Press: Chicaog) p.750-770.
Tanner, J.M., 1964, The Physique of the Olympic Athlete (G. Allen and Unwin Ltd.: London).
Tanner, J.M., 1981, A History of the Study of Human Growth (Cambridge University Press:
Cambridge).
Tanner, J.M., and Whitehouse, R.H., 1976, Clinical longitudinal standards for height, weight, height
velocity and weight velocity and the stages of puberty. Archives of Disease in Childhood, 51, 170-
179.
Tanner, J.M., and Whitehouse, RH., 1982, Atlas of Children'S Growth, Normal Variation and Growth
Disorders (Academic Press: London).
Tanner, J.M., Whitehouse, RH., and Takaishi, M., 1966, Standards from birth to maturity for height,
weight, height velocity and weight velocity, British children 1965. Archives of Disease in
Childhood, 41, 454-471, 613-635.
Tietze-Conrat, E., 1957, Dwarfs and Jesters in Art (Phaidon Press: London Press)
Villerme, L.R., 1828, Memoire sur la mortalite en France dans la classe aisee et dans la classe
indigente. Memoires de I'Academie de Medecine, 1,51-98.
Virey, 1.1., 1816, Geant Dictionnaire des Sciences Medicales, 17,553.
Voss, L.D., 1999, Short but normal. Archives of Disease in Childhood, 81, 370-371.
Voss, L.D., 2000, Growth hormone therapy for the short normal child, who needs it and who wants it?
Journal of Pediatrics, 136, 103-110.
Voss, L.D., and Bailey, B.J.R, 1997, Diurnal variation in stature - is stretching the answer? Archives of
Disease in Childhood, 77, 319-322.
Voss, L.D., and Mulligan, J., 1998, Normal growth in the short normal prepubertal Child. Journal of
Medical Screening, 5, 127-130.
Voss, L.D., Bailey, B.J.R, Cumming, K., Wilkin, T.J., and Betts, P.R., 1990, The reliability of growth
measurement. Archives of Disease in Childhood, 65, 1340-1344.
Voss, L.D., Mulligan, J., and Betts, P.R, 1998, Short stature at school entry - an index of social
deprivation, (The Wessex Growth Study). Child, care, health and development 24,145-156.
Voss, L.D., Wilkin, T.J., Bailey, B.J.R, Betts, P.R., 1991, The reliability of height and height velocity in
the assessment of growth. Archives of Disease in Childhood, 66, 833-837.
Wasse, J., 1724, Concerning the difference in height of a human body, between morning and night.
Philosophical Transactions of the Royal Society of London 33,87-88.
Waterlow, J.C., 1988, Linear Growth Retardation in less developed countries (Raven: New York) pp.
238.
CHAPTER 2
INDIVIDUAL PHYSICAL GROWTH MODELS AND

BIOLOGICAL PARAMETERS OF JAPANESE
T. SHOHOJIt, T. SUMIYAt
tFukuyama University, Fukuyama, Japan, IHiroshima University, Hiroshima, Japan
1. INTRODUCTION
Human physical growth is a dynamically changeable and inherently vital
phenomenon. Individual growth has own characteristics. The physical dimension of
growth is widely influenced by heredity and life environment. The fluctuation of age
and size attaining at a special growth phase is large among subjects. Growth patterns
gradually change over time and geography. Thus, an optimal asymptotic growth
model is useful and effective for both data reduction and the characterisation of
individual and average physical growth.
Falkner and Tanner (1986), Johnston et al. (1980), and Goldstein (1979)
describe the methodology of longitudinal and cross-sectional growth studies from
the viewpoint of theoretical and practical problems. Boyd (1980), Tanner (1981),
and Hauspie et al. (1995) cover various approaches on physical growth, and present
fruitful work on the history of human growth study from all over the world and a
large volume of references. Kshirsagar and Smith (1995) illustrate various statistical
techniques for studying growth science and repeatedly measured experiments.
One of the purposes for studying growth models is to establish individual
growth patterns and to predict future individual growth as well as to characterise and
describe physical growth. In selecting suitable growth models, we look for a model
that has as few number of free growth parameters as possible. After estimating
growth curves, we should evaluate and compare their goodness of fit and check the
independence of residuals. Here, we introduce an extended Count-Gompertz growth
model.
Once we choose a suitable mathematical growth model and estimate individual
growth parameters, we can estimate individual biological parameters of the growth
feature. We can compare the means of biological parameters among the samples and,
also, discuss the relationship among biological parameters. It is not clear whether
the relationship among biological parameters may be correlated to time and area or
not.
Little or no correlation exists apparently between adult height and the timing of
the pubertal growth spurt in height. However, the partial correlation between adult
height and age at peak height velocity is negative when height at peak height
17
18 CHAPTER 2
velocity is controlled (Qin et al. 1996). We study stable relationships between adult
height and some biological parameters (menarcheal age, age at peak height velocity)
of physical growth in detail.
2. APPROACHES
In conducting a long-range follow-up study of human beings, it is difficult to collect

a complete set of serial measurements according to a pre-assigned design. This is
usual due to the human factor, i.e. even if a subject wishes earnestly to collaborate in
the study, there are often unavoidable personal reasons to stop participating. Thus,
we have to develop an effective procedure for analysing such incomplete
measurements.
By eliminating the information on the age at examination so as to make efficient
use of incomplete measurements, we estimate the unknown growth parameters in a
growth model. That is, we may project the unequal incomplete measurements into a
growth parameter space and a biological parameter space by estimating parameters.
As the data (estimates) in these projected spaces become complete sets, we can
easily analyse them by usual statistical methods. The estimators of individual
growth parameters are consistent and asymptotically normally distributed under
some regularity conditions (Shohoji, 1982). The estimates of biological parameters
are empirically normally distributed.
We will discuss growth patterns on a growth parameter space and a biological
parameter space. The mean-constant growth curve is provided by the means of
estimates of individual growth parameters. Also, we consider relationship among
biological parameters.
3. MATERIALS
We use the height and weight of 265 Japanese girls from the Hiroshima Growth
Study sample. Height of these subjects are open to public (Shohoji et ai, 1991).
Most girls were born from 1965 to 1968 and they have roughly one observation in
each spring aged of 6 to 18 years. The age intervals between two successive
measurements are at most 40 months. They all have the records on height and
weight at birth. The maximum and minimum numbers of measurements are 22 and
10, and the average numbers of height and weight are 17.5 and 18.0, respectively.
4. GROWTH MODELS
Let t be the age at examination and let y(t) be the measurement at the age t. We
assume a statistical model such as y(t) = H(t) + et where the random variable et is
normally distributed with the mean 0 and variance al. H(t) is called a growth
(distance) curve. The first derivatives of H(t) with respect to t is called its velocity
curve.
Jenss and Bayley (1937) and Count (1943) propose growth curves during
childhood. Winsor (1932), Marubini et al. (1972) and Marubini and Milani (1986)
GROWTH MODELS AND BIOLOGICAL PARAMETERS 19
A-Bt
compare the Gompertz curve H (t) = e-e and a logistic (autocatalytic) curve
H(t) = ( 1 + e A- Bt ) -1 during adolescence. These two may be generalised as a
-lIF
generalised logistics curve H(t) = ( 1 + Fe A - Bt ) .
One longitudinal growth model, starting at around 1970, describes an
asymptotic lifetime growth from birth to adulthood by mathematical expression.
Bock et al. (1973) propose a double logistic model, and Bock and Thissen (1980)
suggest a triple logistic growth model from the age of 1
H(t)=a l ! 1- p
1 + e -q(t-CJ)
+1+ p
e -i'2(t- c2)
)+ 1+ a2
e -b.3(t-c3)
. (1)
This can represent the mid-childhood growth spurt. Preece and Baines (1978)
successfully develop a new family of growth models, one of which is
H(t)=V- 2(V-he) (2)

eso(t-e) + e SI (t-e)
This is applicable for height growth from the age of 2.

Jolicoeur et al. (1988) propose a seven parameter growth curve which passes
through the origin respect to the total age from the day of fertilisation. Jolicoeur et
al. (1992) improve it as JPA2
H(t) = V 1--------------------------- (3)
'+(';,Et +( '~Et +( ';'Et

where the age t is the age after birth. This model successfully depicts the suitable
mid-growth spurt only if there are enough measurements and is well fitted to
longitudinal measurements of growth. This may be sensitive to fluctuation of
measurements from the viewpoint of the jack-knife. Hauspie (1989) organise a good
summary on growth models in the analysis of human growth record, from the view
point of their applicability, limitation, and merit.
We desire a basic growth process represented by a fundamental functional form
described biologically. This fundamental form consists of structural models for
characterising and summarising serial measurements and growth patterns. The
human physical growth consists of various growth cycles and the corresponding
growth components. We conventionally divide growth period into two growth
cycles such as a childhood growth and an adolescent growth cycle. A fundamental
model H(t) = g(t) + J(t){ V - g(t)} is used where g(t) is a preadolescent growth
curve, V is the adult size and J(t) is a relative measure of maturity at the age t.
20 CHAPTER 2
J(t) may satisfy a second order differential equation taking an ageing process into
consideration (Richards 1959, Bertalanffy 1957, Day 1966).
For such a special case that g(t) is a modified Count model and J(t) is the
Gompertz model, we may introduce the Count-Gompertz growth model
H(t)={C+Dt+Eln(t+l)}.I-e-e ( A-BtJ +Ue-eA-Bt . (4)
where t is the age in months. (A,B,C,D,E,U) is a growth parameter vector, each

component of which is positive. Its growth velocity curve is
dH(t)
h( t) = --;j( ="c
l,(t) + ha(t) (5)
where hc(t) = l-e-e ( E J and

A-BtJ( D+ t+l
ha(t) = BeA-Bte -e
A-Bt {U - C - Dt - Eln(t + l)}.
160
Growth Curve H(f) 1.4
140 -..
1.2 ~
1.0 E
a
-.. 120
E
()
'-"
..... 0.8 ~
-;=1) 100 .G 0
.~
::c
80 ~ .. :
\
....a.
0.6·g
-
Q)
hc(tr~.. >
t
-- ·Age at MHV·--~:
:
:
:,..!..
,'1 '\,
('I ,
~
~
0.4
-=~
60 ---- AgeatPHV ----~----I~ '~~ .. : ~r. 0.2 0
, , : ha(t) ...... ' 'J ....
------ Age at GC------~--.... ---~-------~... , .... d
.- .- -- ---- - - .- .- - .----- - . ---- .~-!":-.--.t----..- ..- ". -"'-.."-;;..._: -... _~.- - 0.0
40
0 2 4 6 8 iO 12 14 16 18 20
Age (years)
Figure 1. The Count-Gompertz growth model and derived biological parameters

1W o r- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -- - - - .
Growth Curve H(I) 1.4
: Height increment
140
Mid·growth spun Length of : adolescence
: dunng
1.2 .s""
. _~~c;l~~:<:W §
, , 1.0 E
""1 20
E
() E
'-' 0.8 ~
~1()()
Adult Height
o
'g
Of}
.Q3 .t.
e- 0.6
::c: o
80 ~
.:.t. ~
..s
~,
I ...... I ' : \ C)' 0.4
~... ~
-.--- Age at MHV--:I.. --- -----; '--:"' :. h \
(1)\ ...~
W , : , 'i" a
----AgeatPHV-;~;:.---.::F:,'~,'"
, t-$.
"I)
0.2 ~
l'J
.Age at GC -.;:~ ...----~-.-. -.,---
I hs~ ,
r~;':'-"""~~ 0 .0
" I 1 --
- - _ ._ - . . _ _• : o'- _...
_ __ __ __ ._ .- . _ - _ _ lflii . _ _ .. _ _~_.. .
40 '
o 2 4 6 8 10 12 14 16 18 20
Age (years)
Figure 2. The extended Count-Gompertz growth model and derived biological

parameters
The solid increasing function in Figure 1 is the growth curve of height H(t) and
the other solid curve is its growth velocity curve h(t). This velocity is divided into
a childhood component 1Ie(t) and adolescent component ha(t) drawn by the dashed
curves. In Figures 1 and 2, the circle (0) is height and the step function is the
average of growth velocity between successive measurements and (A) indicates her
menarche.
The adult size is an asymptote of a growth curve when the age t tends to
infinity. Let the age at peak height velocity (PRV) and the age at minimum height
velocity (MRV) be, respectively, the age of attaining at the local maximum of
growth velocity curve and the local minimum. We conventionally define the age at
growth completion (GC) as the solution of h(t) = 0.1. Once a growth model is fitted
to individual record, the dimensionality of the record is adjusted and regulated. That
is, the record are projected from a vague measurement space to a growth parameter
space and a biological parameter space. We easily apply usual statistical procedures
to the estimates of parameters.
Some children have a mid-growth spurt. The growth curves (1) and (3) may
represent the mid-growth spurt if there is one, but it might be difficult to estimate a
stable and well fitted growth curve. We now introduce a growth model representing
a mid-growth spurt. Let J(t) and get) be a childhood and a school-child growth
curve, respectively. Let H(t) be the height at the age t in months and let U and
U 0 be the adult size and the maximum height during late childhood school-age,
respectively. We have a growth model such as H(t) = J(t) + Ja(t){ U - J(t)} and
22 CHAPTER 2
J(t)=g(t)+Jc(t){Uo-g(t)} where Ja(t) and Jc(t) are monotone increasing

functions between 0 and 1. As the form of J(t) is unknown, we eliminate J(t)
from the system. We have
H(t) = (1- Ja(t) )(1- Jc(t) )g(t) + U oJc(t)(I- Ja(t») + UJa(t). (6)
When we choose suitable functions for g(t), Ja(t) and Jc(t), we can get an
asymptotic lifetime growth model that may represent the mid-growth spurt if there is
one. We define an extended Count-Gompertz growth model
H(t)
{Aa-Bat)(
={C +Dt + Eln(l + t)~ 1- e-e 1- e-e
Ac-Bct)
_ Aa-Bat) _ Ac-Bct _ Aa-Bat

+aU( 1- e e e e + Ue e (7)
where (Aa' Ac,Ba,Bc,C,D,E,U) and a are the growth parameters.

This velocity curve may be divided into three components such as 1Ie(t) during
childhood, hs(t) during late childhood school-age and ha(t) during adolescence as
shown in Figure 2. This Figure shows the growth curve (7) of height and its
biological parameters. This growth model is well fitted to height growth and the
growth model may be applied to weight growth as well as development of height
growth. Also, we have similar relationships among the biological parameters.
However, for the extended Count-Gompertz model (7), we have the mid-
childhood growth spurt for about 30 and 34 percent of boys and girls, and about 22
and 16 percent of boys and girls for JPA2 (3), respectively. Here, we do not discuss
the extended Count-Gompertz growth model and its mid-childhood growth spurt
because we still need a more detailed study on this phenomena.
5. ESTIMATION
The number of measurements and the ages at examination, in a longitudinal human
growth study, usually vary from subject to subject. It is difficult to identify
individual growth and to compare growth patterns directly by the original
measurements. We can reduce the dimensionality by estimating the growth and
biological parameters with the minimum loss of information on measurements.
We may use the least squares method to estimate growth parameters. We set up
a reasonable convergent criterion for non-linear iteration considering the daily and
seasonal variation of growth increment and accuracy of calculation. A normalised
expected amount of Fisher's information (Rao, 1952) on growth parameters per
observation is introduced which is useful for designing a cohort observation scheme
for non-linear regression (Shohoji, 1982). We can show an asymptotic unbiasedness
by obtaining the biases of estimators.
5.1 Goodness offit

As the number of unknown parameters may be different for each model, it is
difficult to compare directly the goodness of fit among non-linear growth models.
However, we may use the Akaike Information Criterion (Akaike, 1973) AIC = -2 x
(the maximum log-likelihood) + 2 x (the number of free parameters in the model) for
comparing the goodness of fit. This criteria considers the penalty for the number of
free parameters in a model. The growth model making the amount of AIC
minimised is chosen as the best model among the target. We can, also, use cross-
validation and Durbin-Watson ratio at selecting models. We had better chosen a
growth model from the bio-mathematical and biological point-of-view as well as
statistically evaluating a model only by the goodness of fit. As an approach selecting
a growth model, Susman et at. (1998) apply the very useful modem mixed model
approach combining both fixed and random effects in the same model (Lindstrom
and Bates, 1990).
We apply the Preece-Baines growth model (2) and Count-Gompertz growth
model (4) to height and weight of 265 Japanese girls. Tables 1 and 2 present the
fundamental statistics of growth parameters and some biological parameters,
respectively. It should be mentioned that there exist statistically significant
differences at the 5% significance level between the means of estimates of biological
parameters in these two growth models from a practical point of view for discussing
the relationship among biological parameters. However, we can consider these two
estimates are almost the same. From now on, we will mainly discuss the results
obtained from the Count-Gompertz model because we have similar results for the
Preece-Baines model; and although we do not go into details, we also get similar
results for estimates due to just linear interpolation.
5.2 Growth parameters and mean-constant growth curves

For Figure 3, the samples are conventionally divided into four menarcheal height
groups. The monotone increasing curves of the top of Figure 3 are the mean-constant
growth curves for the Count-Gompertz growth model by menarcheal height groups.
The curves on the bottom of Figure 3 are their corresponding growth velocity
curves. The horizontal axis is the age in years. The left vertical axis is height in cm
for growth curve and the right is growth velocity in cm/month. The point (.) on
each growth curve mentions the average menarcheal ages of each group. Those who
have a short height at menarche end up shorter as adults than those who have a taller
height at menarche.
Figure 4 presents the mean-constant growth curves of weight broken down by
menarcheal weight groups. Weight growth has similar tendency as height growth.
The menarcheal weight may play an important role for characterising weight growth
as well as the menarcheal height for height growth.
24 CHAPTER 2
Table 1. Statistics of growth parameters for height and weight (265 Japanese girls)
Growth Height Weight

parameters Mean Median SD Mean Median SD
Count-Gompertz model
A 9.233 8.173 5.002 7.278 6.561 2.896
B 0.0629 0.0569 0.0311 0.0510 0.0466 0.0203
C 48.75 48.85 2.295 3.34 3.33 0.494
D 0.3770 0.3768 0.0483 0.1356 0.1341 0.0340
E 7.938 7.963 1.257 1.351 1.353 0.440
U 155.84 156.01 4.649 52.77 52.00 6.837
Preece-Baines model
So 0.0923 0.0901 0.0192 0.0043 0.0042 0.0015

sl 0.0097 0.0098 0.0012 0.0780 0.0733 0.0240
() 139.13 139.10 11.05 147.26 147.45 12.10
he 144.39 144.37 4.68 39.62 39.16 4.56
U 156.68 156.74 4.65 52.06 51.24 6.48
Table 2. Statistics of biological parameters relating to menarche and growth

completion (265 Japanese girls)
Statistics Height at Age at Adult Weight at Age at Adult

menarche growth height menarche growth weight
(cm) completion (cm) (kg) completion (kg)
of height of weight
(years) (years)
Mean 149.41 14.31 155.84 40.56 15.71 52.77
SD 5.98 0.86 4.65 5.02 1.49 6.84
Median 149.61 14.32 156.01 40.57 15.56 52.00
Maximum 166.01 16.51 170.07 57.97 20.52 93.87
Minimum 112.94 11.88 142.42 21.38 11.92 38.19
Preece-Baines model
Mean 149.31 14.35 156.68 40.39 15.41 52.06
SD 5.88 0.94 4.65 5.05 1.27 6.48
Median 149.81 14.36 156.74 40.38 15.29 51.24
Maximum 165.67 18.10 170.48 57.22 19.09 87.56
Minimum 113.32 12.07 142.56 21.22 11.96 38.15
160
1.4 '""'
-=
140 1.2 §
J§
1.0 E
EI20
~ ~
0.8e
~100
' dj
'g
0.6 Qj
::r: :>
80
0.4~
o
60 0.2 0
40 ' .. '· ... i·....·..i ~· .... i.. ·.... ·:·.. ·· ..i···.. ··i··· .. ···i·..;;
, , 0.0
o 2 4 6 8 10 12 14 16 18 20
Age (years)
Figure 3. Mean-constant growth curves and averages of menarcheal ages (oj by

menarcheal height groups.
A(61) menarcheal height:O::; 145 cm 8(76) 145 cm < menarcheal height:O::; 150 cm
C(90) 150 cm < menarcheal height:O::; 155 cm D(38) 155 cm < menarcheal height
~ , ~
1.4 .......
.c
50
1.2§
E
~40
~
1.0~
'-'
0.8e
~30 'g
' dj
0.6 -a
~ 20 :>
0.4 ~
10 0.2 0 e
........ ·~ O.O
2 4 6 8 10 12 14 16 18 20
Age (years)
Figure 4. Mean-constant growth curves and averages of menarcheal ages (oj by

menarcheal weight groups.
A(34) menarcheal weight :0::; 35 kg 8(82) 35 kg < menarcheal weight:O::; 40 kg

C(99) 40 kg < menarcheal weight :0::; 45 kg D(50) 45 kg < menarcheal weight
26 CHAPTER 2
5.3 Biological parameters and their relationships
Table 3 presents the sample correlation coefficients between menarcheal age,

menarcheal and adult body dimension and age at growth completion. The upper and
lower triangles of Table 3 are, respectively, the correlation coefficients for the Count-
Gompertz and the Preece-Baines models. These structures of the correlation
coefficients are almost similar.
Table 3. Correlation coefficients between biological parameters (265 Japanese girls)
Biological Age at Height at Age at Adult Weight at Age at Adult

parameters menarche menarche growth height menarche growth weight
(MA) (MH) completion (AH) (MW) completion (AW)
of height of weight
(AGCH) (AGCW)
(MA) 0.521 0.672 0.026 0.325 0.509 - 0.110
(MH) 0.513 0.117 0.701 0.643 0.249 0.275
(AGCH) 0.694 0.121 0.109 - 0.034 0.595 - 0.048
(AH) 0.060 0.720 0.166 0.384 0.184 0.452
(MW) 0.336 0.654 - 0.025 0.386 0.037 0.573
(AGCW) 0.546 0.260 0.643 0.228 0.008 0.446
{AWl - 0.149 0.287 - 0.087 0.469 0.583 0.329
Preece-Baines model
Notes: The correlation coefficients in the upper triangular and the lower triangular are
respectively, for the Count-Gompertz model and the Preece-Baines model.
Let r (menarcheal age, adult height) = 0.026 denote that the correlation
coefficient between menarcheal age and adult height is 0.026 for the Count-Gompertz
model. Let r (menarcheal age, adult height I menarcheal height) = -0.558 represent
their partial correlation coefficients when menarcheal height is held constant (i.e.,
menarcheal height is statistically controlled). And let r (menarcheal age, adult height
I menarcheal height, age at growth completion of height) = -0.845 describe their
partial correlation coefficient when menarcheal height and age at growth completion
are simultaneously held constants (Tables 4 and 5). Also, we have r (menarcheal
height, age at growth completion of height) = 0.117, r (menarcheal height, age at
growth completion of height I menarcheal age) = -0.369 and r (menarcheal height,
age at growth completion of height I menarcheal age, adult height) = -0.798. For
the Preece-Baines model, we have r menarcheal age, adult height) = 0.060, and r
(menarcheal age, adult height I menarcheal height, age at growth completion of
height) = -0.905.
For both growth models, menarcheal age and adult height are highly correlated
with each other in a cross-section of all subjects having the same amount of
menarcheal height and age at growth completion although menarcheal age and adult
height are almost independent as a whole.
Table 4. The correlation coefficients and partial correlation coefficients among the
contra-pairs of biological parameters (265 Japanese girls) derived from the Count-
Gompertz model
Partial correlation coefficients

Biological raf3 ra f3ey ra f3e8 raf3ey8
parameters
ry8 ry8ea ry8ef3 ry8eaf3
a: adult height
/3: menarcheal age 0.026 -0.558 -0.064 -0.845
r : menarcheal height 0.117 0.058 -0.369 -0.798
8: age at growth completion
a: adult height
/3: age at peak height vel. 0.090 -0.628 0.090 -0.872
r : height at peak height vel. 0.172 0.135 -0.414 -0.819
a: adult weight
/3: menarcheal age -0.110 -0.382 -0.438 -0.902
r : menarcheal weight 0.037 -0.298 -0.158 -0.873
a: adult weight
/3: age at peak weight vel. 0.044 -0.206 -0.363 -0.845
r : weight at peak weight vel. 0.190 -0.377 0.094 -0.835
8: age at growth comQletion
Notes: raf3 is the correlation coefficient between variables a and /3, ra f3ey is the
partial correlation coefficient between a and /3 when hold constant variable r, and
ra f3ey8 is the partial correlation coefficient when hold constant r and 8.
Let two growth phases be Fl and F2 and let (al,fl) and (a2,h) be a pair of
the corresponding biological parameters, respectively. For example, we take the ages
for al and a2 and heights for II and h. When the correlation coefficient between
al and h is almost zero or positive and its partial correlation coefficient is negative
in a cross-section of all subjects having the same amounts of the pair (a2,/l), the
special pair of biological parameters (al,h) is called the contra-pair of the
biological parameters (a2, 11) with the negative partial correlation.
From Tables 4 and 5, adult height and menarcheal age are a contra-pair of
menarcheal height and age at growth completion of height. Adult height and age at
peak height velocity are also a contra-pair of height at peak height velocity and age
28 CHAPTER 2
at growth completion of height. A pair of adult height and age at minimum height
velocity is also the contra-pair of height at minimum height velocity and age at
growth completion. Adult height and age interval from peak height velocity to
growth completion are the contra-pair of age at growth completion and height
increment after peak height velocity. These are true for weight growth as well as
height growth and also true for Japanese boys as well as girls.
Table 5. The correlation coefficients and partial correlation coefficients among the
contra-pairs of biological parameters (265 Japanese girls) derived from the Preece-
Baines model
Partial correlation coefficients

Biological raf3 ra f3ey ra f3 e8 ra f3ey8
parameters ry8 ry8 ea ry8 ef3 ry8eaf3
a: adult height
{3: menarcheal age 0.060 -0.518 -0.077 -0.905
Y : menarcheal height 0.121 0.003 -0.380 -0.885
/5: age at growth completion
a: adult height
{3: age at peak height vel. 0.114 -0.519 -0.013 -0.910
Y : height at peak height vel. 0.151 0.031 -0.352 -0.890
a: adult weight
{3: menarcheal age -0.149 -0.450 -0.415 -0.908
Y : menarcheal weight 0.008 -0.239 -0.221 -0.875
a: adult weight
{3: age at peak weight vel. -0.033 -0.255 -0.356 -0.873
Y : weight at peak weight vel. 0.197 -0.258 0.125 -0.849
/5: age at growth comQletion
Notes: raf3 is the correlation coefficient between variables a and {3, ra f3ey is the
partial correlation coefficient between a and {3 when hold constant variable y, and
ra f3ey8 is the partial correlation coefficient when hold constant y and /5.
Figure 5 presents the scatter diagram of menarcheal age and adult height and 80%
concentrated ellipsoids by menarcheal height groups. In the legends of Figures 5 and
6, the numbers in the parenthesis between alphabet and symbol show the sample
sizes of the groups. The dotted ellipsoid "A" presents the 80% concentrated
ellipsoid for the menarcheal height group (A: :::;145 cm, 61 girls), menarcheal height
of which is shorter than or equal to 145 cm. Figure 6 shows a scatter diagram of
menarcheal age and adult weight and the 80 % confidence ellipsoids by the
menarcheal weight groups.
170 i C
t;.
165 I +
~
::: 160
-=01)
'Q)
:r:: 155
"3
"0
< 150
? o
+ + .
145 + ...t.. ++.... +
. A .. .+···
9 10 11 12 13 14 15 16
Menarcheal age (years)
Figure 5. Scatterdiagram of menarcheal age and adult height. and the 80% confidence
ellipsoids by menarcheal height groups.
A(61) + menarcheal height:<:; 145 em B(76) 0 145 em < menarcheal height:<:; 150 em
C(90)· 150 em < menarcheal height:<:; 155 em D(38) A 155 em < menarcheal height
75
70
,.....,
~ 65
'-"
~1) 60
.~
~ 55
....
:g 50
<
45
40 L ' ..+0
9 10 11 12 13 14 15 16
Menarcheal age (years)
Figure 6. Scatterdiagram of menarcheal age and adult height. and the 80% confidence
ellipsoids by menarcheal weight groups.
A(34) + menarcheal weight :<:; 35 kg B(82) 0 35 kg < menarcheal weight:<:; 40 kg
C(99) • 40 kg < menarcheal weight :<:; 45 kg D(50) A 45 kg < menarcheal weight
30 CHAPTER 2
Although the correlation coefficient between menarcheal age and adult height is
almost zero apparently for the whole sample, adult height linearly associates with
menarcheal age in a cross section of all girls having the same menarcheal height.
Those who have taller menarcheal height may shift into later menarche and taller
adult than those who have shorter menarcheal height. The taller the menarcheal
height is, the shorter the average length from menarche to growth completion and
the average of growth increment of height during the period are. When we restrict
any menarcheal height groups, those who have younger menarcheal age end up on
average taller adults than those who reach menarcheal age later.
6. DISCUSSION
When selecting a structural growth model, it desires to choose a growth model that
• is evaluated from the viewpoint of goodness of fit (e.g., AIC)
• is interpretable from biological and medical points-of-view
• obtain stable estimates of growth parameters by fitting the model (the asymptotic
variances of the estimators are relatively small).
The sample distributions of estimates of many biological parameters are empirically
normally distributed. The current mixed model technique and other recent
approaches are effectively used for analysing these estimates.
If there exists a mid-growth spurt of growth, the triple logistic (2), JP A2 (3) and
the extended Count-Gompertz (7) may accurately describe the mid-growth spurt.
However, we have to improve the growth models because the coefficient of variation
of the estimates for biological parameters relating to the mid-growth spurt is very
large.
The growth models developed for height growth may be applicable for weight
growth even if the fluctuation of weight is relatively large. Moreover, we need such a
non-linear growth model where height and weight are simultaneously considered for
reducing the number of unknown growth parameters.
When the biological parameters are estimated for some assigned growth models,
there may be statistically significant differences among the means of estimates of the
biological parameters of the target growth models in the case the sample sizes are
large. However, we will get some simple and consistent relationships among
biological parameters for any growth models whenever their growth models are well
fitted within a certain extent.
When studying the relationships among the biological parameters, we have to
consider the true relationship among the biological parameters as well as spurious
relationships. It is, needless to say, regarded as one of essential parts of any growth
model analysis for characterising physical growth from a biological point of view, to
study only true relationship among biological parameters. When we consider the
contra-pair of biological parameters in the case that the relationship between
biological parameters may not be explained directly, it becomes sometime easily to
interpret the relationship of these biological parameters. This contra-pair is very
useful for interpreting physical growth and for predicting the future growth. To study
the relationship among biological parameters is a sort of statistical characterisation of
physical growth from a biological point of view.
Acknowledgement. This work was partially supported by Grant-in-Aid for
Scientific Research (B), Japan Society for the Promotion of Science.
7. REFERENCES
Akaike, H., 1973, Information theory and an extension of the maximum likelihood principle. In the 2nd
International Symposium on Information Theory, edited by 8.N. Petrov and E. Csaki (Budapest:
Akademiai Kiado), p. 267.
Bertalanffy, 1. von, 1957, Quantitative laws in metabolism and growth. The Quarterly Review of
Biology, 32, 217-231.
Bock, R. D., and Thissen, D., 1980, Statistical problems of fitting individual growth curve. In Human
Physical Growth and Maturation: Methodologies and Factors, edited by F.E. Johnston, A.F. Roche,
C. and Susanne (New York: Plenum Press), p. 265.
Bock, R. D., Wainer, H., Petersen, A., Thissen, D., Murray, J. and Roche, A., 1973, A parametrization
for individual human growth curves. Human Biology, 45, 63-80.
Boyd, E., 1980, Origins of the study of human growth. (Portland, Oregon: University of Oregon Health
Sciences Center Foundation).
Count, E. W., 1943, Growth patterns of the human physique: An approach to kinetic anthropometry,
Part I. Human Biology, 15, 1-32.
Day, N. E., 1966, Fitting curves to longitudinal data. Biometrics, 22, 276-291.
Falkner, F., and Tanner, 1. M., 1986, Human Growth: A Comprehensive Treaties. 2nd ed., Volume 3:
Methodology. (New York: Plenum Press).
Goldstein, H., 1979, The design and analysis of longitudinal studies: Their role in the measurements of
change (London: Academic Press).
Hauspie, R. C., 1989, Mathematical models for the study of individual growth patterns. Revue
d'Epidemiologie et de Sante Publique, 37, 461-476.
Hauspie, R., Lindgren, G., and Falkner, F., 1995, Essays on Auxology, (Welwyn Garden City:
Castlemead Publications).
Jenss, R. M., and Bayley, B., 1937, A mathematical method for studying the growth of a child. Human
Biology, 9, 556-563.
Johnston, F. E., Roche, A. F., and Susanne, C., 1980, Human Physical Growth and Maturation:
Methodologies and Factors (New York: Plenum Press).
Jolicoeur, P., Pontier, J., and Abidi, H., 1992, Asymptotic models for the longitudinal growth of human
stature. American Journal of Human Biology, 4, 461-468.
Jolicoeur, P., Pontier, 1., Pernin, M.-O., and Sempe, M., 1988, A lifetime asymptotic growth curve for
human height. Biometrics, 44, 995-1003.
Kshirsagar, A. M., and Smith, W. 8., 1995, Growth curves (New York: Marcel Dekker).
Lindstrom, M. J., and Bates, D. M., 1990, Non linear mixed effects models for repeated measures data,
Biometrics, 46, 673-687.
Marubini, E., and Milani, S., 1986, Approaches to the analysis of longitudinal data, In Human Growth
Vol. 3: Methodology, ecological, genetic and nutritional effects on growth, edited by F. Falkner,
and 1.M. Tanner (New York: Plenum Press) p.79.
Marubini, E., Resele, 1. F., Tanner, J. M., Whitehouse, R. H., 1972, The fit of Gompertz and logistic
curves to longitudinal data during adolescence on height, sitting height and biacromial diameter in
boys and girls of the Harpenden growth study. Human Biology, 44, 511-524.
Preece, M. A., and Baines, M. 1., 1978, A new family of mathematical models describing the human
growth curve. Annals of Human Biology, 5,1-24.
Qin, T., Shohoji, T., and Sumiya, T., 1996, Relationship between adult stature and timing of the pubertal
growth spurt. American Journal of Human Biology, 8,417-426.
Rao, C. R., 1952, Advanced statistical methods in biometric research. (New York: John Wiley & Sons,
Inc), p. 131.
Richards, F. J., 1959, A flexible growth function for empirical use. Journal of Experimental Botany, 10,
290-300.
Shohoji, T., 1982, On the asymptotic properties of the maximum likelihood estimates on a growth curve,
Hiroshima University Statistical Research Group Technical Report, No. 63.
Shohoji, T., Kanefuji, K., Sumiya, T., and Qin, T., 1991, A prediction of individual growth of height
according to an empirical Bayesian approach. Annals of Institute of Statistical Mathematics, 43,
607-619.
Susman, E. P., Murphy, J. R., Zerbe, G. 0., and Jones, R. H., 1998, Using a nonlinear mixed model to
evaluate three models of human stature. Growth, Development and Aging, 62, 161-171.
Tanner, 1. M., 1981, A history of the study of human growth (Cambridge: Cambridge University Press).
32 CHAPTER 2
Winsor, C. P., 1932, The Gompertz curve as a growth curve, Proceeding of the National Academy of
Sciences, U.S.A., 18, 1-8.
CHAPTER 3
SMOOTHING CENTILE CURVES OF HEIGHT OF

BASQUE BOYS AND GIRLS BY THE APPLICATION
OF THE LMS-METHOD
1. ROSIQUEt, L. SAN MARTINt, J.R. FERNANDEZ-LOPEZt, I. SALCESt,

E. REBATOt, A. VINAGREt, c. SUSANNEt
t Laboratory of Physical Anthropology, Department of Animal Biology and Genetics,
Faculty of Sciences, University of the Basque Country, :f: Laboratory of
Anthropogenetics, Free University of Brussels
l. INTRODUCTION
The methods of drawing references for height, using advanced statistical
methodology based on fitting mathematical models of growth are useful to show
both the pattern and timing of growth in individuals and populations (Ledford and
Cole 1998). There is a set of growth models proposed to fit longitudinal and cross-
sectional growth data of attained height by age which are not based on a set of
parameters. They are often called non-structural models (Bock and Thissen 1980).
The underlying assumption, in models depending on a set of parameters (Count
1943, Jenss and Bayley 1937, Preece and Baines 1978, Shohoji and Sasaki 1987,
Jolicoeur et al. 1988, Deming 1957, Marubini et al. 1971, Thissen and Bock
1990), is that the growth pattern has a basic functional form to which a direct
biological interpretation can be attributed (Berkey 1982, Hauspie, 1989). The
disadvantage of these models is that they sometimes impose a too rigid shape upon
the form of the growth curve (Wember et al. 1992, Zemel and Johnston 1994). This
type of functions are normally used to model individual growth curves rather than to
summarise a whole sample; however, they can be also applied to means (Tanner et
al. 1982) or to selected centiles and, in this way, a set of smooth centi1e curves can
be generated. However, the independent fitting of the centile curves is not a good
solution because some centile lines describe heterogeneous patterns of growth and
the distribution of the centile curves often lacks homogeneity. On the other hand,
fitting non-structural models such as polynomials (Welch 1970) and cubic splines
(Largo et al. 1978), the LMS (Cole 1988, Cole, 1990) and Healy's method (Healy
et al. 1988), give also a good description of the growth process and are good
approaches for estimating centile line curves, spaced appropriately relative to
neighbouring curves. However, this kind of models have the drawback of not being
flexible enough or being uncontrollable at the lower and upper end of the age range.
33
34 CHAPTER 3
The LMS-method has been proposed to fit either original data or anthropometric
standards (Cole 1988, Cole and Green 1992) of height and weight and is suitable for
traits which do not necessarily have a monotonously increasing pattern as the BM!
or skinfolds for instance (Cole et al. 1995). The LMS-method performs well in large
samples, but in samples of moderate size, or dealing with reference data collapsed
into few data points, can be difficult to find the median as a steady asymptote, so
some authors (Altman 1993) have proposed an alternative to optimise the
smoothing by applying a function to find the median and then normalise the
standard deviation of height by age by computing the residuals to each data point
and normalising these values by the LMS.
This study was conducted in order to construct cross-sectional growth charts by
applying the LMS-method on the original data of height by age to give a local
reference for Biscay, the western province of the Basque Country. The method was
chosen because it affords suitable smooth centile values of cross-sectional samples of
moderate size. Finally, centiles derived were compared with those obtained from a
mixed-longitudinal sample from the same region (Herruindez et al. 1988).
2. SUBJECTS AND METHODS

The cross-sectional collection of height data from 1625 subjects (768 boys and 857
girls) was part of a research project on the height growth of school children and
adolescents of the province of Biscay (Basque Country, Spain). The exact age in
days was computed as the difference between the date of measurement and that of
birth. Ages ranged from 7 to 19 years in both sexes. Height data were collected by
following the IBP criteria (Weiner and Lourie 1981) during 1997-8. The total size of
the collected sample was near to the 1% of the census size of the schooling
population of the province.
The normalised mean (M), the coefficient of variation (S) and the Box and Cox
(1964) power (L) curves were obtained as smooth functions of age by fitting the
LMS-method using a program written in standard FORTRAN 77 (Cole and Green
1992) where the parameters L, M and S are estimated by maximum penalised
likelihood. A Chi-square test was used to measure the goodness of fit, testing if the
number of data points between centile lines corresponds to the expected number of
points (Cole et al. 1995).
The solution proposed by Altman (1993) to normalise the standard deviation of
height by age, as an alternative to optimise the smoothing was followed because the
fitting of LMS carried out on the whole sample did not result in a pattern with a
central tendency (median) tending towards an upper asymptote. Preece-Baines Model
1 (PB 1) better fitted the central tendency of the sample. Therefore the fitting of LMS
to the residuals of the PB 1 fitting of the central tendency yielded a better solution to
estimate the centile lines. As Cole's method only admits positive values for carrying
out the centiles, the residuals obtained have been rescaled to positive values.
3. RESULTS
The LMS values obtained after fitting the original data (height by age points) are
shown in Table 1. The PB 1 was used to obtain the standard height by age
SMOOTIllNG CEN'TILES BY TIlE LMS ME1HOD 35
distribution. The fit of the PB 1 on the whole sample divided by sex (Table 2) was
considered successful due to the values of the SEE (0.27 in boys and 0.22 in girls)
and the random distribution of the residuals. In fact, for both sexes the runs test did
not indicate systematic bias in the residuals.
Table 1. Values of the IMS parameters fitted to the original data and to the data after
standardisation with PBl fitted to the central tendency.
Boys Girls
Original data Standardised Original data Standardised
data data
L = -0.104 L = 0.849 L = -0.132 L = 0.904
Age, M S M S M S M S
years
7.5 124.85 0.046 124.26 0.228 124.99 0.051 124.41 0.297
8.5 130.21 0.047 130.55 0.246 130.45 0.051 129.65 0.300
9.5 136.06 0.048 136.59 0.265 135.91 0.050 135.08 0.303
10.5 141.06 0.049 141.21 0.282 141.01 0.048 140.61 0.305
11.5 146.63 0.050 145.99 0.299 146.33 0.047 146.98 0.305
12.5 152.32 0.051 15l.l4 0.314 150.82 0.045 152.55 0.304
13.5 157.43 0.051 156.99 0.321 154.54 0.043 156.75 0.298
14.5 162.27 0.051 163.97 0.321 158.02 0.041 159.77 0.292
15.5 167.23 0.049 170.20 0.315 160.45 0.039 161.21 0.288
16.5 171.29 0.048 172.78 0.307 162.52 0.039 16l. 95 0.285
17.5 175.20 0.047 173.64 0.298 164.19 0.041 162.27 0.285
18.5 178.88 0.047 173.86 0.290 165.84 0.043 162.42 0.283
chi- 1.836 2.757 3.401 6.929
square
Table 2. Values of the function parameters and standard error (ME), obtained by
fitting Preece Baines model I (PBI) to the central tendency of the data.
Parameter Boys Girls

hI 173.94 ±0.78 162.57 ±0.66
he 164.73 ±0.78 151.29 ±OA1
so 0.14 ±0.01 0.12 ±0.05
sl 1.42 ±0.24 0.85 ±0.20
e 15.55 ±0.17 12.26 ±0.17
Residual var. 0.27 0.22
The LMS values served to generate sets of seven centiles, from the 3rd to the
97th for boys and girls, displayed in Figures 1 and 2. Figure 1 displays the first step
of the fitting of height by age based on the whole sample leading to a bad estimation
of final mean height. Figure 2 displays the growth chart obtained the LMS-method
applied to the residuals of the PB 1 fit to the central tendency.
36 CHAPTER 3
200
P97
Boys P90
P75
180 P50
P25
P10
P3
§ 160
1:
0>
om
I
140
120
100
6 8 10 12 14 16 18 20
Age, years
200
Girls
----
P97
180 ~
P90
r
P75
! 0>
160
~~~i
om
I
140
120
100
6 18 20
Age, years
Figure 10 Growth charts for boys and girls showing centile lines fitted by EMS on
unstandardised height by ageo
SMOO1HING CENTILES BY TIlE LMS MErnOD 37
200
Boys
P97
P90
180 P75
P50
P25
5 160 P10
-
..c::
.Q>
Q)
P3
I
140
120
100
6 8 10 12 14 16 18 20
Age, years
200
Girls
180 ~ P97
P90
1'60 t
P75
~
P50
P25
P10
P3
Q)
I
140
120
100
6 8 10 12 16 18 20
Age, years
Figure 2. Growth charts for boys and girls showing centile lines fitted by IMS on the
residuals of height after fitting PBI to the median.
38 CHAPTER 3
Table 3. Smoothed centiles for Boys, obtained by applying the IMS-method to the
standardised data
age, P3 PI0 P25 P50 P75 P90 P97

years
7.5 113.90 117.11 120.45 124.26 128.15 13l.73 135.32
8.5 119.44 122.87 126.46 130.55 134.75 138.61 142.48
9.5 124.66 128.34 132.19 136.59 14l.l2 145.29 149.49
10.5 128.58 132.45 136.53 14l.21 146.02 150.46 154.93
11.5 132.62 136.71 141.03 145.99 15l.l1 155.83 160.59
12.5 137.17 141.43 145.94 15l.l4 156.51 161.47 166.47
13.5 142.70 147.05 151.66 156.99 162.49 167.57 172.69
14.5 149.67 154.02 158.64 163.97 169.47 174.56 179.69
15.5 156.17 160.45 164.98 170.20 175.59 180.58 185.60
16.5 159.08 163.26 167.69 172.78 178.04 182.89 187.79
17.5 160.31 164.39 168.69 173.64 178.75 183.46 188.20
18.5 160.89 164.87 169.05 173.86 178.82 183.39 187.99
Table 4. Smoothed centiles for Girls, obtained by applying the LMS-method to the
standardised data
age, P3 PI0 P25 P50 P75 P90 P97

years
7.5 112.53 116.22 120.05 124.41 128.85 132.92 136.98
8.5 117.65 121.37 125.25 129.65 134.14 138.25 142.36
9.5 122.97 126.73 130.64 135.08 139.61 143.76 147.90
10.5 128.42 132.20 136.13 140.61 145.17 149.35 153.52
11.5 134.77 138.56 142.50 146.98 151.56 155.74 159.93
12.5 140.41 144.18 148.10 152.55 157.10 161.26 165.42
13.5 144.80 148.51 152.37 156.75 161.22 165.31 169.39
14.5 148.08 151.71 155.48 159.77 164.14 168.14 172.13
15.5 149.68 153.26 156.98 161.21 165.51 169.45 173.38
16.5 150.50 154.06 157.75 161.95 166.22 170.13 174.03
17.5 150.85 154.40 158.08 162.27 166.53 170.43 174.32
18.5 151.07 154.60 158.26 162.42 166.66 170.53 174.40
The unstandardised height by age (Figure 1) showed departures from the

expected monotonous asymptotic growth in the final age groups for almost all
centiles, more clearly shown by males than by females. The M values follow the
50th percentile and show an asymptotic trend only after standardisation of height
(Table 1). Therefore the centile curves generated by the LMS from the standard
height showed a better solution to construct the charts (Figure 2). The
standardisation of height by age removed the skewness of the original data (Table 1)
as shown by the change in the value of the Box-Cox power (L). The original data in
both sexes showed right skewness (low values of L) while the distribution of the
standard height was almost normal (L is near 1). S values increased with the
standardisation because the change of scale of height. While Table I presents S
SMOOTHING CENTILES BY 1HE LMS METHOD 39
values without rescaling, the final M was rescaled to show the improvement in the
asymptotic trend.
The LMS goodness of fit was checked by comparing the expected and observed
frequencies between the various centile bands, using the chi-square test. The chi-
square test was not significant in boys and girls whether or not standardised height
was employed to fit the model (Table 1).
The final centile values proposed as height by age references for boys (Table 3)
and girls (Table 4) were compared with those previously obtained from the same
region (Hernandez et al. 1988, Ruiz 1989) by computing the percentage of log
differences between centile values: Log (centile IIcentile 2) x 100, taken as
denominator de previous reference. In general, in both sexes, for all ages and centiles
the values of the present study where higher than the previous references as results
from the positive Log differences, with some exceptions at the beginning of the age
range in girls (from 7.5 to 8.5 years in 3rd and 10th centiles) and at the end of the
age range in boys (from 16.5 to 18.5 years in 3rd, 10th, 25th and from 17.5 to 18.5
years in 50th and at 18.5 in 75th).
4. DISCUSSION
In the last years, some longitudinal (Moreno and Carri6 1993, Bemis 1976), mixed-
longitudinal (Hernandez et al., 1988) and cross-sectional (Sandin 1980, Codina
1983, Bemis et al. 1984, Prado et al. 1985, Rosique 1992, Gonzalez 1997, De la
Puente et al. 1997) studies addressed to develop growth charts of height by age have
been conducted in Spain. However, national references for the whole country have
not been proposed and the country tendency is to develop local references by regions
as is the case of the present study. Previously the growth charts used in Spain were
international standards such as those from Tanner and Whitehouse (1976)
recommended by WHO (1978, 1979). Nowadays, the most popular growth charts
used in Spain are those obtained by Hernandez et al. (1988) by means of a mixed-
longitudinal study based on a sample from Biscay. The discussion whether this
region is sufficiently representative of the whole of Spain, to be used as reference, is
already open because a large amount of immigrant children from almost all of the
different Spanish regions can be found in Madrid. Therefore, Madrid should be more
representative of the Spanish population than Biscay, which gathered immigrants
mainly from three distinct regions of Spain (Galicia, Extramadura and Castilla) and
not from the whole country. In spite of this, while lacking other national designs
Biscaian references are suitable. The present cross-sectional growth study could
improve the references of Hernandez et al. (1988) mainly because the ability of cross-
sectional studies of accounting for a greater amount of variance can be useful to
improve longitudinal standards (Tanner and Davies 1985, Hoey et al. 1987), and
last but not least reason, the possible presence of secular trend in height, due to the
lapse oftime past (more than 10 years) between the two surveys.
In spite that Pan and Goldstein (1997) pointed out the usefulness of LMS to
analyse longitudinal data, originally Cole (1988) proposed this method to be used
in a cross-sectional way. Recently, the LMS-method has been applied by Zoppi et
al. (1996) to construct growth charts from a cross-sectional sample of Verona and
proposed centile values as standards for Italy. Cross-sectional growth charts do not
40 CHAPTER 3
have most of the advantages of longitudinal studies (Tanner and Davies 1985) but
they are used on general health evaluations because the recognised great
sensitiveness of height by age data (Berkey et al. 1993). In the present study the
application of LMS on the residuals of the central tendencies of height by age can be
proposed as a good method to construct growth charts for Biscay. It has the
advantages of fitting by LMS the normalised shape of the curves of the centiles of
height distribution and by PB 1 the central tendencies of height by age. Although,
pure cross-sectional growth charts have to be derived from sufficiently large groups of
children (Sinclair 1989), due to the difficult estimation of true values of the extreme
centiles, the combination of PB 1 and LMS to smooth centiles afforded a good
solution in the present Biscaian sample of moderate size. The generated centile lines
have been successfully representative of normal land-marks (Tables 3 and 4) and
proposed as references for height growth of boys and girls
The growth charts based on unstandardised height (Figure 1) showed a poor
solution for drawing centile lines because the right skewness of original data (Table
1), as L values can show, influenced the departure from the monotonous asymptotic
growth of the 50th percentile in the final ages. A larger sample size could correct this
effect. However, Figure 2, based on standardised height by age showed a better chart
because the combination of LMS and PB 1 can cope with the normalisation, in spite
that the distance among centile lines, across ages, was rather homogeneous in both
charts (Figure 1 and 2). In fact, the easier normalisation of centile lines by non-
structural growth models has widely encouraged researchers for using such methods.
The shape of the centile lines (Figure 2) followed a more homogeneous pattern
than those generated by fitting PB 1 on independent raw centi1es as showed other
studies (Rosique et al. 2000). It is well known that for certain ages height does not
present a normal distribution due to increases in popUlation variability and PB 1
cannot afford a solution for this. LMS coped successfully with the excess of
variability of the present sample around puberty by carrying out the necessary
transformation of height distribution and thus determined the various exact centi1es
(Box and Cox 1964).
The application of the LMS to the unstandardised height yields good
correspondence between observed and expected values as shown by non-sigrrificant
chi-square values in both sexes (see Table 1). Thus, the standardisation of height
does not result in a better modelisation, but only a better normalisation, because
LMS better removes the right skewness of the distribution when height is
standardised (change in L coefficient). L with this second approach is near 1 (almost
a normal distribution), as shown in Table 1.
In Spanish studies, centi1e curve smoothing was done, in the past, mainly by
drawing instruments and sometimes done by applying mathematical models only on
selected centile values (Rebato et al. 1993, Martinez et al. 1994), or by free
estimation of centile distribution by non-structural models (De la Puente et al.
1997). Structural models of growth have also been used in order to describe
regional, utban/rural and other ecological differences in Spain (Rosique and Rebato
1995a) by fitting medians or means of height by age of samples taken from the
literature. Limb dimensions and other body segments have also been studied by
means of fitting the PB1 to mean dimensions by age (Rosique and Rebato 1995b)
providing good estimates of mean adult size and average age at peak velocity as in
other studies (Tanner et al. 1982, Cameron et al. 1982, Dasgupta and Das 1997).
SMOOTHlNG CENTILES BY THE LMS ME1HOD 41
Log differences from the mixed-longitudinal study confirmed a shift upwards of

almost all cross-sectionally determined centiles in both sexes, maybe due to a
secular increment in height for Biscaian population at all ages. When secular trend
has little effect, cross-sectional and longitudinal-based growth charts of height by age
can be very near one another. In fact, height by age growth charts longitudinally-
based for the Belgian population (Wachholder and Rauspie 1987) did not differ
significantly from cross-sectional growth standards (Vercauteren 1984) for the same
population. The age at peak height velocity (PRV) of the mixed-longitudinal study
of Biscay was 13.7 in boys and 11.8 in girls (Hernandez et al. 1988, Ruiz 1989).
Therefore, boys of the present study had 0.70 years of delay in the age at PRV with
respect to the previous study, and girls had 0.72 years of advancement with respect
to the age at PRV of the previous study (Table 2). The first parameter of the PB 1
(hI) of the mixed-longitudinal study was estimated in other study by Rosique and
Rebato (1995a) and therefore final height can also be compared with the hI of the
present study (Table 2). Boys had lower final heights (1.62 cm difference) in spite
that girls were taller (2.27 cm difference) in the present study. The decrease in final
height of boys and the increase of girls is a consistent trend found when comparing
the present study hI with other cross-sectional previous studies from Biscay as
those of Rosique (1992) and Gonzalez (1997). The observed trend in final height and
age at PRV could be enough to advising the revision of the mixed-longitudinal
references from Biscay.
In summary the use of LMS after standardising the sample variation of height by
age by the PB 1 permitted to construct good local references for the Biscaian
population and revealed that the secular trend of height growth existing in the
Basque population comparing with previous studies affected almost all centiles and
ages in both sexes. Male population in the Basque Country has probably reached a
limit of height growth while female population continue to improve their biological
potential of growth. The need of continuous revision of Spanish standards by further
research growth projects is obvious to better fulfilling the aims of the auxological
study.
Acknowledgements. This work was partly supported by the Basque Government

(Department of Education, Universities and Research) through two pre-doctoral
grants and by the UPV 154.31O-EA049/99 project of the University of the Basque
Country (UPV/EHU).
5. REFERENCES
Altman, D.G., 1993, Construction of age-related reference centiles using absolute residuals. Statistics in
Medicine, 12,917-924.
Berkey, C.S., 1982, Comparison of two longitudinal growth Models for preschool children. Biometrics,
38,221-234.
Berkey, C.S., Dockery, D.W., Wang, X, Wypij, D., and Ferris, B., 1993, Longitudinal height velocity
standards for US adolescents. Statistics in Medicine, 12,403-414.
Bemis, C., 1976, Sobre el aumento secular de la estatura en Espana. Trabajos de Antropologia, XVIII,
27-32.
Bemis, C., Ochoa, F., Perez, C., Prado, C., Gomez, A., Sandin, M., Sariflena, V., and Varea, C., 1984,
Crecimiento y Desarrollo en niflos rurales de Segovia. Boletin de la Sociedad Espanola de
Antropologia Biologica, 5,7-18.
42 CHAPTER 3
Bock, R.D., and Thissen, D., 1980, Statistical problems of fitting individual growth curves. In Human
Physical Growth and Maturation: Methodologies and Factors, edited by F.E. Johnston, A.F. Roche
and Susanne (New York: Plenum) pp.265-290.
Box, G.E.P., and Cox, D.R, 1964, An analysis of transformation. Journal of the Royal Statistical Society.
Series B, 26, 211-252.
Cameron, N., Tanner, J.M., and Whitehouse, R.H., 1982, A longitudinal analysis of the growth of limb
segments in adolescence. Annals of Human Biology, 9, 211-220.
Codina, M., 1983, Crecimiento y asimetrias en varones barceloneses de edad escolar (6-16 aiios). Ph.D.
Thesis, Universidad Autonoma de Barcelona. Spain.
Cole, T.J., 1988, Fitting smoothed centile curves to reference data. Journal of the Royal Statistical
Society. Series A, 151,385-418.
Cole, T.J., 1990, The LMS method for constructing normalized growth standards. European Journal of
Clinical Nutrition, 44, 45-60.
Cole, T.J., Freeman, J.V., and Preece, M.A., 1995, Body Mass Index reference curves for the UK.
Archives of Disease in Childhood, 73, 25-29.
Cole, T.J., Green, P.J., 1992, Smoothing reference centile curves: the LMS method and penalized
likelihood. Statistics in Medicine, 13,2359-2367.
Count, E.W., 1943, Growth patterns of human physique: an approach to kinetic anthropometry. Human
Biology, 15, 1-32.
Dasgupta, P., Das, S.R, 1997, A cross-sectional growth study of trunk and limb segments of the Bengali
boys of Calcuta. Annals of Human Biology, 24, 363-369.
De la Puente, M.L., Canela, J., Alvarez, J., Salleras, L., and Vicens-Calvet, E., 1997, Cross-sectional
growth study of the child and adolescent population of Catalonia (Spain). Annals of Human
Biology, 24, 435-452.
Deming, J., 1957, Application of the Gompertz curve to the observed pattern of growth in length of 48
individual boys and girls during the adolescent cycle of growth. Human Biology, 29, 83-122.
Gonzalez, A., 1997, Antropologia del crecimiento en la poblacion escolar de la Villa de Bilbao.
Variacion antropometrica e infuencias ambientales. Ph.D. Thesis, Universidad del Pais Vasco -
Euskal Herriko Unibertsitatea. Spain.
Hauspie, R.C., 1989, ModeJes structurels dans l'analyse de la courbe de croissance. Anthropologie et
Prehistoire, 100, 123-128.
Healy, M., Rasbash, J., and Yang, M., 1988, Distribution-free estimation of age-related centiles. Annals
of Human Biology, 15, 17-22.
Hernandez, M., Castellet, J., Narvaiza, J.L., Rincon, I.M., Ruiz, I., Sanchez, E., Sobradillo, B., and
Zurimendi, A, 1988, Curvas y Tablas de Crecimiento. Instituto de Investigacion sobre Crecimiento
y Desarrollo, Fundacion F. Orbegozo (Madrid: Garsi).
Hoey, H.M.C.V., Tanner, J.M., and Cox, L.A, 1987, Clinical growth standards for Irish children. Acta
Paediatric a Scandinavica, Supplement, 338, 3-31.
Jenss, RM., and Bayley, N., 1937, A mathematical method of studying the growth of a child. Human
Biology, 9, 556-563.
Jolicoeur, P., Pontier, J., Pernin, M.O., and Sempe, M., 1988, A lifetime asymptotic growth curve for
human height. Biometrics, 44, 995-1003.
Largo, R.H., Gasser, T., Prader, A, Stuetzle, W., and Huber, P.J., 1978, Analysis of the adolescent
growth spurt, using smoothing spline functions. Annals of Human Biology, 5, 421-434.
Ledford, A.W., and Cole, T.J., 1998, Mathematical Models of growth in stature throughout childhood.
Annals of Human Biology, 25, 101-115.
Martinez, R, Nielsen, A.H., Prado, C., Acevedo, P., Cuesta, R., and Gomez-Lobo, P., 1994, Aplicacion
del modelo de Preece-Baines-I para los datos transversales del crecimiento de la talla en una
poblacion rural (Cuenca, Espana). In Biologia de las poblaciones humanas: problemas
metodologicos e interpretacion ecologica. Actas del VIII Congreso de la Sociedad Espanola de
Antropologia Biologica, edited by C. Bemis, C. Varea, F. Robles and A. Gonzalez (Madrid:
Ediciones de la Universidad Autonoma de Madrid) pp. 671-681.
Marubini, E., Resele, L.F., and Berghini, G., 1971, A comparative fitting of the Gompertz and logistic
functions to longitudinal height data adolescence in girls. Human Biology, 43, 237-252.
Moreno, P., and Carrie., R., 1993, The height and weight of girls in Barcelona (Spain): Longitudinal
data. Acta Medica Auxologica, 25, 59-66.
Pan, H., and Goldstein, H., 1997, Multi-level models for longitudinal growth norms. Statistics in
Medicine, 16, 2665-2678.
SMOOTHING CENTILES BY 1HE LMS ME1HOD 43
Prado, C., Martinez, R., and Nielsen, A.H., 1985, Panimetros longitudinales y transversales en la
poblacion rural y urbana de Cuenca: Estudio ontogenetico y dimorfismo sexual. Actas del IV
Congreso Espafiol de Antropologia Biologica-1. (Barcelona: Ediciones de la Universidad de
Barcelona) pp. 243-252.
Preece, M.A., and Baines, M.J., 1978, A new family of mathematical Models describing the human
growth curve. Annals of Human Biology, 5, 1-24.
Rebato, E., Gonzalez, A., and Rosique, J., 1993, Application du modele I de Preece-Baines (PBI) it
l'etude de la croissance staturale chez deux populations biscaiennes. Bulletin et Memoires de la
Societe d' Anthropologie de Paris, 5,93-102.
Rosique, J., 1992, Estudio transversal del crecimiento en escolares vizcainos. La variacion
antropometrica como componente de la estructura biologica de la poblacion. Ph.D. Thesis.
Universidad del Pais Vasco - Euskal Herriko Unibertsitatea. Spain
Rosique, J., and Rebato, E., 1995a, Comparative study of statural growth in Spanish populations.
American Journal of Human Biology, 7, 553-564.
Rosique, J., Rebato, E., 1995b, Modelizacion del desarrollo somatico diferencial de una muestra de
chic os y chicas de la costa de Vizcaya. Cuadernos de Seccion. Antropologia-Etnografia, 13, 273-
285.
Rosique, J., Gordon, P., Rebato, E., Gonzalez-Montero de Espinosa, M., Callejo, L., Moreno-Heras, E.,
and Marrodan, M.D., 2000, Aplicacion del modelo I de Preece-Baines al estudio auxologico de
muestras contemporaneas e historicas de la poblacion madrileiia. Investigaciones en Biodiversidad
Humana, edited by T.A. Varela. (Santiago de Compostela: Servicio de Publicaciones e Intercambio
Cientifico, Universidad de Santiago de Compostela).
Ruiz, I., 1989, Estudio longitudinal del crecimiento en la etapa puberal en una muestra de escolares del
Gran Bilbao. Ph. D. Thesis. Universidad del Pais Vasco-Euskal Herriko Unibertsitatea. Spain.
Sandin, M., 1980, Consecuencias biologic as de la emigracion: analisis en poblacion escolar extremeiia.
Ph. D. Thesis. Universidad Autonoma de Madrid. Spain.
Shohoji, T., and Sasaki, H., 1987, Individual growth of Japanese. Growth, 51,452-460.
Sinclair, D., 1989, Human Growth after Birth, Fifth edition (Oxford: Oxford Medical Publications) p.27.
Tanner, J.M., and Davies, P., 1985, Clinical longitudinal standards for height and height velocity for
North American children. Journal of Paediatrics, 107, 317-328.
Tanner, J.M., and Whitehouse, R.H., 1976, Clinical longitudinal standards for height, weight, velocity
and stages of pUberty. Archives of Disease in Childhood, 51, 170.
Tanner, J.M., Hayashi, T., Preece, M.A., and Cameron, N., 1982, Increase in length of leg relative to
trunk in Japanese children and adults from 1957 to 1977: comparison with British and with Japanese
Americans. Annals of Human Biology, 9,411-423.
Thissen, D., and Bock, R.D., 1990, Linear and nonlinear curve fitting. In Statistical Methods in
Longitudinal Research, Vol. 2, Time Series and Categorical Longitudinal Data, edited by A von
Eye (New York: Academic Press) pp. 289-318.
Vercauteren, M., 1984, Evolution seculaire et normes de croissance chez des enfants belges. Bulletin
de la Societe royale beige d' Anthropologie et de Prehistoire, 95, 109-123.
Wachholder, A., and Hauspie, R.C., 1987, Clinical standards fro growth in height of Belgian boys and
girls, aged 2 to 18 years. International Journal of Anthropology, 1,327-338.
Weiner, J.S., and Lourie, J.A., 1981, Practical Human Biology (London: Academic Press).
Welch, Q.B., 1970, Fitting growth and research data. Growth, 34, 293-312.
Wember, T., Goddemeier, T., and Manz, F., 1992, Height growth of adolescent German boys and girls.
WHO, 1978, A Growth Chart for International Use in Maternal and Child Health Care: Guidelines for
Primary Health Care Personnel (Geneva, Switzerland: World Health Organization).
WHO, 1979, Measurement of Nutritional Impact. WHOIFAP179.1 (Geneva, Switzerland: World Health
Organization).
Zemel, B.S., and Johnston, F.E., 1994, Application of the Preece-Baines growth Model to cross-
sectional data: Problems of validity and interpretation. American Journal of Human Biology, 6, 563-
670.
Zoppi, G., Bressan, F., and Luciano, A., 1996, Height and weight reference charts for children aged 2-
18 years from Verona, Italy. European Journal of Clinical Nutrition, 50, 462-468.
CHAPTER 4
A NEW INTERNATIONAL GROWTH REFERENCE FOR

YOUNG CHILDREN
M. DE ONISt, C.G. VICTORAt, C. GARZA~, E.A.J.R. FRONGILLO~, T.

COLE*
for the WHO Working Group on the Growth Reference Protocol
tDepartment of Nutrition for Health and Development. World Health Organisation,

Geneva, Switzerland, /Post-Graduate Programme in Epidemiology, Universidade
Federal de Pelotas. Pelotas, Brazil. ~Division of Nutritional Sciences, Cornell
University, Ithaca, New York, USA, *Institute of Child Health, London, UK
1. INTRODUCTION
Anthropometric references for infants and young children are among the most used
tools for assessing paediatric well-being. At the population level growth references
are useful for predicting general emergencies related to food and nutrition, assessing
the equity of distribution of economic resources within and between communities,
evaluating the suitability of general weaning practices, and for screening and
following at-risk groups. For individuals, growth references are the mainstay to
growth monitoring and promotion, help identify the optimal timing of the
introduction of complementary foods, often are used to assess maternal lactation
performance, are applied in the detection of growth faltering, and help in the
diagnosis of failure to thrive and excessive growth.
Recently, several concerns were raised regarding the adequacy of currently
existing growth references, particularly that developed by the National Center for
Health Statistics (NCHS) and currently supported by the World Health Organisation
(WHO 1995a). The NCHS data for the first two years of life originated from a
single, ethnically homogeneous North American community where most babies
were artificially fed, while recent research shows that babies following current feeding
recommendations show different growth trajectories (WHO 1995b) while presenting
excellent health status. Additional concerns with the NCHS-WHO curves were that
measurements were taken at wide intervals and that curve-fitting procedures were
outdated. In 1995, the WHO Expert Committee on "Physical Status: the use and
interpretation of anthropometry" recommended the replacement of the NCHS-WHO
reference data with a new international growth reference. The main conclusions of the
Expert Committee have been summarised in an earlier publication (de Onis and
Habicht 1996).
45
P. Dasgupta and R. Hauspie (eds.), Perspectives in Human Growth. Development and Maturation, 45-53.
46 CHAPTER 4
A WHO Working Group was assembled to develop the protocol for the new
study, which is aimed at building a new international growth reference for assessing
the growth status of populations and of individual children. The present chapter
summarises the main features of this new study. The full protocol (WHO 1997) is
available from the authors on request.
The new reference is intended for assessing the growth of singleton children
under five years of age, throughout the world. Its basic assumption is that infants
following current feeding recommendations are growing optimally. The curves may
therefore be considered as prescriptive (or optimal) standards, as opposed to
traditional normative (or descriptive) standards based on samples of children,
regardless of feeding or other behaviours.
A major concern when proposing a reference based on recommended practices is
how such restrictions may affect other characteristics of the sample. The less
common the recommended practices are in the society where the reference data are
being collected, the more it is likely that other, possibly unmeasurable selection
forces will also be operating. If a reference population is highly homogeneous, the
distribution of values will be unacceptably narrow, resulting in cut-off values closer
to the mean than would occur given an appropriately heterogeneous reference
population. Measures to avoid the selectivity problem will be further addressed
below.
The new curves are aimed at both population and individual uses. Population
uses will include providing a reference for comparing groups of children, either in
terms of means (or medians) or of the proportions of children below or above a given
cut-off point. Individual uses include screening for attained size (weight, length, etc)
on a single occasion, and monitoring growth over time.
2. STUDY DESIGN
The WHO Multicentre Growth Reference Study (MGRS) will include a longitudinal
study from birth to 24 months of age and a cross-sectional study of children aged 18
to 71 months. In the longitudinal study, cohorts ofnew-borns will be followed up
for the first two years of their lives, with frequent assessments of feeding practices
and growth. It will also allow the assessment of self-selection bias (see below) and
provide incremental measurements. Children will be identified at birth and visited at
weeks 2, 4, 6, 8 and then monthly until their first birthday. In the second year,
bimonthly measurements will be taken.
After two years of age, growth is considerably more linear than earlier, and
feeding patterns appear to be no longer critical for growth curve construction. A
cross-sectional design will therefore be adopted for children aged 18 to 71 months,
to avoid the time and cost of carrying a longitudinal study in that age range. Using
18 months as the lower age limit for this group will allow an overlap between the
two studies. Although the curves will be built for children aged up to 59 months, it
is necessary to extend data collection to 71 months in order to obtain reliable
estimates of growth at the end of the fifth year oflife (see Sample Size section).
INTERNATIONAL GROWTH REFERENCE FOR YOUNG CHILDREN 47
3. POPULATION SAMPLES
There will be different sets of eligibility criteria for subpopulations and for individual
children.
3.1 Eligibility of study subpopulations:

The following characteristics will be used for selecting potential study sites:
• Socio-economic status that does not constrain growth. This restriction will
ensure that the curves reflect the true growth potential of children, not being
constrained by factors related to maternal nutrition, to the quality of
complementary foods, nor to other social or environmental restrictions.
• Low altitude. Centres located at an altitude above 1,500 m will not be included.
• Low mobility of the target population. The need to follow up children
prospectively for two years renders populations with high migration rates
inappropriate. Ideally, previous studies carried out in the area will show that it is
possible to follow up at least 80% offamilies with children for a 12-month period.
• Minimum of 20% of mothers willing and able to follow feeding
recommendations. This refers to the issue of selectivity. Ideally, investigators
should be able to demonstrate that 20% of eligible mothers already follow the
recommendations. If this cannot be demonstrated, investigators should satisfy the
Steering Committee that this goal is likely to be achieved during the actual study.
• Existence of breast feeding support system. Breast feeding support will ensure
lower selectivity by allowing a large proportion of mothers willing to breastfeed to
actually do so.
• Local presence of collaborative institutions. Data collection requires expertise in
both epidemiology - particularly in longitudinal studies - and anthropometry. In
addition, adequate know ledge of the local epidemiology of breast feeding and of
growth deficits is essential for selecting children for the study and for
understanding selection pressures.
It is not necessary for the whole population from the study area to have the above
characteristics, since these restrictions would probably preclude the participation of
any centre outside developed countries. These characteristics, however, should be
present among the subpopulation from which study participants will be drawn. For
example, the upper classes (and possibly part of the middle classes) would qualify in
a less developed country, while most of the population would qualify in a developed
setting. Since both the longitudinal and cross-sectional studies will be carried out in
the same sites, the subpopulation eligibility criteria apply to both.
The mean birth weight in the target population was not included as an eligibility
criterion. However, this was taken into account when selecting sites, in order to
ensure a spread of mean birth weights across centres.
3.2 Eligibility of individuals:

Within each selected site, the following characteristics will be applied for selecting
new-borns:
48 CHAPTER 4
• No health, environmental or economic constraints on growth. The issue of

economic and environmental liabilities has been discussed above. Local criteria
will be used in each site to define adequate socio-economic status. In addition,
children should not have any illnesses affecting growth.
• Mother willing to follow feeding recommendations. Mothers willing to comply
with current feeding recommendations (see below).
• Term birth. Gestational age at birth between 37 and 42 full weeks.
• Single birth.
• Maternal smoking. Since smoking can affect both lactation performance and infant
growth (Meyer 1979; Vio, Salazar and Infante 1991; Mansbach, Greenbaum and
Sulkes 1991), as well as birth weight (Kramer 1987), mothers who smoked either
before or after delivery are excluded from the study.
Low birth weight babies born at term are not excluded from the sample. This
restriction would artificially distort the lower centiles of the curve in the early
months.
4. COMPLIANCE WITH FEEDING RECOMMENDA nONS

The study takes the breast-fed infant as the biological "norm". Given that strict
adherence to the current feeding recommendations might cause important selectivity
problems with the cohort to be followed, and also due to economic and logistic
considerations, the following operational criteria are applied in the study:
• Exclusive or predominant breast feeding for at least four months (120 days);
• Introduction of complementary foods by the age of six months (180 days) and
• Partial breast feeding to be continued for at least 12 months (365 days).
To increase compliance, lactation counselling is an essential part of the study. In
each site, trained counsellors visit all participating mothers frequently in the first
months after delivery, to help successful breast feeding initiation and to advise on
eventual problems. The first visit takes place within 24 hours from delivery, and
subsequent visits at 7, 14 and 30 days, and monthly thereafter until at least the sixth
month. Additional visits are carried out if required. A 24-hour hotline is made
available to mothers, for emergency support.
The restriction of the study to subpopulations with a high socio-economic status
and no evidence of growth retardation suggests that complementary foods will be
generally adequate. Nevertheless, the lactation counsellors also advises mothers on
complementary foods with emphasis on energy density, feeding frequency and micro-
nutrient content. Compliance will be assessed by the types of foods given to each
individual child. Food frequency data are collected at all visits. In particular, the
volumes of non-breast milk fluids given and the use of vitaminim inera I supplements
are being assessed.
In addition to compliance with feeding recommendations, the following criteria
are applied for excluding mother-infant pairs from the growth-curve set of the
longitudinal study:
• Serious illness. Children developing a serious illness leading to substantial
growth faltering will be excluded from the growth-curve set of the study.
• Age limit. Children will be discontinued from the longitudinal study when they
complete two years of age.
• Voluntary exclusion. At any time, mothers not willing to comply with the visit
schedule may leave the study. The reasons for the refusal will be recorded
Children who are excluded from the growth-curve set of the study, due to lack of
compliance or serious illness, are continue to be visited as scheduled, so that
information will be available on their growth. Mothers who refuse to continue are
being asked if they would allow one more visit on the next birthday of the child.
These visits will allow to document how the growth of children who remain in the
study may differ from the growth of other children in the target subpopulation.
5. ANTHROPOMETRIC MEASUREMENTS
Anthropometric measurements taken in the longitudinal study include weight,
crown-heel length and head circumference, arm circumference, and triceps and
subscapular skinfold thicknesses. The timing for these measurements is described in
Table 1. More frequent data collection takes place during the first year - when
growth velocity is greater - with a total of 21 visits per child, including the first
visit within the first 24 hours of life.
Table 1. Time schedule for the collection of anthropometric measurements in the

longitudinal study.
Measurement Time frame Frequency No. o/visits

Weight, length, head
circumference Birth Once
Weeks 2-8 Biweekly 4
3-12 months Monthly 10
14-24 months Bimonthly 6
Arm circumference, 3-12 months Monthly 10

skinfold thicknesses
14-24 months Bimonthly 6
In the longitudinal study, measurements are obtained with a maximum delay of

10% of the child's age (e.g., three days at one month, or 18 days at six months). In
the cross-sectional study, children aged 18-71 months have their weight, standing
height, head and arm circumference, and triceps and subscapular skinfolds measured
once. In addition, children aged 18-35 full months also have their supine length
measured. This will allow the precise estimation of the lengthiheight disjunction. It
is envisaged that the fmal curve will have a clear disjunction at 24 months to warn
the health worker to move from length to height, thus avoiding some of the
problems observed with the NCHS-WHO reference.
The anthropometric instruments and techniques used in all centres are
standardised, i.e., similar equipment and measuring procedures are used by all study
teams. Children are weighed with the Uniscale (UNICEF, Copenhagen), an
electronic scale with precision of 100g that obtains the child's weight by
subtraction, by weighing first the mother and then the mother holding the child.
50 CHAPTER 4
Length is measured with Harpenden Infantometers with digit counter readings to

1 mm. The Harpenden Portable Stadiometer with digit counter reading is being
used for measuring both parental and children's height. Circumferences are measured
with a flat metal tape 0.7 cm wide calibrated to 1 mm. For skinfolds, the Holtain
Tanner/Whitehouse calliper is being used Uaw face area: 35 mm2; pressure between
the jaws: 1O±2 gimm2 ). All equipment is calibrated daily.
Teams are standardised by an international anthropometry expert who visits
every participating site prior to the beginning of data collection to train all field
workers. In addition, the international expert visits every site annually. Every two
months, a standardisation session is held in each study site, in which 10 children
are measured twice by each observer, for all measurements. The purpose of the
bimonthly standardisation is to ensure the observers are not departing from the
recommended techniques, monitor their reliability, and to take corrective measures if
required. Observers showing inadequate intra- or inter-observer agreement are
retrained. This also allows reliability statistics to be monitored within and between
study sites.
During the study, every anthropometric measure is carried out twice, once by
each of a pair of observers, who record their results independently. Each set of two
measures is then compared, and if the difference exceeds the maximum allowed
difference, both measurements are repeated until the measurements are within the
allowable difference. If there is no agreement after three sets of duplicate
measurements, the process is stopped and the last measurements are recorded.
6. QUESTIONNAIRES, DATA MANAGEMENT AND QUALITY CONTROL

All questionnaires include closed questions with pre-coded answers when applicable.
Questionnaires are centrally prepared in English and translated to the local
languages. These versions are then back-translated into English for comparison with
the original forms, and any discrepancies corrected. Detailed interviewers' guides
with guidance on all questions are prepared for training and field use.
Data entry takes place simultaneously with data collection at each study site. To
standardise data entry, centrally-prepared data entry routines automatically carry out
range and consistency checks for immediate correction. All records are entered twice
for validation purposes. Data being collected is transferred monthly to the co-
ordinating centre, located in the Department of Nutrition of WHO, where further
quality control analyses take place and compliance with the study protocol is
assessed.
The main measures to ensure data quality control include:
• Use of pre-tested, standardised data collection forms and detailed interviewer
guides
• Translation into local languages and back-translation of questionnaires and other
forms
• Careful selection and evaluation of interviewers
• Thorough training course on interviewing techniques
• Training workshop on data management
• Training on anthropometric measurements by a single expert in all sites, followed
by initial standardisation sessions with assessment of intra- and inter-observer
variability
• Frequent calibration of measuring equipment

• Bimonthly standardisation sessions throughout the data collection period
• Repetition of 5-10% of all interviews
• Simultaneous data entry with centrally-prepared range/consistency checks
7. SAMPLE SIZE
The precision of growth chart centiles is determined by several factors, of which the
most important is sample size. Other factors are also relevant, including study
design (cross-sectional versus longitudinal), the timing of measurements, and the
method of curve-fitting. Many criteria can be used to set the sample size, but four
were considered in the present study. These were a) the precision of a given centile
at a particular age, b) the precision of the slope of the median curve over a given age
range, c) the precision of the median curve overall and the influence of data at
particular ages, and d) the precision of the correlation between measurements in the
same subjects at different ages. The latter criterion is relevant for velocity references.
Sample sizes were calculated for each of these four criteria, and it was found that a
sample size of 200 for the longitudinal study and 200 per 3 months for the cross-
sectional study should provide adequate precision. These are the numbers for each
sex, and are to be obtained by combining data from several sites.
A relevant finding of the sample size calculations was that the first few
measurements, particularly birth weight, have high variance, while between 1 and 4
years this is low. In addition, limiting the study to children under five years results
in increased imprecision during the fifth year. To address the imprecision of the
curve at the extremes, birth weight needs to be over-sampled and the upper age limit
raised. A four-times larger sample at birth and an upper limit of 71 full months for
the cross-sectional study will considerably improve the precision of the curve
throughout the whole age range of interest.
To obtain 400 children of both sexes, given six participating sites, 70 children
will have to fulfil all criteria in each centre. The number of infants to be initially
recruited will depend on the proportions expected to remain eligible (i.e., complying
with feeding recommendations, healthy and willing to participate) until the age of
two years. For example, if a completion rate of 30% is envisaged, then 233 infants
are to be recruited at birth (233 x 0.3 = 70). As a working figure for estimating the
logistics of the study, a sample of 300 eligible new-borns per centre was used. This
will fulfil the requirement that the sample size at birth should be at least four times
larger than the group of70 children to be followed up longitudinally for two years.
In the cross-sectional study, as for the longitudinal study, 70 children per age
range are required. As mentioned, to improve the estimates of growth in the fifth
year, it will be necessary to expand the upper age limit of the cross-sectional study
to 71 full months. The lower age limit was set at 18 months to provide some
overlap with the longitudinal study. Since three-month groupings are used, then
1260 children aged 18 to 71 full months are needed in each centre, allowing for
refusals, a sample of 1400 per centre was used. The sampling methodology for the
cross-sectional study is adapted to each site depending on local circumstances (e.g.,
a household survey in the catchment areas of the maternity hospitals included in the
study).
52 CHAPTER 4
8. POSSIBLE SOURCES OF BIAS

Building a prescriptive reference may be affected by methodological problems. This
section reviews two major sources of bias and how these will be addressed.
8.1 Dropouts and self-selection bias

The issue of selectivity was a major concern of the Expert Committee (WHO
1995a). Whenever only a small proportion of the population is eligible for entering
the study, other important selection biases may operate. To counteract this
possibility, several steps are being taken: operational definitions of the feeding
recommendations are somewhat relaxed to ensure that a greater proportion of
children are included in the study; the study is restricted to sites where 20% or more
of the mothers in the subpopulation of interest are likely to comply with these
definitions; and breast feeding support is being provided to enhance compliance.
Possible differences between children selected for inclusion in the curve and the
remaining children will be assessed by comparing the baseline measurements at
birth (weight, length and head circumference) of children selected for the study with
all children born in the same facilities as well as all children in the socio-economic
subpopulation of interest who refused to participate or were not eligible due to lack
of intent to breastfeed. In addition, a sub-sample of the latter group of children is
visited at 12 months, to compare their attained weights and lengths with those of
the cohort children.
Finally, the growth of children who complied with the feeding recommendation
will be compared with that of children who entered the study but failed to comply.
This entails continuing to follow up to the age of two years all children included in
the study, whether or not they comply.
8.2 Variability
The Expert Committee expressed concern that selecting children who were "too
similar" (e.g., in feeding practices) might lead to low variability and therefore to
outer percentile lines that are too close to the median (WHO 1995a). This would
result in over-diagnoses of both growth deficits and excess (e.g., obesity) when the
curve is used in a general population. In order to minimise this problem, data will
be combined from different parts of the world, thus allowing the international data
set to include ethnic and environmental variability that is not present in most
existing growth curves. In addition, there will be no anthropometric restrictions on
eligibility: the subpopulations to be included will have a spread of mean birth
weights, and no anthropometric restrictions will be made on individual children,
therefore ensuring substantial variability in anthropometric measurements at birth.
Recent research showed that breast-fed infants growing in different parts of the world
have remarkably similar growth patterns and low variability compared to the current
NCHS-WHO reference (WHO 2000). This fmding confirmed the earlier results of the
Expert Committee (WHO 1995a) suggesting that concern with lack of variability
may not be as important as originally believed.
9. CONCLUDING REMARKS
The WHO Multicentre Growth Reference Study is an ambitious undertaking. By
prescribing the nature of the sample to the degree proposed by the protocol, the
recommended approach represents a significant departure from approaches used in the
past to construct growth references. This approach has been adopted because of the
stated intent to create a reference that also approximates a standard. Thus the
proposed reference should be representative of a group whose care reasonably
approaches recommended health practices. At the time of this writing data collection,
involving about 10,000 children, is well underway in six countries representing all
the major world regions: Brazil, Ghana, India, Norway, Oman and the USA. It is
envisaged that the new reference will be available in the year 2004.
10. REFERENCES
de Onis, M., and Habicht, J.P., 1996, Anthropometric reference data for international use:
recommendations from a WHO Expert Committee. American Journal Clinical Nutrition, 64, 650-
658.
Kramer, M.S., 1987, Determinants of low birthweight: methodological assessment and meta-analysis.
Bulletin of the World Health Organisation, 65, 663-737.
Mansbach, l.K., Greenbaum, C.W., and Sulkes, J., 1991, Onset and duration of breast feeding among
Israeli mothers: relationship with smoking and type of delivery. Social Science and Medicine, 33,
1391-1397.
Meyer, M.B., 1979, Breast-feeding and smoking. Lancet, 1,975-976.
Vio, F., Salazar, G., and Infante, c., 1991, Smoking during pregnancy and lactation and its effect on
breast milk volume. American Journal Clinical Nutrition, 54, 1011-1016.
WHO, 1995a, Physical status: the use and interpretation of anthropometry. Report of a WHO Expert
Committee. Technical Reports Series No. 854. Geneva, WHO.
WHO, 1995b, Working Group on Infant Growth. An evaluation of infant growth: the use and
interpretation of anthropometry in infants. Bulletin ofthe World Health Organisation, 73, 165-174.
WHO, 1997, Working Group on the Growth Reference Protocol. A growth curve for the 21st Century:
the WHO Multicentre Growth Reference Study. Geneva: WHO Department of Nutrition.
WHO, 2000, Working Group on the Growth Reference Protocol and the WHO Task Force on Methods
for the Natural Regulation of Fertility. Growth patterns of breast-fed infants in seven countries.
Acta Paediatrica, 89, 215-222.
CHAPTER 5
PELS AND TANNER-WHITEHOUSE SKELETAL AGES

OF SCHOOL CHILDREN 7-13 YEARS IN OAXACA,
MEXICO
M.E. PENA REYEst, R.M. MALINA:j:
fEscuela Nacional de Antropologia e Historia, Mexico, DF, :f: Michigan State University,
East Lansing, Ml
1. INTRODUCTION
In an earlier report, the growth and skeletal maturity of primary school children 6-14
years of age living under low socio-economic conditions in a peripheral urban
settlement (colonia) in Oaxaca, southern Mexico, was considered (Malina et al.
1976). Skeletal maturity was assessed with the Tanner-Whitehouse II (TW II) 20
bone method (Tanner et at. 1975). Mean height and weight of the children
approximated the 3rd and 10th percentiles of British reference data, but TW II skeletal
ages (SA) indicated only a slight lag or delay. About 60% of the children had skeletal
ages that lagged behind their chronological ages. The seeming discrepancy between
the growth status of the children (3rd and 10th percentiles) and the relatively slight
delay in skeletal maturity suggested, perhaps, a dissociation of growth in body size
and skeletal maturity associated with chronic mild-to-moderate undernutrition. There
is the possibility that the TW II method for the assessment of skeletal maturity may
not be sufficiently sensitive in samples living under conditions of chronic, mild-to-
moderate undernutrition. On the other hand, there may be population differences in
the rate of skeletal maturation which need to be considered in selecting a method for
assessment.
This report considers the application of the Fels method for the assessment of the
skeletal maturity of the hand and wrist in this sample of primary school children in
Oaxaca and compares skeletal ages based on the two methods. Do the two methods
of assessing maturity of the hand and wrist provide similar estimates of skeletal age
(SA) in children living under impoverished conditions? Three specific issues are
considered. First, differences in skeletal age and chronological age (SA minus CA,
SA-CA) based on the Fels method are compared to previous assessments with the
TW II method. Second, the distributions of SA-CA differences are compared. And
third, the growth status of the children plotted by CA, Fels SA and TW II SA,
respectively, are compared to United States reference data.
55
P. Dasgupta and R. Hauspie (eds.), Perspectives in Human Growth, Development and Maturation, 55-{j5.
56 CHAPTER 5
2. METHODS
Height, weight, and hand-wrist radiographs of primary school children in a peripheral
urban settlement of the city of Oaxaca in southern Mexico were studied in 1972
(Malina et al. 1976, see also Malina and Little 1981). The city of Oaxaca de Juarez,
the capital of the state, is located centrally in the Valley of Oaxaca. The population
was approximately 79,000 inhabitants in 1960 (Welte 1973) and increased to about
160,000 in the mid-1970s (Selby and Murphy 1979). With the rapid urban growth,
colonias populares, which are often initially settled by a sqm~tters' invasion on the
city's edges, accounted for a major portion of the area of the city.
Conditions in the Valley of Oaxaca at the time of the study reflected, in part,
those which characterised the State of Oaxaca in the early 1970s - a predominantly
rural state in which most of the population was poor. The economy of the state was
among the poorest in Mexico and ranked last in per capita production in 1970. The
colonias populares have been characterised as ".. vital, young, bustling, and energetic
places where a heterogeneous mix of classes and ethnic groups try to make a decent
life for themselves despite appalling public health conditions, poor nutritional
status, and a lack of those services that make urban living possible and sometimes
pleasant" (Selby and Murphy 1979, p. 8).
Most residents of the colonia were migrants from predominantly rural areas,
although the growth status of children who moved into the colonia did not differ
from those born and raised in the colonia. The community was, on average, poor
socio-economically. The median income of residents coincided with the average for
unskilled and semiskilled labour in the city of Oaxaca (Chance 1971), and living
conditions were marginal in terms of nutrition and public health amenities (Graedon
1976).
Dietary information obtained from a 24-hour recall by sixth grade children in the
colonia indicated a limited diet which was quite similar to that of a rural community
where corn, beans and chiles were the basic foods of the daily diet (pefia Reyes, et al.
1995). Compared to children in the rural community, the urban sample reported a
diet which included slightly greater intakes of protein and bread, in contrast to
tortillas.
The growth and skeletal maturity of 351 primary school children, 181 boys and
170 girls, 6 through 13 years of age, comprise the basis of this report. The data were
collected in 1972. The protocol of the study was approved by authorities in the
colonia both at the community and school levels. The total sample of school
children included a broader age range, 5 to 18 years; however, sample sizes under 6
years and older than 13 years of age were small. Radiographs of the left hand and
wrist were taken with the co-operation of a local radiologist, using a portable X-ray
unit and non-screen film. The focal distance was 76 cm with the tube centred above
the head of the third metacarpal.
All films were initially rated with the TW II 20 bone method (Tanner et al. 1975)
and subsequently with the Fels method (Roche et al. 1988). The methods of skeletal
maturity assessment are similar in principle. Both entail matching the hand-wrist
radiograph to a set of criteria, but differ in criteria for making assessments and
procedures used to construct a scale of skeletal maturity from which skeletal ages are
assigned. In the TW II and Fels methods, the film is matched to specific criteria for
each bone. The TW method uses 20 bones - the radius, ulna, seven carpals, and the
SKELETAL AGE IN MEXICAN CHILDREN 57
metacarpals and phalanges of the first, third and fifth digits of the hand. The Fels
method uses the same 20 bones, and the pisiform and adductor sesamoid of the first
metacarpal. The criteria are based upon shape differentiation of individual bones,
epiphyseal union, and the attainment of adult morphology. The Fels method utilises
ratios between linear measurements of epiphyseal and metaphyseal widths of the
long bones. Criteria for long bones in the TW method include observations of the
widths of the epiphyses and corresponding metaphyses, e.g., the epiphysis is as wide
as the metaphysis or the epiphysis is wider than the metaphysis, but ratios of linear
measurements of epiphyseal and metaphyseal widths are not used.
The methods differ in scoring. The scores for each of the 20 bones rated in the
TW II method are summed to provide a maturity score. The total score can be
converted to a skeletal age (SA), although the maturity score can be used by itself.
The carpal bones and the long bones, by definition, each contribute 50% to the total
score and in turn to the SA. The scoring system in the Fels method statistically
weighs the contributions of specific indicators depending on the sex and age of the
child. For example, epiphyseal union of the radius may occur over several years,
while the appearance of a specific shape of another bone may be present only for a
short period of time. Hence, the radius is given less statistical weight and the other
bone more weight in calculating the SA at this point in time. In addition, the Fels
method provides a standard error of the estimate for the SA.
3. RESULTS
Children were grouped into whole-year age categories, i.e., 6.00 to 6.99, etc. Sample
sizes per age group, and mean chronological ages (CA) and Fels and TW II 20 bone
SAs are summarised in Table 1. Fels SAs differ significantly (p < 0.05 and p <
0.01) from CA in all age groups of boys and girls (except 13 year old girls). In
contrast, TW II SAs do not differ significantly from CA in all age groups of girls
and do not consistently differ from CA across age groups in boys (differences are
significant, p < 0.05, in three of the seven age groups of boys).
Differences between SA and CA, i.e., SA minus CA, for each method of
assessment are illustrated in Figure 1. Within each CA group, Fels SAs lag
consistently behind CA more so than TW II SAs in both sexes. The SA-CA
differences between methods are significant except at 13 years of age in girls (p <
0.05 and p < 0.01). SA-CA differences are reasonably stable for the TW II method in
both boys and girls from 6 to 13 years. On the other hand, SA-CA differences for the
Fels method increase with age from 6-13 years in boys and tend to decrease with age
in girls after 8 years.
The extent of deviations between SA and CA with each method of skeletal
maturity assessment was also considered in the distributions around the expected,
i.e., SA = CA. Children were grouped into five categories. One category was defined
as "on time" or average with SA within ± 1 year of CA (±1 year). Two categories of
late (delayed) maturity, SA more than 1 year but less than 2 years behind CA (> -1
year) and SA more than 2 years behind CA (> -2 years), and two categories of early
(advanced) maturity, SA more than 1 year but less than 2 years in advance of CA (>
+ 1 year) and SA more than 2 years in advance of CA (> + 2 years), were also
defined.
58 CHAPTER 5
Table 1. Sample sizes and descriptive statistics for chronological ages and skeletal ages
(years) based on the Fels (Fels SA) and TW 11 (TW 11 SA) methods in Oaxaca children
Chronological age Fels SA TW II SA

n mean SD mean SD mean SD
Boys
6+ 12 6.5 0.2 6.2 0.6 6.7 0.7
7+ 20 7.5 0.2 6.5 1.1 6.9 1.3
8+ 23 8.5 0.3 7.6 1.2 8.4 1.3
9+ 33 9.5 0.3 8.1 1.2 9.2 1.4
10+ 30 10.5 0.3 9.2 1.2 10.0 1.2
11+ 20 11.5 0.3 10.3 1.5 11.4 1.5
12+ 31 12.5 0.3 10.9 1.1 12.1 1.1
13+ 12 13.4 0.3 11.8 1.5 12.9 1.3
Girls
6+ 9 6.7 0.2 5.8 0.9 6.4 1.1
7+ 18 7.5 0.3 6.8 1.3 7.6 1.3
8+ 25 8.5 0.3 7.3 1.1 8.2 1.0
9+ 32 9.5 0.3 8.6 1.4 9.3 1.2
10+ 27 10.4 0.3 9.6 1.4 10.3 1.2
11 + 24 11.6 0.3 11.0 1.4 11.4 1.2
12+ 20 12.5 0.3 11.6 1.3 12.1 0.9
13+ 15 13.5 0.3 12.9 1.5 13.1 1.6
~ 1 ~OYS • TWII SA-CA

Fels SA-CA
----€)----
<Ii
()
c::
0 \3.
-. • .." -"' .............. _...... .. ._o.,:J_ .....
~
i3
Q)
-1
«
0
~ -26
i I 1
Girls • TWII
_.. -€)-.-.
Fels SA-CA
SA-CA
(3- .... -
~.... 01 ". ...(3.........
(3 ,. ·····(3·····---e ·······.e--..... ~
i3 ·1
Q)
«
0
<i:: -2
CI) 6
Figure 1. Skeletal age-chronological age (SA minus CA, SA-CAY differences with the Fels
and TW11 methods in Oaxaca boys (top) and girls (bottom)
60 60
Boys ~ Fels
TWII ~ Boys I2l Fels
501- 6·9 years 50 10·13 years • TWII
....
Q)
40 ....
Q)
40
..c ..c
§ 30 § 30
z z
20
±, >+ > -;.:: > -, ±, >+

SA-CA Differences, years SA-CA Differences, years
Figure 2. Distribution of skeletal age-chronological age (SA-CA) differences with the Fels
and 1WIl methods in Oaxaca boys 6-9 (left) and 10-13 (right) years of age
60r 60
Girls
501- 6·9 years I • I2l Fels
TWII
~ Girls
50 10·13 years
~ 0
•
Fels
TWII
240[
-
.... 40
Q)
..c
§ 30 § 30
z z
20 ~ 20
> -;.:: > -, :t 1 > +1 >

SA-CA Differences, years
Figure 3. Distribution of skeletal age-chronological age (SA-CA) differences with the Fels
and 1WIl methods in Oaxaca girls 6-9 (left) and 10-13 (right) years of age
Distributions of SA-CA differences for the Fels and TW II methods in 6-9 and
10-13 year old age groups are illustrated in Figures 2 and 3 for boys and girls,
respectively. Overall, about 37% of boys and 50% of girls have an SA within ±1
year of CA with the Fels method. Corresponding percentages for the TW II method
are 61 % and 68%, respectively. Distributions of TW II SAs are more symmetrical
than those for Fels SAs. More Fels SAs are classified as late or delayed, but among
boys 10-13 years of age, approximately equal numbers have SAs classified as "on
time" (± 1 year), > -1 year and> -2 years. No boys have an SA > +2 years of CA
with the Fels method compared to 4% with the TW II method. About 2% of girls
have an SA > +2 years of CA with the Fels and TW II methods.
60 CHAPTERS
170_
1601-
Boys
/: P10
1SOt ~~
§ 140
E
f:::~.
•
D
t:;.
CA
FelsSA
TWIISA
100 1 I I
6 7 8 9 10 11 12 13 14
Years
Figure 4. Mean height of Oaxaca boys plotted by chronological age (CA), Fels skeletal
age (Fels SA) and TW II skeletal age (TWll SA), respectively, related to selected
percentiles of United States reference data (Kuczmarski et al. 2000)
170
Girls
_ _ _ P50
1601-
P25
150
1 /~~O
E
0 140
E
.~ 130
I
120
/'" ~""""""'-LJo_
•
I ~
CA
D FelsSA
t:;. TWIISA
6 7 8 9 10 11 12 13 14
Years
Figure 5. Mean height of Oaxaca girls plotted by chronological age (CA), Fels skeletal
age (Fels SA) and TW II skeletal age (TWll SA), respectively, related to selected
60
Boys
P50
50
P25
P10
~ 40 P5
E
Cl
.Qj
~ 30
• CA
D Fels SA
t:, TWII SA
10LI----~------L-----~----~-----L----~~----~----~
6 7 8 9 10 11 12 13 14
Years
Figure 6. Mean weight of Oaxaca boys plotted by chronological age (CA), Fels skeletal
age (Fels SA) and TW II skeletal age (TWII SA), respectively, related to selected
50 P50
Girls
40
I P25
P10
P5
Cl
..:.::
30
E
Cl
.Qj
~
10 • CA
D FelsSA
t:, TWII SA
0
5 6 7 8 9 10 11 12 13 14
Years
Figure 7. Mean weight of Oaxaca girls plotted by chronological age (CA), Fels skeletal
age (Fels SA) and TW II skeletal age (TWII SA), respectively, related to selected
62 CHAPTER 5
Mean height and weight of the children based on CA, Fels SA and TW II SA are
shown relative to United States reference values (Kuczmarski et at. 20(0) in Figures
4 through 7. With the exception of 6 year old boys, mean height of the Oaxaca
children falls below the 5th percentile of the reference data when plotted by CA
(Figures 4 and 5). When plotted by TW II SAs, mean height moves closer to the 5th
percentile, more so in boys than in girls. When plotted by Fels SAs, mean height of
boys is at or above the 10th percentile, whereas those of girls is above and below the
10th percentile.
Mean weight of the Oaxaca school children, with the exception of the sample of
6 year old boys, fluctuate between the 5th and 10th percentiles of the reference values
when plotted by CA. Among boys, mean weight plotted by TW II SAs is at or close
to the 10th percentile whereas mean weight plotted by Fels SAs is at the 25th
percentile of the reference data (Figure 6). Among girls, mean weight shift only
slightly when plotted by TW II SAs, except in the 12 and 13 year old samples, but
fluctuates between the 10th and 25th percentiles when plotted by Fels SAs (Figure
7).
- - - Ohio boys
2 r: ............... • - ••••••• Ohio girls
_ Oaxaca boys
I!?
I1l
V- .... _
--
----8---- Oaxaca girls
...... _- .... __ ._---- ....... .
--._-.
~
~
(J)
= 0
~ 1J
I-;' ••••• '21- •• -••• 0 ... ··· '21- ---. _.<:r ••••••
«(J)
CI)
-1
Qi
u..
-2
6 7 8 9 10 11 12 13 14
Chronological Age, years
Figure 8. Differences between Fels and TW II skeletal ages (Fels SA minus TW II SA) in
Oaxaca children compared to Ohio children in the Fels Longitudinal Study (Roche et al.
1988)
4. DISCUSSION
Comparisons of Fels and TW II SAs (Fels SA less TW II SA) in the sample of
primarily American White children of the Fels Longitudinal Study (Yellow Springs,
Ohio) showed that Fels SAs are consistently greater than TW II SAs at each
chronological age during childhood and adolescence (Roche et al. 1988). The
differences increased with age to 6 years in boys and to 7-8 years in girls, and then
declined. The differences were attributed to slower rates of skeletal maturation in the
standardising sample for the TW II method.
Fels SAs were more delayed relative to CA than TW II SAs in the sample of
Oaxaca children (Figure 1). Fels SAs also had distributions that were generally
skewed towards late or delayed SAs with few early or advanced SAs (Figures 2 and
3). In contrast to observations for the sample of well-off children in the Fels
Longitudinal Study, Fels SAs were consistently less than TW II SAs in each age
group of boys and girls. The contrasting trends in absolute differences between Fels
and TW II SAs (Fels SA minus TW II SA) in the Fels and Oaxaca samples are
shown in Figure 8. It appears that Fels SAs of Oaxaca boys deviate from TW II SAs
with age from 7 to 9 years and then the difference between methods is rather stable.
On the other hand, differences between Fels SAs and TW II SAs in Oaxaca girls tend
to decrease with age from 8 to 13 years.
Studies utilising the Fels method of skeletal maturity assessment of the hand and
wrist are not available for other samples of Mexican children living under conditions
similar to those of the Oaxaca sample. Both methods of maturity assessment were
applied to a sample of 50 youth soccer players, 7 to 17 years, from low to middle
class socio-economic backgrounds in two urban centres in north-central Mexico
(Pefia Reyes et at. 1994). The height and weight of the boys were, on average,
similar to a mixed-longitudinal sample of middle class Mexican boys. Mean SA-CA
differences with the Fels and TW II methods were virtually identical, -0.49 and -0.47
years, respectively, among 12 boys <11.0 years, and +0.27 and +0.30 years,
respectively, among 14 boys 11.0-12.9 years. Thus, in this small sample of active
boys from better-off conditions than the Oaxaca sample, differences between Fels and
TWII assessments of skeletal maturity were, on average, small. However, Fels SAs
were more heterogeneous - 13 (50%) of the boys had SAs within ±1 year of their
CAs, 8 (31 %) had SAs delayed by > 1 year relative to CA, and 5 (19(%) had SAs
advanced by > 1 year relative to CA. Corresponding numbers and percentages for TW
II SAs were 18 (69%),4 (15%), and 4 (15%), respectively. The trends in this small
sample of active boys must be viewed in the context of the selective nature of the
sport. Older boys (> 13.0 years) in this sample of Mexican youth soccer players (n =
24) and in a sample of elite Portuguese soccer players (Malina et at. 2(00) tend to be
advanced in skeletal maturity assessed by both methods.
Two other studies have applied the TW II 20 bone method to rural samples of
Mexican children living under conditions generally similar to those of the Oaxaca
sample, one in the north of the Mezquital Valley in Hidalgo (Cahuich and Rosado
1989) and the other in Northwest Morelos (Cervantes 1989). TW II SAs in the three
samples of Mexican children 7-13 years of age do not, on average, consistently
differ, and, perhaps more importantly in the context of the present analysis, mean
TW II SAs are not significantly delayed relative to CA. Mean height and weight of
the three samples are also similar, i.e., short and light relative to reference data for
well nourished children. This would suggest, perhaps, that skeletal maturation as
assessed by the TW II method is maintained at the expense of growth, which
suggests some degree of stunting. It is also possible that the rate of skeletal
maturation in low socio-economic, mild-to-moderately undernourished Mexican
children is similar to that of the British sample upon which the TW II method was
standardised.
64 CHAPTER 5
In the earlier analysis of the Oaxaca sample (Malina et al. 1976), it was
hypothesised that nutritional stress during infancy and early childhood may have a
more severe effect on growth in size but may have a minimal influence on skeletal
maturation when ossification centres of the hand-wrist are ossifying, i.e. "appearing"
on a hand-wrist radiograph. However, there is minimal information available for the
assessment of skeletal maturity of the hand-wrist during the first 3 or 4 years of life
when the impact of chronic, compromised nutrition is most profound. The knee, in
contrast, may be the more appropriate area for the assessment of skeletal maturity
early in life, i.e., the pre-school years (Roche, Wainer and Thissen, 1975). Data on
the skeletal maturity of the knee in young children reared under conditions of chronic
undernutrition are not available.
In the context of public health, it has been suggested that " ... comparisons of
skeletal maturity between samples of populations reveals degrees of environmental
disadvantage perhaps more finely than growth velocities ... " (Tanneret at. 1983). If
this suggestion is accepted, one can inquire why, in the case of children raised under
unfavourable conditions, is skeletal maturity assessed by the TW II method not
affected to the same extent as their growth in height? On the other hand, trends in the
Fels SAs of the Oaxaca sample suggest that skeletal maturity is in fact affected by
adverse nutritional and general health conditions during childhood. This is especially
clear in comparisons of the height of the Oaxaca sample plotted by CA, Fels SA and
TW II SA relative to United States reference data, which indicate that their small
body size is consistent with delayed Fels SAs (Figures 4 and 6). In contrast, the
relatively slight delay in TW II SAs among Oaxaca children is not commensurate
with their small body size.
In summary, the two methods for assessing skeletal maturity of the hand and
wrist, the Fels and TW II methods, do not provide similar estimates of SA in this
sample of Mexican children living under generally impoverished health and
nutritional circumstances. Fels SAs are significantly delayed relative to TW II SAs
and relative to CA, and the delay or lateness in Fels SAs is consistent with the small
body size of the children. Application of the Fels method for the assessment of
skeletal maturity to other samples living under similar conditions is needed to
confirm the observations in the present analysis.
Acknowledgements. The field research was supported in part by a summer

research grant from the Institute of Latin American Studies of the University of
Texas as Austin. Maria Eugenia Pefia Reyes was supported by a Fulbright-LASPAU
scholarship at the University of Texas at Austin at the time of the analysis.
5. REFERENCES
Cahuich, M., and Rosado, E., 1989 Los habitos alimenticios en una comunidad rural del Valle del
Mezquital. Tesis, Escuela Nacional de Antropologia e Historia, Mexico, OF.
Cervantes, M.C., 1989 Estudio de crecimiento y maduracion esqueletica en una comunidad rural. Tesis,
Escuela Nacional de Anthropologia e Historia, Mexico, OF.
Chance, J.K., 1971 Kinship and urban residence: Household and family organization in a suburb of
Oaxaca, Mexico. Journal of the Steward Anthropological Society, 2 (no. 2), 2 pp. (mimeo).
Graedon, T.L.F., 1976 Health and Nutritional Status in an Urban Community of Southern Mexico.
Unpublished Doctoral Dissertation, University of Michigan, Ann Arbor.
Kuczmarski, R.I., Ogden, c.L., Grurnmer-Strawn, L.M., Flegal, K.M., Guo, S.S., Wei, R., Mei, Z.,
Curtin, L.R., Roche, A.F., and Johnson, C.L., 2000 CDC growth charts: United States. Advance
Data from Vital and Health Statistics, no 314 (Hyattsville, MD: National Center for Health
Statistics).
Malina, R.M., Himes, J.H., and Stepick, C.D. 1976 Skeletal maturity of the hand and wrist in Oaxaca
school children. Annals of Human Biology, 3, 211-219.
Malina, R.M., and Little, B.B., 1981 Comparison of TWI and TW2 skeletal age differences in
American Black and White and in Mexican children 6-13 years of age. Annals of Human Biology,
8,543-548.
Malina, R.M., Pena Reyes, M.E., Eisenmann, J.C., Horta, L., Rodrigues, J., and Miller, R., 2000 Height,
mass and skeletal maturity of elite Portuguese soccer players ages 11-16 years. Journal of Sport
Sciences, 18, 685-693
Pena Reyes, M.E., Cardenas-Barahona, E., and Malina, R.M., 1994 Growth, physique and skeletal
maturation of soccer players 7-17 years. Humanbiologia Budapestinensis, 25, 453-458.
Pena Reyes, M.E., Malina, R.M., Little, B.B., and Buschang, P.H., 1995 Consumo de alimentos en una
comunidad rural Zapoteca en el Valle de Oaxaca. In Estudios de Antropologia Biologia, Volumen
5, R.M. Ramos Rodriguez and S. Lopez Alonso, eds. (Mexico, DF: Instituto Nacional de
Antropologia e Historia), pp. 407-414.
Roche, A.F., Chumlea, W.e., and Thissen, D., 1988 Assessing the Skeletal Maturity of the Hand-Wrist:
Fels Method (Springfield, IL: CC Thomas).
Roche, A.F., Wainer, H., and Thissen, D., 1975. Skeletal Maturity: The Knee Joint as a Biological
Indicator (New York: Plenum).
Selby, H.A., and Murphy, A.D., 1979 The City of Oaxaca. Final Technical Report, Office of Urban
Development, Technical Assistance Bureau (Washington, DC: Agency for International
Development).
Tanner, J.M., Whitehouse, R.H., Marshall, W.A., Healy, MJ.R., and Goldstein, H., 1975 Assessment of
Skeletal Maturity and Prediction of Adult Height (New York: Academic Press).
Tanner, J.M., Whitehouse, R.H., Cameron, N., Marshall, W.A., Healy, M.J.R., and Goldstein, H., 1983
Assessment of Skeletal Maturity and Prediction of Adult Height, 2nd edition (New York: Academic
Press).
Welte, e.R., 1973 Population of Selected Localidades in the Valley of Oaxaca: Censuses of 1960 and
1970 (Oaxaca de Juarez, Mexico: Oficina de Estudio de Humanidad del Valle de Oaxaca).
CHAPTER 6
ASSESSMENT OF CHILDHOOD AND ADOLESCENTS

BODY COMPOSITION: A PRACTICAL GUIDE
A. PIETROBELLI t:I:, M. HEot, M.S. FAITHt
fPaediatric Clinic, "GB. Rossi Hospital", Verona University, Verona, Italy, :t Obesity
Research Centre, St. Luke'slRoosevelt Hospital, Columbia University, College of
Physicians and Surgeons, New York, USA
1. INTRODUCTION
Quantifying the main body components is integral to the study of growth, as the
assessment of human physical characteristics is important both in the
anthropological and medical fields (Forbes 1962). The human body consists of over
thirty components, sometimes also referred to as compartments, distributed across
four main organisational levels: atomic, molecular, cellular and tissue system (Wang
et al. 1992). The sum of all components at each level of body composition is
equivalent to total body mass (Wang et al. 1995). Body composition is influenced by
endogenous and environmental factors, and is a valid indicator of pathological
changes especially when diagnosing obesity and other nutritional disorders (Le.:
bulimia and anorexia) in humans (Frisancho 1984). Studies of body composition and
various risk factors in growing children should have important implications for
preventive medicine (Forbes 1987 ).
Why is there such concern with paediatric obesity not alone with adulthood
obesity? The reason why may be that the prevalence of paediatric obesity is
increasing in the United States (Kucsmarski 1993, Troiano and Flegal 1998) and
European countries (Crepaldi et al. 1991, Organon et at. 1988). Moreover, even in
developing and newly industrialised countries, the increasing prevalence of childhood
obesity is pronounced (Rossner 1998, Mo-Suwan and Greater 1996). This rising
obesity prevalence is a concern given the health risks associated with paediatric
obesity, including elevated blood pressure, glucose intolerance, hyperinsulinemia,
dyslipidemias and cardiac disease (Faith et at. 1996, Rocchini 1993, Williams et at.
1992). Accurate assessment of body composition is important in many areas of
obesity and nutrition-related research.
The aim of this chapter is to describe and discuss methods for paediatric body
composition assessment. This chapter is broadly divided into three sections. The first
section discusses several cutting-edge methods that are primarily available to research
centres. The second section reviews low cost, practical methods for estimating body
composition. Specifically, we will focus on three methods that can be used in
67
68 CHAPTER 6
everyday practice: Body Mass Index (BMI), skinfold measurements, and

bioimpedance analysis (BIA). Finally, we examine the usefulness of body
composition assessment in order to establish an individual child's health.
2. PAEDIATRIC BODY COMPOSITION METHODS IN RESEARCH SETTING
2.1. Dual energy X-ray absorptiometry (DXA)
Recent advances in techniques for measuring body composition have provided DXA
for assessment of whole-body as well as regional measurements of bone mass, lean
mass and fat mass (Mazess et al. 1992). For a more in-depth review of the physical
concept of DXA, see Pietrobelli et al. (1996).
The DXA method evolved from the simpler single-photon absorptiometry
method used to quantify bone mass in areas with little overlying soft tissue
(Cameron and Sorensen 1963). When soft tissue is minimal, photons, emitted from
a radioactive source, are attenuated almost exclusively by higher atomic weight
elements such as calcium and phosphorus in bone.
In 1981, Peppler and Mazess (1981) first introduced the concept of dual-photon
absorptiometry. Two photon energies were passed through tissue and then
attenuation could be measured. This technique has the potential to substantially
improve the feasibility of body composition analysis with children for whom other
laboratory techniques may be impractical (e.g., underwater weighing). Briefly DXA
measurement is based on the differential attenuation of two photon beams as the
various tissues of the body absorb them. DXA requires minimal co-operation from
the participant and is relatively quick (20 minutes for a child). Radiation dosages are
very low « 1 mrem) (Pietrobelli et al. 1998).
The greatest advantage of DXA may be the ability to assess regional body
composition (i.e., trunk, arms, and legs). Nutritional status of diseased individuals
can be evaluated by analysing the individual compartments of the body. To date,
several studies were performed using DXA in paediatric samples and the results have
shown encouraging evidence that DXA is an accurate method of assessing body
composition also in paediatric sample (Goran et al. 1995). Costs vary and are higher
than those for simpler methods such as BIA or anthropometry. DXA is a stationary
instrument suitable for laboratory research and at present its clinical use for soft
tissue analysis is limited (Pietrobelli et al. 1998). However, DXA offers a new
method for the study of skeletal maturation, mineral homeostasis, environmental,
and nutritional factors involved in development and growth (Lapillonne et al. 1997)
because it measures the three body compartments (i.e. bone mineral, fat mass and
lean soft tissue).
2.2 Imaging: CT and MRI

Computerised axial tomography (CT) and magnetic resonance imaging (MRI)
provide investigators opportunities to evaluate tissue-system level components in
vivo (Sjostrom et al. 1986, Foster et al. 1984). CT and MRI can produce cross-
sectional high-resolution images and multiple cross-sectional images can be used to
BODY COMPOSmON IN PAEDIATRIes 69
reconstruct various tissue volumes including total, subcutaneous and visceral adipose
tissue, skeletal muscle, brain, organs, skin, and bone. Imaging techniques are
expected to provide new insights into the physiology of intra abdominal adipose
tissue and its relation to health (Goran et at. 1995).
These techniques are widely used in research studies to evaluate lipid, hormonal
and other relations to body composition (Heymsfield et at. 1997). CT and MRI have
the ability to quantify 3-dimensional tissue volume. The images are sharp enough
for tissue boundaries to become clear (Sjostrom 1991). None of other currently
available methods can assess tissue-system level body composition components with
the same accuracy as CT and MRI. Both CT and MRI are costly tests and their
application to low-budget and large-scale epidemiological studies is limited.
3. PRACTICAL METHODS FOR ESTIMATING PAEDIATRIC BODY

COMPOSITION
3.1. Body Mass Index (BM/)

Quetelet's Body Mass Index (BMI) is an expression of relative weight to height and
is calculated as weight (W, in kg) divided by the square of height (H, in meters)
(Quetelet 1869), that is
BMI=W/H2
This index has the advantage of being inexpensive, safe, and easy to obtain. It
only requires a standard scale and an accurate way to measure height. A home
bathroom scale can provide a reliable measurement of body weight. Recently we
compared BMI estimates of body fat against body fat measured by DXA in a large
paediatric sample and found a significant association between the two methods
(Pietrobelli et al. 1998). In addition, BMI is reported in many epidemiological data
sets, making it commonly available for population-level analyses (Harlan 1993). At
the same time, a limitation of BMI is that this index cannot differentiate fat mass
from fat-free mass (i.e., lean tissue) (Lohman 1992, Ellis et al. 1999).
Weight and height reference charts are important for monitoring height and excess
weight, but do not assess nutritional status in children (Rolland-Cachera 1995).
During growth period weight increases with both age and height, and the weight
changes are mainly due to changes in stature rather than those in fatness. However,
changes in BMI reflect changes in body fatness (Rolland-Cachera 1995).
According to previous studies (Himes and Dietz 1994, Harlan 1993, Dietz and
Bellizzi 1999), body weight and height and subsequently BMI are essential
parameters to assess growth and development in children. It is suggested by some
that these parameters be measured regularly from birth to better detect deviations
from normal development (i.e. growth retardation or overweight / obesity) (Forbes
1987).
In 1997 the International Obesity Task Force (lOTF) convened a Workshop on
childhood obesity to explore the strengths and limitations of existing approaches to
the measurement of childhood obesity. The workshop concluded that BMI is a
70 CHAPTER 6
reasonable and, for clinical purposes, preferred measure of fatness in children and
adolescents. The standards used to identify overweight and obesity in children and
adolescents should agree with the standards used to identify Grade I and 2 overweight
in adults (BMI of 25 or 30, respectively) (Dietz and Bellizzi 1999).
3.2 Skin/old measurements

A long-standing method for evaluating growth and body fatness is skinfolds
measurements. Some investigators advocate anthropometric (skinfold) measurements
for practical "field" work assessments of children and adolescents (Lohman 1992).
This technique uses special callipers to grasp a skinfold which is held between the
tester's thumb and fingers to provide a measurement in millimetres for a double fold
of skin and subcutaneous fat (Lohman et at. 1988). A high-quality skinfold calliper
and a flexible measuring tape are required for data collection.
Table 1. Practical issues for taking skinfold measurements.
• To take all measurements on the right side of the body.

• To take the skin between the thumb and the index finger of the left hand.
• To lift the fold by placing the thumb and index finger approximately three inches
apart on a line that is perpendicular to the long axis of the skinfold.
• To keep the fold elevated while the measurement is taken.
• To place the jaws of the calliper perpendicular to the fold and release the jaw
pressures slowly.
• To take a minimum of two measurements at each site and averaged.
• To read the results of the calliper to the nearest measurement.
• To have a well-trained skilled assessor.
Table 1 presents some practical issues for a correct skinfold measurement.

Validity of this approach has been demonstrated in children, with two or more
skinfolds at different sites having high correlations (r = 0.65 to 0.90) with estimated
body fat from underwater weighing (Harsha et at. 1978). Age- and race-specific
equations for estimating body fat were developed by Slaughter et al. (1988) who used
multi component model reference measures. The equations use the sum of two
skinfolds (triceps + subscapular for equation 1 and triceps + calf skinfolds for
equation 2) to predict percent body fat. The prediction errors for these equations
ranged from 3.6% to 3.9% body fat. These equations may be used to assess body
composition of boys and girls 8 to 17 years of age. Janz et al. (1993) cross-validated
the Slaughter equations for girls and boys, using hydrostatic weighing to determine
total body density and subsequently derived percent body fat. Dezenberg et al. (1999)
evaluated the accuracy and precision of Slaughter equations and concluded that
Slaughter equations did not accurately predict fat mass in a heterogeneous group of
children in terms of age and race. Dezenberg et al. (1999) proposed a new
anthropometric predict equation that provides accurate estimates of body fat not only
in Caucasian but also in African-American children with a wide range of body
composition.
BODY COMPOSmON IN PAEDIATRIes 71
Rolland-Cachera (1995) notes that anthropometric measures taken at different

sites correlate differently with total-body fat and percent body fat (Roche et at. 1981).
Specifically, triceps have a better correlation with percent body fat (r =0.6) whereas
subscapular skinfolds correlate better with absolute body fat (r = 0.7).
Cross-validation and testing of anthropometric equations in independent groups is
of particular importance. Thus, it is important to identify anthropometric measures
that are robust to inter-user variability and equally reflective of body composition in
all subgroups of the population (Ooran 1998).
We need also to take into account the importance of the reliability of each
observers' measurements. For this reason Ulijaszek and Kerr (1999) suggested the
development of appropriate protocols for anthropometry training for enhancement of
accuracy and precision of measurement.
3.3 Bioimpedance analysis (BIA)
Bioelectrical impedance analysis (BIA) is an alternative technique for assessment

of body composition in clinical and population-based studies. This method is based
on a simple concept: tissues rich in water and electrolytes are much more resistant to
the passage of an electrical current than lipid-rich adipose tissue (Lukaski 1992).
Typically, four electrodes are attached to the individual during the measurement: one
each to the ankle and foot, and one each to the wrist and back of the hand
(Heymsfield et al. 1997). A small current (undetectable to the subject) is passed
trough the electrodes on the hand and foot. The electrodes on the wrist and ankle
received the current flow. Conduction of the electrical current through the body
tissues is related to the water and electrolyte content of the tissue (Heymsfield et at.
1997). It is important to note that measurement conditions are fundamental for
obtaining correct body composition estimates. Room and subject temperature, body
position, electrode placement and several other factors (e.g. eating or drinking,
dehydration, exercise, and menstrual cycle) can influence measurement and must be
standardised during measurement. Table 2 summarises the BIA measurement
procedure.
Table 2. BIA measurement procedure.
• Take on the right side of the body with the child supine on a non-conductive
surface.
• Clean the skin at the electrodes sites.
• Place the sensor electrodes on the dorsal side of the wrist and ankle.
• Place the source electrodes at the base of the second third metacarpaVmetatarsal-
phalangeal joint of the hand and foot.
• The arms and the legs of the child will be abducted approximately 45° to each
other.
A limitation of BIA is that it provides an estimate of total water, which is then

transformed into fat-free mass. For this reason, the hydration of fat-free mass must
72 CHAPTER 6
be known. It is constant in adults (73.2%) but varies in children (Fomon et at.

1982). We summarise the relative hydration of fat-free mass in children in Table 3.
Table 3. Hydratation offat-free mass in pre-pubertal children
% hydratation offat-free mass

Age, years Boys Girls
5-6 77.0 78.0
7-8 76.8 77.6
8-9 76.2 77.0
Table 4. Features that can be taken into consideration for appropriate choice of a body
composition method.
Method Cost Features

Laboratory methods
DXA Expensive Quick and simple to use. Separates bone
equipment from tissue.
($70,000) Provides data on fat distribution and
regional assessment.
Easy for participant. Requires very low x-
ray radiation.
CT Very expensive Measures tissue area in specific anatomic
locations.
Involves radiation. Not suitable for kids.
MRI Very expensive Provides excellent precision.
Available only limited on research base.
Field methods
BMI Inexpensive Simple and inexpensive. Good for health-
($20 cost for risk stratification.
measurement)
Skinfolds Inexpensive Quick and simple procedure. Useful for
($80-100 cost for large studies.
measurement) Provides information on fat pattern.
Good tool for nutritional assessment.
BIA $1,000 - $5,000 Quick and simple procedure. Useful for
large studies.
Estimates body water.
BODY COMPOSmON IN PAEDIATRICS 73
DXA: Dual energy X-ray Absorptiometry; CT: Computerised axial Tomography;

MRI: Magnetic Resonance Imaging; BMI: Body Mass Index; BIA: Bioimpedance
Analysis.
Age-specific equations have been recommended (Deurenberg et al. 1989), because

age-related differences in electrolyte concentration in the extra-cellular space relative
to the intracellular space may alter the relationship between bioelectrical resistance
and total body water (Deurenberg et al. 1990).
We may conclude that BIA, despite its limitation described above, can be used to
accurately assess the body composition of pre-pubertal children, pubertal children,
and adolescents using the appropriate equations (Slaughter et al. 1988, Houtkooper et
al. 1992).
4. CONCLUSION
There is no single body composition method, which is universally the "best" for
paediatric samples; rather, the clinicians or researcher must weigh the practical
considerations of their assessment needs with the limitations of the methods.
Regardless of which instrument is chosen to assess body composition, it is crucial
for the clinician to meticulously follow the standard guidelines (Dietz and Bellizzi
1999) and protocol associated with each method to limit measurement error. In
addition the conversion formulas and prediction equations selected must be restricted
to the populations from which they were derived and validated (Wagner and Heyward
1999).
The clinician should consider the various factors that can influence body
composition. It is important to know nutritional factors, energy intake, and
composition of the diet, nutrition and hormonal status, food preferences and
behaviour, and the influence of non-nutritional factors. When these are taken all
together with an accurate and precise body composition assessment, it may be
possible to have the possibility to control growth process and to predict adult status
in order to reduce the risk factors of various diseases. We have reported in Table 4
some features that can be taken into consideration for appropriate choice of body
composition method.
5. REFERENCES
Cameron, J.R., and Sorensen, J., 1963, Measurement of bone mineral in vivo. Science, 42, 230-232.
Crepaldi, C., Belfiore, F., Bosello, 0., Caviezle, F., Contaldo, F., Enzi, G., and Melchionda, N., 1991,
Consensus Conference ltaliana: sovrappeso, obesita' e salute (CCI S.O.S.). Annali ltaliani
Medicina Interna, 6, 359-364.
Deurenberg, P., Kusters, C.S.L., and Smith, H.E. 1990, Assessment of body composition by bioelectrical
impedance in children and young adults in strongly age-dependent, European Journal of Clinical
Nutrition, 44, 261-268.
Deurenberg, P., Van der Kooy, K., Paling, A., and Withagen, P., 1989, Assessment of body composition
in 8-11 year old children by bioelectrical impedance. European Journal of Clinical Nutrition, 43,
623-629.
Dezenberg, C.V., Nagy, T.R., Gower, B.A., Johnson, R., and Goran, M.I., 1999, Predicting body
composition from anthropometry in pre-adolescent children. International Journal of Obesity, 23,
253-259.
Dietz, W.H., and Bellizzi, M.e., 1999, Assessment of childhood and adolescent obesity. American
Journal of Clinical Nutrition, 70 (Suppl), 117-175.
74 CHAPTER 6
Ellis, K.J., Abrams, S.A., and Wong, W.W., 1999, Monitoring childhood obesity: assessment of the
weightlheight2 index. American Journal of Epidemiology, 150,939-946.
Faith, M.S., Pietrobelli, A., Allison, D.B., and Heymsfield, S.B., 1996, Prevention of paediatric obesity:
examining the issues and forecasting research directions. In Preventive Nutrition edited by A.
Bednich and R Deckelbaum (Totowa, (NJ), Humana Press) p. 471-486.
Fomon, S.J., Haschke, F., Ziegler, E.E., and Nelson, S.E., 1982, Body composition ofreference children
from birth to age 10 years. American Journal of Clinical Nutrition, 35, 1169-1175.
Forbes, G.B., 1962, Methods for determining composition of the human body. Pediatrics, 29: 477-94.
Forbes, G.B., 1987, Body composition in infancy, childhood and adolescence. In Human body.
composition: Growth, Aging, and Activity edited by G.B. Forbes (New York: Springer Verlag),
125-196.
Foster, M.A., Hutchinson, J.M.S., Mallard, J.R, and Fuller, M., 1984, Nuclear magnetic resonance pulse
sequence and discrimination of high- and low-fat tissue, Magnetic Resonance Imaging, 2, 187-192.
Frisancho, A.R, 1984, New standard of weight and body composition by frame size and height for
assessment of nutritional status of adults and the elderly. American Journal of Clinical Nutrition, 40,
808-819.
Goran, M.I., 1998, Measurement issue related to studies of childhood obesity: assessment of body
composition, body fat distribution, physical activity, and food intake, Pediatrics, 101,505-518.
Goran, M.I., Figueroa, R, McGloin, A., Nguyen, V., Treuth, M.S., and Nagy, T.R., 1995, Obesity in
children: recent advances in energy metabolism and body composition. Obesity Research, 3, 277-
289.
Harlan, W.R, 1993, Epidemiology of childhood obesity. Annals of the New York Academy of Science,
699, 1-5.
Harsha, D.W., Frerichs, RR, and Berenson, G.S., 1978, Densitometry and anthropometry of black and
white children. Human Biology, 50, 261-280.
Heymsfield, S.B., Nunez, C., and Pietrobelli, A., 1997, Bioimpedance analysis: what are the next steps?
Nutritional in Clinical Practice, 12, 201-203.
Heymsfield, S.B., Ross, R., Wang, Z.M., and Frager, 0.,1997, Imaging techniques of body composition:
advantages of measurement and new uses. In Emerging technologies for nutrition research by S.J.
Carlson-Newberry and RB. Costello (New York: National Academic Press), 127-150.
Himes J.H., and Dietz, W.H., 1994, Guidelines for overweight in adolescent preventive services:
recommendations from an expert committee. American Journal of Clinical Nutrition, 59, 307-316.
Houtkooper, L.B., Going, S.B., Lohman, T.G., Roche, A.F., and Van Loon, M., 1992, Bioelectrical
impedance estimation of fat free body mass in children and youth: a cross validation study. Journal
of Applied Physiology, 72, 366-372.
Janz, K.F., Nielsen, D.H., Cassady, S.L., Cook, J.S., Wu, Y., and Hansen, J.R, 1993, Cross-validation of
the Slaughter skinfold equations for children and adolescents. Medicine and Science in Sports and
Exercise, 25, 10780-1076.
Kucsmarski, R.J., 1993, Trends in body composition for infants and children in U.S. Clinical Review
Food Science Nutrition, 33, 375-387.
Lapillonne, A., Braillon, P.M., Delmas, P.O., and Salle, RL., 1997, Dual energy x-ray absorptiometry in
early life. Hormone Research, 48, 43-49.
Lohman, T.G., 1992, Advances in body composition assessment. Current Issues in Exercise Science
(#3), Human Kinetics Press, Champaign, IL
Lohman, T.G., Roche, A.F., and Martorell, R, 1988, Anthropometric standardization reference manual.
Champaign, IL, Human Kinetics.
Lukaski, H.C., 1992, Body composition using impedance methods. In Obesity edited by P. Bjontorp and
B.N. Brodoff (Philadelphia: J.B. Lippincott Company), 67-79.
Mazess, RB., Chesnut, C.H., McClug, M., and Genant, H., 1992, Enhanced precision with dual-energy
X-ray absorptiometry. Calcification Tissue, 33, 353-359.
Mo-Suwan, L., and Greater, A.F., 1996, Risk factors for childhood obesity in a transitional society in
Thailand. International Journal of Obesity, 20, 697-703.
Organon, R.G., Tubol, LB., Zhukovskii, G.S., Perova, N.V., Ilchenko, LN., 1988, Epidemiological
characteristics of dyslipoproteinaemia and certain other risk factors of atherosclerosis and
aeschaemic heart disease in 11- and 14-years children in different climatogeographic zones:
results of a cooperative study. Cor Vasa, 30, 248-256.
Peppler, W.W., and Mazess, R.B., 1981, Total body bone mineral and lean body mass by dual-photon
absorptiometry. I. Theory and measurements procedure. Calcif Tissue, 33, 353-59.
Pietrobelli, A., Faith, M.S., Allison, D.B., Gallagher, D., Chiumello, G., and Heymsfield, S.B., 1998,
Body mass index as a measure of adiposity among children and adolescent: a validation study.
Journal of Paediatric, 132,204-210.
PietrobeJli, A., Formica, C., Wang, Z.M., and Heymsfield, S.B., 1996, Dual energy X-ray
absorptiometry body composition model: Review of physical concepts. American Journal of
Physiology, 271, 941-951.
BODY COMPOSmON IN PAEDIA TRIes 75
Pietrobelli, A., Wang, Z.M., Formica, c., and Heymsfield, S.B., 1998, Dual energy x-ray
absorptiometry: fat estimation errors due to variation in soft tissue hydration. American Journal of
Physiology, 274, 860-866.
Pietrobelli, A., Wang, Z.M., and Heymsfield, S.B., 1998, Techniques used in measuring human body
composition. Current Opinion in Clinical Nutrition and Metabolic Care, I, 439-448.
Quetelet, L.A.J., 1869. Physique Sociale, Vol 2, C. Muquardt, Brussels.
Rocchini, A.P., 1993, Hemodynamic and cardiac consequences of childhood obesity. In Prevention and
treatment of childhood obesityedited by c.L. Williams and S.Y.S. Kimm.The New York Academy
of Sciences, 3, 46-56.
Roche, A.F., Siervogel, R.M., Chumlea, W.B., and Webb, P., 1981, Grading body fatness from limited
anthropometric data. American Journal of Clinical Nutrition, 34, 2831-38.
Rolland-Cachera, M.-F., 1995, Prediction of adult body composition from infant and child
measurements, In Body composition techniques in health and disease edited by P.W.D. Davies and
T. Cole T. (Cambridge: Cambridge University Press), 100-145.
Rossner, S., 1998, Childhood obesity and adulthood consequences. Acta Paediatrica, 87:1-5.
Sjostrom, L., 1991, A computer tomography based multicomponent body composition techniques and
anthropometric predictions of lean body mass, total and subcutaneous adipose tissue, International
Journal of Obesity, 15, 589-599.
Sjostrom, L., Kvist, H., Cederblad, A., and Tylen, U., 1986, Determination of total adipose tissue and
body fat in women by computed tomography, 40K, and tritium. American Journal of Physiology,
250,736-745.
Slaughter, M.H., Lohman, T.G., Boileau, R.A., Horswill, C.A., Stillman, R.I., Van Loon, M.D., and
Bemben, D.A., 1988, Skinfold equations for estimation of body fatness in children and youth,
Human Biology, 60, 709-723.
Troiano, R.P., and Flegal, K.M., 1998, Overweight children and adolescents: description, epidemiology,
and demographics. Pediatrics, !OI, 497-504.
Ulijaszek, S.1., and Kerr, D.A., 1999, Anthropometric measurement error and the assessment of
nutritional status. British Medical Journal, 82, 165-177.
Wang, Z.M., Heshka, S., Pierson, R.N. Jr., Heymsfield, S.B., 1995, Systematic organization of body
composition methodology: overview with emphasis on component-based methods. American
Journal of Clinical Nutrition, 61, 457-65.
Wang, Z.M., Pierson, R.N. Jr., and Heymsfield, S.B., 1992, The five level model; a new approach to
organizing body composition research. American Journal of Clinical Nutrition, 56, 19-28.
Wagner, D.R., and Heyward, V.H., 1999, Techniques of body composition assessment: a review of
laboratory and field methods. Research Quarterly for Exercise and Sport, 70, 135-149.
Williams, D.P., Going, S.B., Lohman, T.G., Harsha, D.W., Sriinivasan, S.R., Webber, L.S., and
Berenson, G.S., 1992, Body fatness and risk for elevated blood pressure, total cholesterol, and
serum lipoprotein ratios in children and adolescents. American Journal of Public Health, 82, 358-
362.
GENETIC AND ENVIRONMENTAL
FACTORS
CHAPTER 7
GENETICS OF COMPLEX TRAITS WITH

PARTICULAR ATTENTION TO FAT PATTERNING
tD.C. RAO, :j:P.P. MAJUMDER

fDivision of Biostatistics and Departments of Psychiatry and Genetics, Washington
University School of Medicine, St. Louis, USA, :f:Anthropometry and Human Genetics
Unit, Indian Statistical Institute, Calcutta, India
1. INTRODUCTION
Obesity is a major health concern throughout the world (e.g., Bouchard, 1994). Body
fat is variously measured through under water weighing or through bioelectric
impedance. While these measures provide accurate direct measurements of total body
fat or percent body fat, often highly correlated surrogates are used in epidemiological
studies, such as the sum of skinfolds measured at multiple sites or the body mass
index (BMI), computed as the weight (kilograms) over the squared height (meters).
Several indices of fat patterning are in wide use: Trunk-to-Extremity Ratio (TER,
computed as the sum of skinfolds measured close to the trunk area over the sum of
skinfolds measured in distal areas) provides a measure of relative fat patterning.
Waist-to-hip ratio and abdominal obesity (visceral plus subcutaneous fat) are also
commonly used. Ironically, the BMI is highly correlated with both total body fat as
well as measures of fat patterning. Accordingly, BMI tends to be used in most
studies because of its simplicity of measurement. Several large family studies are
currently involved in an evaluation of the genes underlying adiposity (e.g., Bouchard
et ai., 1995). Since most family studies include BMI, it provides a good example for
investigating the genetics of fat patterning. Like most quantitative traits and disease-
related risk factors, BMI is a complex trait that involves the action of multiple genes
and environmental effects. In this paper, we will review the general problems
involved in genetic studies of complex traits in general, and review some of the
known results for BMI in particular.
Genetic investigations of simple Mendelian traits have been relatively
straightforward, although investigators are familiar with potential complications
even with simple traits. In contrast, everything about complex traits has been real
difficult, and often even large scale investigations end up with frustrating results
(perhaps all that is simple about complex traits begins and ends with the spelling).
Experienced investigators realise that numerous genes and environmental factors
interact with one another to produce the traits. Not surprisingly, simple-minded
approaches have largely been unsuccessful in finding genes for complex traits and
79
80 CHAPTER 7
have generated much controversy with "conflicting" findings from multiple studies.
It is helpful to distinguish between three types of gene effects: major genes, each
with a large effect; oligo genes, each with a moderate effect; and, minor genes or
polygenes, each with a modest effect. For complex traits like fat patterning, there
may not be any major genes at all, and even if there are major genes, their effects are
likely to be mediated by other interacting determinants (other genes and/or
environments). Efforts to date have been relatively more successful in finding major
gene effects, when they exist, but finding genes of small effect sizes (oligo genes and
polygenes) has been very difficult. Numerous large scale studies are underway for
finding genes for hypertension, heart disease, asthma, psychiatric diseases, just to
name a few. Likewise, several groups of investigators are working frantically to map
genes for obesity and other body size phenotypes. Several findings are emerging,
none too compelling. Most often, these studies end up reporting suggestive levels of
evidence.
The effect-sizes of any of the multiple genes involved are likely to be rather
modest. Therefore, methodologies meant for detecting genes with large effects (major
genes) are unlikely to be very successful with complex traits (except in rare
instances), as the experience of recent years has shown. Even though the individual
gene effects may be small, interactions among the genes and environments could
make a substantial contribution to the final manifestation of the trait. Failure to
recognise and accommodate such interactions may often mask the effects of the
individual genes.
A common approach to enhance the power of any study is to utilise larger
sample sizes. The concept of multi-centre genetic studies (e.g., Higgins et aI., 1996)
is rapidly evolving as a means of generating large samples of standardised family
data. Even in pre-planned collaborations of this sort where common protocols are
used and data collection is standardised, one must remain cognisant that the frequency
and distribution of risk factors - both genetic and environmental - may well be
different among the study centres. To pool data from studies that were conducted
independently encompasses even greater challenges as there may be considerable
differences in the sampling strategy, the phenotypic measurement, the particular
genetic markers that were typed, or in the ancillary information available for
classification or phenotypic adjustment. Therefore, sometimes the data may not be
directly poolable, but it may be useful to pool the results from different studies.
Some of these issues have been considered in the development of meta-analytic
methods for pooling results from multiple linkage studies (e.g., Li and Rao, 1996;
Gu et al., 1998b).
For complex traits where we expect etiologic heterogeneity, one wishes to
maximise the signal to noise ratio by analysing the largest possible sample of
families sharing the same predominant etiologic factor(s). Strategies that enable
investigators to sort families into relatively more homogeneous subgroups are
extremely desirable. An approach that holds promise is an application of the
classification and regression trees (CART) methodology, which we refer to as "tree
linkage" method (e.g., Rao, 1998; Shannon et al., 2000; Province et aI., 200la). We
believe that a combination of the lumping (as done by pooling data from multiple
studies or through meta-analysis) and splitting (as done in the tree linkage)
approaches provides an optimal strategy for genetic studies of complex traits.
GENETICS OF COMPLEX TRAITS 81
2. VARYING CAUSES OF PHENOTYPIC VARIATION

Let us first review the causes underlying phenotypic variability. Figure 1 shows how
multiple causes, both genetic and non-genetic, and interactions among them,
contribute to the variation in a given phenotype like the body mass index (BMI).
Some investigators believe that incorporating the full complexity of causation into
our analytical models is necessary and important for finding the genes and
understanding their effects. Although that can be a powerful approach with
appropriate types of family data, such complexity often renders the models
intractable or indeterminate in most prevailing data sets. It is also important to
realise that genetic heterogeneity often underlies a complex phenotype and that gene-
environment interactions may also vary across populations or even families. Often,
lack of data on appropriate family structures and/or on the relevant interacting
determinants, like smoking and physical activity (to name just two), make it
impossible to even entertain full-blown models. Therefore, despite the awareness that
identification of important interactions involving multiple genetic and non-genetic
determinants is necessary for the detection of the very genes we are seeking, the
complex reality is often approximated by simple but feasible models as shown, for
example, at the bottom of Figure 1.
Gene
~~l~
\( /r
I p I
Figure f. Varying causes of phenotypic variation: Hypothetical model of the underlying

genetic and environmental effects giving rise to a complex phenotype like the body mass
index (BMf). Modelling approximations are shown at the bottom. P is phenotype, h2 is
genetic heritability, c2 is familial environmental component, r2 is residual, and t2 is
maximal heritability (due to both genetic and familial environmental effects) {Reproduced
with permission from D.C. Rao and Treva Rice, Path Analysis in Genetics, in "Encyclopedia of Biostatistics",
Volume 4 (Peter Armitage and Theodore Colton, eds.), 1998. Copyright John Wiley & Sons Limited}.
82 CHAPTER 7
Most often it is possible to determine the strength or magnitude of familial

resemblance, but it is difficult to resolve this into genetic versus other components.
This difficulty is reflected in certain ambiguous models, as shown at the bottom left
of Figure 1, which gives rise to the notion of maximal heritability. Maximal
heritability includes all sources of familial resemblance, including familial
environmental effects. It provides an upper bound for the total genetic effect (e.g.,
see Rice et at., 1997). It is important to note that, when data are limited to only
certain family structures like intact nuclear families, it is not possible to resolve the
effects of genes from those of the environment(s).
3. STUDY DESIGN
One can hardly overemphasise that study design is perhaps the single most important
issue in the planning of any genetic study. Some would argue that choice of analysis
methods is less important than a carefully developed study design. Feasibility of the
study, statistical power, and cost-effectiveness all depend critically on the design. It
is important that all the available information about the diseaseltrait (e.g.,
physiology, aetiology etc.) be used fully when decisions are made about the
sampling schemes, sampling units, and analytical methods. More information
should lead to better designs. The major steps involved in a study design are:
definition of the phenotype, sampling unit and the method of sampling, sample size
and power, and cost benefit analysis. Additional issues for genome wide scans
involve various genotyping issues such as the quality of large scale genotyping,
marker density, type of markers, and whether to undertake a linkage scan or an
association scan.
Although definition of the phenotype may at first seem to be a trivial issue,
some thought should be given to whether the current definition of the phenotype,
however expertly done originally, is still the right one to use in gene finding studies.
After all, our goal is one of finding the trait genes, not whether we follow a
traditional or a revolutionary approach. Different definitions of the phenotype do lead
to different results. Certain definitions tend to dwarf the signal while others might at
least have the potential to sharpen, and relatively enhance, the signal. For example,
when studying BMI, it would be prudent to take multiple measurements of both
height and weight. Using the averages will reduce external noise. Likewise, when
studying a disease like essential hypertension, it would be preferable to use early
onset families since they are more likely to be of genetic origin. For phenotypes that
are not highly reproducible, it would seem desirable to study them in smaller family
units rather than in extended pedigrees. Often, end-point phenotypes may be difficult
to define or may not be highly reproducible, but causally-related intermediate
phenotypes may be more easily handled. In such cases, it may be prudent to study
the intermediate phenotypes. Several features of any genetic study are highly
interdependent. The most critical among them are the sampling unit, the sampling
method, and the sample size. One should not be decided independently of the other
two in particular. For genome scans involving complex traits, sib pairs of one type
or another are commonly used in conjunction with model-free methods of analysis.
When using sib pairs, the total number of participants needed in a study can be
minimised by sampling larger sibships as opposed to sampling independent sib pairs
(Todorov et ai., 1997). Other more powerful sampling units such as extremely
discordant (ED) sib pairs (Eaves, 1994; Risch and Zhang, 1995), or extremely
discordant and extremely concordant (EDAC) sib pairs (Gu et at., 1996), can reduce
the sample size even more. Sampling some sibs from above the 90th percentile of a
trait distribution and other sibs from below the 30th percentile appears to provide an
optimum strategy. This includes ED sib pair(s), sib pairs both above the 90th
percentile (high concordant, HC), and sib pairs both below the 30th percentile (low
concordant, LC). The HC is analogous to the affected sib pair (ASP) method.
However, for discrete diseases, an affected individual and an unaffected sib do not
constitute an ED sib pair; they would represent a 'discordant' sib pair, not
necessarily an 'extremely discordant' sib pair. Simply discordant sib pairs do not
constitute a good design. For an extended discussion of study design issues, see Gu
and Rao (2001).
Another crucial issue is the choice of populations from which to sample families
for genetic epidemiological studies of complex phenotypes. It has been argued that
genetically simplified isolates are more informative than diverse admixed populations
for mapping genes for complex phenotypes (Wright et aI., 1999). If strong linkage
disequilibria (LD) exist between marker loci and trait loci in isolated populations,
such an approach should ease the task of gene-mapping. However, it is not clear if
any of the known isolates demonstrate the level of LD needed for gene mapping.
Finally, genetic and environmental heterogeneity may be less in isolated populations
than in conglomerate populations, thus making population isolates more attractive
for gene finding studies using approaches that do not rely heavily on LD. This also
raises a related issue about pooling data from multiple studies. While pooling
increases the sample size and hence the power, it also risks increasing the amount of
heterogeneity. It is therefore highly desirable that splitting the pooled data into
relatively more homogeneous subgroups should be a part of the overall approach to
data pooling, as outlined later.
4. METHODS OF ANALYSIS
Three classes of methods are most commonly used today for linkage analysis: the
model-based classical LOD score method (Morton, 1955); the so-called model-free
relative pair methods (see Elston and Cordell, 2001); and finally the hybrid variance
components methods (Blangero and Almasy, 1997; Province et aI., 2000).
Especially for complex traits, one lacks reasonable trait models and therefore
routine use of the (strongly) model-based LOD score method may not be always
appropriate. This realisation has given rise to the development of alternative methods
that are not based on strong assumptions about the trait inheritance. It is natural to
reason that the existence of a susceptibility gene should lead to an elevated
probability that a pair of affected siblings would inherit the same allele(s) from their
parents. Based on this premise, a class of model-free methods has been developed
based on the sharing of alleles identical by descent (IBD) among relative pairs (see
Elston and Cordell, 2001).
For quantitative traits, the first insightful method was presented by Haseman and
Elston (1972), who took the squared difference of trait values of a sib pair as the
outcome variable, and regressed it on the proportion of alleles shared IBD by the sib
pair using the model E(YjJnj) = BO + BInj- A significantly non-zero negative
84 CHAPTER 7
regression coefficient B1 implies genetic linkage to the marker. Goldgar (1990) and
Amos (1994) used maximum likelihood methods to model directly the covariance
structure of sib pairs and arrived at a variance components method that was more
powerful than the original H-E method. Fulker and Cardon (1994) extended the HE
method to estimate the location of a QTL using flanking markers by applying the
interval mapping method of Goldgar (1990). More recently, Ghosh and Majumder
(2000) have proposed another model-free method for QTL mapping based on
genome-wide scan data on sib pairs. Based on rank correlations of squared differences
of trait values of sib pairs and IBD scores at marker loci and kernel-smoothing
techniques, these authors have proposed a method and have shown that it performs
efficiently under a wide variety of scenarios. As a hybrid of the model-based and
model-free methods, the variance components method combines the strengths of both
methods and provides one of the most powerful methods for linkage analysis of
complex traits. These methods use all the data available within a pedigree, without
excluding subjects with partially missing information or producing redundancy in the
statistics by counting all relative pairs.
Methods for the analysis of associations have undergone many enhancements in
the last decade. Although candidate genes have been the primary focus of association
studies, genome-wide association scans have been proposed as a promising approach
(Risch and Merikangas, 1996). It remains to be seen to what degree the unpredictable
pattern of linkage disequilibrium proves to be a limitation for genome-wide
association scans. While the full promise of this approach remains unclear, partial
association scans in the narrow genomic regions identified by linkage seem to be
particularly attractive. For a comprehensive collection of state-of-the-art methods for
linkage and association analyses, see Rao and Province (2001).
5. LUMPING AND SPLITTING AS A STRATEGY

We believe that lumping and splitting is a very promising approach for complex
traits, whereby data are first pooled from multiple studies (pooling), followed by a
methodical approach to subdividing the pooled data into relatively more
homogeneous subgroups (splitting). This will enable a variety of analytical
approaches like meta-analysis, multivariate methods, context dependent effects, and
classification methods.
5.1 Meta-analysis
When raw data from multiple studies are not available for pooling, results from
those studies may be pooled using meta-analysis methods. The term "meta-analysis"
is used for a wide variety of statistical procedures developed for summarising results
from multiple studies (see Olkin, 1995 for a review). The application of meta-
analysis techniques to genetic studies began only recently (Li and Rao; 1996; Rice,
1998; Gu et aI., 1998b). For meta-analysis of linkage results, Gu et al. (1998b)
proposed using the proportion of alleles shared IBD at a marker locus by a sib pair
(with specified trait outcomes) as the common effect, and presented methods for
pooling results from model-free sib pair analyses. A random effects model was used
to characterise the among-study variability, and a weighted estimate of the overall
effect and the variance components were given using the weighted least-squares
method. A heterogeneity test was also proposed to assess variability among studies.
For details, see Gu et at. (2001).
5.2 Multivariate methods
Another characteristic feature of complex traits is that the effects of individual loci
often manifest in a battery of correlated traits, and this additional information can be
exploited using appropriate multivariate methods of analysis (e.g., Blangero and
Almasy, 1997; Todorov et aI., 1998; and Province et al., 2000). Full multivariate
methods for simultaneous analysis of multiple traits may be used with greater
power, as argued by Ghosh and Majumder (2001). Other multivariate methods such
as principal component analysis (PCA) and factor analysis are often used to reduce
the dimensionality of data (e.g., Bartholomew, 1987). With respect to genetic
studies, the method of PCA can be used to construct a few "summary phenotypes"
(explaining most of the variance) from a large number of correlated traits, which can
be used in turn for genetic analysis. Alternatively, the method of "principal
components of heritability" (Ott and Rabinowitz 1999) can be directly built into
variance components models such as SEGPATH (Province et at., 2000).
5.3 Regression Tree Linkage

Although analysis of aggregate samples employing sophisticated models have the
potential to uncover complex trait genes, analyses of relatively more homogeneous
subgroups should enhance gene finding. While context dependency (Turner et al.,
1999) provides one type of subgrouping, application of the classification and
regression trees methodology (Breiman et al., 1984) offers an algorithmic approach.
One way of doing this would be to subdivide the data into potentially more
homogeneous subgroups, with the expectation that a simpler model with very few
interacting determinants might suffice for analyses of individual subgroups.
An inherent attraction of the CART methodology is that it assumes that
interactions among the independent variables (the predictors) are more the rule than
the exception. The CART methods typically partition the data through a series of
binary splits using one predictor at a time. In genetic studies, CART can be used to
focus attention on those families where the signal is the greatest. This could be done
by using relevant covariate information to identify clinically and/or biologically
more homogeneous subgroups, within each of which the disease aetiology may be
more homogeneous. Applications of this methodology for linkage studies, which we
call the regression tree linkage, are in the early stages (Rao, 1998; Shannon et at.,
2000; Province et al., 2001a). Recent application of the tree linkage methods to
hypertension data is very encouraging (Province et at., 2001b).
6. GENETICS OF FAT PATTERNING

Familial resemblance for anthropometric traits has been pursued for over two
decades. As discussed earlier, familial resemblance can arise from both genetic and
familial environmental influences. Whereas any observed familial resemblance for
86 CHAPTER 7
some anthropometric traits such as height is largely genetic in origin, for many
other traits (such as body mass index or BMI, sum of skin folds at multiple sites
etc), the resemblance in families arises from the joint action of genes and
environments (e.g., Rice et at., 1995). As noted earlier, BMI (as estimated by the
body weight in kilograms over the square of height in meters) represents a good
surrogate for both adiposity and regional fat distribution. We shall briefly review
here the current status of the genetics of BML
Familial resemblance of BMI has been investigated extensively by several
investigators in multiple family studies using different analytical methods such as
familial correlations and heritability, segregation analysis, and linkage analysis based
on genome-wide scans with hundreds of anonymous markers. Heritability for BMI
has been estimated around 40% in several studies (e.g., see Rice et at., 1999a; Price
et at., 1994), which has been shown to be remarkably stable over time (Rice et at.,
1999a). That is, about 40% of the variability in BMI can be attributed to the action
of genes and familial environments. Segregation analyses in multiple family studies
have also demonstrated the existence of one or possibly two major genes with
additional sources of familial resemblance (e.g., Price et aZ., 1994; Borecki et at.,
1998; Borecki et at., 1993; Rice et aZ., 1999b). Thus, BMI appears to involve the
joint influences of one or two genes with sufficiently large effects, some minor gene
effects, and possibly some familial environmental effects.
More recently, several genome-wide linkage scans have been undertaken to
localise the genes involved. Chagnon et at. (2001a) have reported linkage evidence
for BMI to several genomic regions in the Quebec Family Study in which
segregation analysis has indicated the existence of one or two major genes (Borecki
et at., 1993). A similar genome scan was also performed in the HERITAGE family
study, and the results are reported in Chagnon et at. (2001b). The pertinent results
for BMI are presented in Table 1.
Table 1. Linkage evidence for BMI from genome-wide scans in two family studies: the
Quebec Family Study (QFS, Chagnon et al. 200Ia) and the HERITAGE Family Study
(HERITAGE, Chagnon et al. 200Ib). Evidence is shown only for those genomic regions
that yielded multipoint LOD scores ;(? 2.0 using the variance components linkage model as
implemented in SEGPATH (Province et al. 2000)
Study Genomic region LODscore

QFS IP31 (96 CM) 2.22
QFS 1Q42-43 (267 CM) 2.12
QFS 7p15.3 (41 eM) 2.81
QFS 8q23 (128 eM) 2.45
QFS 15q25-26 (101 eM) 2.35
HERITAGE 8q23.3 (120 eM) 2.00
HERITAGE lOpI5.3-15.1 (8 cM) 2.70
HERITAGE 14ql1-11.2 (26-31 eM) 2.2 - 2.4
In all, evidence suggests that as many as 6 genomic regions may be involved,

each with a LOD score ~ 2.0. Although the evidence is not compelling for anyone
of them, we should not expect high levels of evidence for complex traits for which
even the so-called major gene effects may be relatively subdued because of
interactions that have not been incorporated in these analyses. In any case, some of
the findings are indeed very interesting. For example, on chromosome 1, the leptin
receptor gene (LEPR), known to be associated with obesity, is also located in Ip31
(at 96 cM), at the same location showing linkage evidence (LOD = 2.22). Likewise,
the angiotensinogen (AGT) gene, known to be involved in hypertension, is located
in lq42-43 (at 267 cM), where the linkage evidence is quite suggestive (LOD =
2.12). If this is true, then AGT likely has a pleiotropic effect on BMI and
hypertension. Among the other findings, the one on chromosome 8 is very
interesting since evidence for linkage was found in 8q23 in both family studies (LOD
= 2.45 in QFS and 2.00 in HERITAGE). Clearly, all the evidence suggests that
there are genomic regions involved in the inheritance of BMI. One may argue that
BMI is not a biological phenotype as it is derived from two other primary variables.
However, it is not entirely meaningless to study BMI both because it is highly
correlated with adiposity and regional fat distribution and because it is one of the
easiest variables to study in any epidemiological study. Although nothing is proven
until the gene variants are found, the progress to date seems to be in the right
direction.
7. DISCUSSION
The study of human disease has fully arrived in the molecular age. The steady and
swift progress of the Human Genome Project has provided molecular tools necessary
for the genetic dissection of complex traits. Yet significant challenges remain. Many
complex traits are also common in the population, accounting for a significant
proportion of the public health burden which emphasises the importance of the
endeavour. We begin the new millennium with challenges, but with it come
enormous opportunities. This is what makes it an exciting time to be a genetic
epidemiologist.
A genetic epidemiologist can hardly be over equipped with tools, and a tool that
works in one case may not work in the next. As experienced investigators realise,
reliance on a single method of analysis, however appropriate it may seem under the
circumstances, is not very optimal. So long as our primary objective is to find genes
for complex traits, we should be willing to consider alternative strategies toward
achieving the objective. In particular, we believe that a "lumping" and "splitting"
strategy can be useful. Lumping involves pooling data and/or results from multiple
studies. Although pooling data may sometimes introduce greater heterogeneity, and
therefore may at first seem counter-intuitive, it is capable of providing substantially
more statistical power when pooling is followed by splitting. Splitting involves
subdividing the aggregate data into multiple homogeneous subgroups, and this may
be done through an application of the CART methodology or through other
multivariate clustering techniques. In this case, separate analyses within subgroups
may actually come with inherently more power on the whole. These strategies would
be particularly useful if, for example, different trait genes or subsets thereof operate
within the subgroups.
88 CHAPTER 7
Lumping and splitting assumes that no single study may be capable of

successfully dissecting the genetic architecture of a given complex trait/disease.
There is tremendous opportunity for meaningful collaborations, and this may be the
limiting factor in terms of whether or not we succeed. For such collaborations to be
productive, one must go beyond a mere willingness to share data with each other.
Only when the investigators interact actively without bars can the added benefits of
constructive synergism be pitted against the complex challenges, with the promise
that we may win at least some of the time. After all, the question is not whether
there are genes, only when and how they might be found.
Acknowledgements. This work was partly supported by a grant from the

National Institute of General Medical Sciences (GM 28719) of the National
Institutes of Health.
8. REFERENCES
Amos, C.L, 1994, Robust variance-components approach for assessing genetic linkage in pedigrees.
American Journal of Human Genetics, 54, 535-543.
Bartholomew, D.J., 1987, Latent variable models and factor analysis (Oxford: Oxford University
Press).
B1angero, J., and Almasy, L., 1997, Multipoint oligogenic linkage analysis of quantitative traits. Genetic
Epidemiology, 14,959-964.
Borecki, LB., B1angero, J., Rice, T., Bouchard, C., and Rao, D.C., 1998, Evidence for at least two loci
affecting human fatness. American Journal of Human Genetics, 63, 831-38.
Borecki, 1.8., Bonney, G.E., Rice, T., Bouchard, c., and Rao, D.C., 1993, Influence of genotype-
dependent effects of covariates on the outcome of segregation analysis of the body mass index.
American Journal of Human Genetics, 53, 676-687.
Bouchard, c., 1994, The genetics of obesity (Boca Raton: CRC Press).
Bouchard, C., Leon, A.S., Rao, D.C., Skinner, J.S., Wilmore, J.H., and Gagnon, J., 1995, The
HERITAGE family study: Aims, design, and measurement protocol. Medicine and Science in
Sports and Exercise, 27, 721-729.
Breiman, L., Friedman, J.H., OIshen, R.A., and Stone, C.H., 1984, Classification and Regression Trees
(Belmont, CA: Wadsworth International Group, Inc).
Chagnon, Y.c., Borecki, LB., Perusse, L., Roy, S., Lacaille, M., Ho-Kim, M.A., Chagnon, M., Rice, T.,
Collier, G.R., Rao, D.C. and Bouchard, c., 2001a, Genome-wide search for genes related to
adiposity in the Quebec Family Study. Submitted to Obesity Research.
Chagnon, Y.C., Rice, T., Wilmore, 1.H., Ho-Kim, M.A., Lacaille, M., Pare, C., Bouchard, C., Borecki,
LB., Gagnon, J., Perusse, L., Leon, A.S., Skinner, 1.S., Rao, D.C., and Bouchard, c., 200lb
Genome-wide search for genes related to body composition in sedentary caucasians from the
Heritage Family Study. Pending review in Journal of Applied Physiology.
Eaves, L.l., 1994, Effect of genetic architecture on the power of human linkage studies to resolve the
contribution of quantitative trait loci. Heredity, 72, 175-192.
Elston, R.c., and Cordell, H., 2001, Overview of model-free methods for linkage analysis. In Genetic
Dissection of Complex Traits, edited by D.C. Rao and M.A. Province MA (San Diego: Academic
Press) pp 135-150.
Fulker, D.W., and Cardon, L.R., 1994, A sib-pair approach to interval mapping of Quantitative Trait
Loci. American Journal of Human Genetics, 54, 1092-1103.
Ghosh, S., and Majumder, P.P., 2000, A two-stage variable-stringency semiparametric method for
mapping quantitative-trait loci with the use of genome wide-scan data on sib pairs. American
Journal of Human Genetics, 66,1046-1061.
Ghosh, S., Majumder, and P.P., 2001, Deciphering the genetic architecture of a multivariate phenotype.
Advances in Genetics, 42, 323-347.
Goldgar, D., 1990, Multipoint analysis of human quantitative genetic variation. American Journal of
Human Genetics, 47, 957-967.
Gu, C., and Rao, D.C., 2001, Optimum study designs. In Genetic Dissection of Complex Traits, edited
by D.C. Rao and M.A. Province MA (San Diego: Academic Press) pp. 439-457.
Gu, C., Province, M.A., and Rao, D.C., 2001, Meta-analysis for model-free methods. In Genetic
Dissection of Complex Traits, edited by D.C. Rao and M.A. Province MA (San Diego: Academic
Press), pp 255-272.
Gu, c., Province, M.A., Todorov, A.A., and Rao, D.C., 1998b, Meta-analysis methodology for
combining non-parametric sibpair linkage results: Genetic homogeneity and identical markers.
Genetic Epidemiology, 15,609-626.
Gu, C., Todorov, A.A., and Rao, D.C., 1996, Combining extremely concordant sibpairs with extremely
discordant sibpairs provides a cost-effective way to linkage analysis of QTL. Genetic
Haseman, J.K., and Elston, RC., 1972, The investigation of linkage between a quantitative trait and a
marker locus. Behavior Genetics, 2, 3-19.
Higgins, M., Province, M.A., Heiss, G., Eckfeldt, J., Ellison, RC., Folsom, A.R, Rao, D.C., Sprakfa,
J.M., and Williams, R, 1996, The NHLBI Family Heart Study: Objectives and design. American
Journal of Epidemiology, 143, 1219-1228.
Li, Z., and Rao, D.C., 1996, A random effect model for meta-analysis of mUltiple quantitative sibpair
linkage studies. Genetic Epidemiology, 13,377-383.
Morton, N .E., 1955, Sequential tests for the detection of linkage. American Journal of Human Genetics,
7,277-318.
Olkin, I., 1995, Statistical and theoretical consideration in meta-analysis. Journal of Clinical
Ott, J., and Rabinowitz, D., 1999, A principal-components approach based on heritability for combining
phenotypic information. Human Heredity, 49, \06-11 \.
Price, RA, Charles, M.A., Pettitt, D.J., Knowler, and W.C., 1994, Obesity in Pima Indians: Genetic
segregation analysis of body mass index complicated by temporal increases in obesity. Human
Biology, 66, 251-274.
Province, M.A., Rice, T., Borecki, LB., Gu, C., and Rao, D.C., 2000, A multivariate and multilocus
variance components approach using structural relationships to assess quantitative trait linkage via
SEGPATH. Genetic Epidemiology (In press).
Province, M.A., Shannon, W.A., and Rao DC., 2001a, Classification Methods. In Genetic Dissection of
Complex Traits, edited by D.C. Rao and M.A. Province MA (San Diego: Academic Press), pp 273-
286
Province, M.A., Shannon, W.A., and Rao, D.C., 2001b, Application of tree linkage to hypertensive
sibship data (in preparation).
Rao, D.C., 1998, CAT scans, PET scans, and genomic scans. Genetic Epidemiology, 15, 1-18.
Rao, D.C., and Province, M.A., 2000, Genetic Dissection of Complex Traits (San Diego: Academic
Press).
Rice, J.P., 1998, The role of meta-analysis in linkage studies of complex traits. American Journal of
Medical Genetics, 74, 112-114.
Rice, T., Bouchard, c., Perusse, L., and Rao, D.C., 1995, Familial clustering of multiple measures of
adiposity and fat distribution in the Quebec Family Study: A trivariate analysis of percent body fat,
body mass index, and trunk-to-extremity skinfold ratio. International Journal of Obesity, 19, 902-
908.
Rice, T., Despres, J.P., Daw, E.W., Gagnon, J., Borecki, I.B., Perusse, L., Leon, A.S., Skinner, J.S.,
Wilmore, J.H., Rao, D.C., and Bouchard, c., 1997, Familial resemblance for abdominal visceral fat:
The HERITAGE Family Study. International Journal of Obesity, 21, \024-103 \.
Rice, T., Perusse, L., Bouchard, C., and Rao, D.C., 1999a, Familial aggregation of body mass index and
subcutaneous fat measures in the longitudinal Quebec Family Study. Genetic Epidemiology, 16,
316-334.
Rice, T., Sjostrom, C.D., Perusse, L., Rao, D.C., Sjostrom, L., and Bouchard, C., 1999b, Segregation
analysis of body mass index in a large sample selected for obesity: The Swedish Obese SUbjects
study. Obesity Research, 7, 246-255.
Risch, N., and Merikangas, K., 1996, The future of genetic studies of complex human diseases.
Science, 273, 1516-1517.
Risch, N., and Zhang H, 1995, Extreme discordant sib-pairs for mapping quantitative traits in humans.
Science, 268, 1584-1589.
Shannon, W.A., Province, M.A., and Rao, D.C., 2001, Tree-based recursive partitioning methods for
subdividing sibpairs into relatively more homogeneous subgroups. Genetic Epidemiology, In press.
Todorov, A.A., Province, M.A., Borecki, I.B., and Rao, D.C., 1997, Trade-off between sibship size and
sampling scheme for detecting quantitative trait loci. Human Heredity, 47,1-5.
Todorov, A.A., Vogler, G.P., Gu, c., Province, M.A., Li, Z., Heath, A.C., and Rao, D.C., 1998, Testing
causal hypotheses in multivariate linkage analysis of quantitative traits: General formulation and
application to sibpair data. Genetic Epidemiology, 15,263-278.
Turner, S.T., Boerwinkle, E., and Sing, C.F., 1999, Context-dependent associations of the ACE I/D
polymorphism with blood pressure. Hypertension, 34, 773-778.
Wright, A.F., Carothers, A.D., and Pirastu, M., 1999, Population choice in mapping genes for complex
disease. Nature Genetics, 23, 397-404.
CHAPTER 8
SUBCUTANEOUS ADIPOSE TISSUE DISTRIBUTION

IN 7- TO 16-YEAR OLD BOYS OF CALCUTTA IN
RELATION TO SOCIO-ECONOMIC LEVEL
E. REBATOt, J. ROSIQUEt, A. PIETROBELLI:j:, M. CHATTERJEECj[, S.

CHATTERJEECj[, R. SAHA*, AND P. DASGUPTA*
t Laboratory of Anthropology, Department of Animal Biology and Genetics, University

of Basque Country, Bilbao, Spain, :f: Paediatric Unit. Policlinico G.B. Rossi, Verona,
Italy, 'f[ Anthropological Survey of India, Calcutta, India, * Anthropology and Human
Genetics Unit, Indian Statistical Institute, Calcutta, India
1. INTRODUCTION
The changes of fat distribution during the growth period are the result of changes in
body shape, proportions and body composition and can be modulated by nutrition
and several cultural and socio-economic factors present in human environments. The
distribution of body fat can be studied in different ways, for example, i) from
centripetal fat ratio or trunk extremity ratio, ii) by combining waist-hip ratio and
skinfold ratios (Fiori et at. 2000), iii) by Principal Component Analysis (PCA) on a
set of skinfolds (Mueller and Reid 1979, Ramirez and Mueller 1980), iv) from the
residuals of the regression of each log transformed skinfold on the mean log skinfold
thickness for an individual (Healy and Tanner 1981), v) from ratios of circumferences
and skinfolds (Hattori et al. 1987, Rosique et al. 1994, Rebato et at. 1998) which is
also a heuristic device able to extract fat patterns. However, when input variables are
indices or regressions of the residuals, the size factor is avoided and the shape factor
is first extracted (Baumgartner et at. 1990, Johnston et at. 1991).
One of the earliest attempts to apply a multivariate technique of factor analysis to
study the changes of fat distribution during growth was made by Hammond (1955).
Later on many authors continued to study this basic aspect, often called fat
patterning, by applying PCA, and consequently the literature on this topic has
meanwhile been abundant (Garn and Clark 1976, Johnston 1988, Malina and
Bouchard 1988, Kaplowitz et al. 1988, Deutsch et at. 1985, Mueller 1982, 1985,
1988, Bailey et at. 1985, Baumgartner et at. 1986, Malina et at. 1982, Cameron et
al. 1992, Johnston et al. 1995, Rosique et at. 1994, Robson et al. 1971, Ramirez
and Mueller 1980, Mueller and Wohleb 1981, Demarchi and Marcellino 1999).
Further, to reveal the developmental aspects of fat patterning from the auxological
perspective, normative studies have been found to be increasingly important (Norgan
1987, Johnston 1992, Bouchard et at. 1990). Some of these studies have established
91
92 CHAPTER 8
that during the period of adolescence, redistribution of body fat from the extremities
toward the trunk takes place more among the males than among females. The
phenomena has been popularly designated as the Centripetal Fat Distribution (CFD),
and sometimes Centripetal Fat Patterning, which has developed the interests of the
researchers to explain the variety of biological mechanisms involved (Bouchard
1992). Eveleth and Tanner (1990) have stated that subcutaneous fat over the limbs
specially in males, is controlled by different physiological mechanisms from those
regulating subcutaneous fat in thorax particularly during adolescence. Moreover, it
has been recently shown that risk factors for disease association with various types
of fat patterns rise during the period of adolescence (Sarria 1992). Thus generating
more information on the growth in fat patterning during adolescence in particular,
and specifically for the unstudied populations, has been suggested to be necessary for
making proper clinical comparisons (Norgan 1987). Cronk et at. (l983a,b)
emphasised that auxological studies of fat distribution are important in the
peripubertal period when changes in terms of subcutaneous fat in skinfolds are
pronounced.
Genetic factors playa relatively greater role than environment in determining the
fat pattern (Garn 1955, Norgan 1991, Malina et at. 1982, Bouchard 1988, Mueller
and Reid 1979, Selby et at. 1990). However the socio-economic status (SES) is one
of the major environmental determinants. A cross-cultural interpretation of the socio-
economic levels is needed because only in some unwesternised cultures there is a
positive view on fat accumulation in the highest socio-economic level as a sign of
well-being, while in western societies a more slim body is a sign of well-being. For
example, while among the Tokelau migrants and the Dogrib Indians the higher
socio-economic status has been found to be associated with greater CFD (Ramirez
and Mueller 1980, Szathmary and Holt 1983). Further, many workers have reported
that low SES is strongly associated with greater centralisation of subcutaneous fat in
various populations particularly from rural Columbia (Mueller 1986), some ethnic
groups of Guatemala (Bogin and Sullivan 1986, Haffner et at. 1986), Punjabi
adolescents of north India (Johnston et at. 1991), and also other populations,
predominantly from the Western World (Larsson et at. 1989, Gillum 1987, Georges
et at. 1993, Mueller 1986). On the other hand, The National Centre for Health
Statistics Report has shown a significant association of high SES with low CFD
(Georges et at. 1991). Some association between low SES and CFD are also found
to be sex-specific, as shown by Bjorntorp (1988) among the Swedish adult males and
Rebato et at. (1998) on Basque females of Spain. Thus the role of sociocultural
environment on fat patterning has still remained to be one of the important topics of
renewed interest (Georges et at. 1993). It seems that in the western and the most
westernised societies the high SES has a negative view on the accumulation of
central fat and these can influence the frequencies of CFD found among high SES
samples in these kind of populations. The research on this topic can show the
relationship between culture and biology from the Human Ecology perspective.
In spite of scanty information on fat patterning of the children and adults,
recently several recommendations have been made to continue further studies in this
direction of which some may be relevant:
i) As fat pattern has been found to be associated with diseases, ethnic specific
measures of body composition should be developed for the popUlations,
ADDIPOSE TISSUE AND SOCIO-ECONOMIC LEVEL 93
particularly of the multiethnic society, such as in India. This has been given
primary importance for knowing the limits of the anthropometric measurements
and their interpretation in the matters of public health policy instead of the
universal criteria (Solomons and Kumanyika 2000).
ii) Continued efforts should be made towards revealing the effects of various aspects
of socio-economic environment on the relative fat distribution (Wagner and
Heyward 2000).
It may however be noted that due to methodological limitations, it sometimes
becomes difficult to differentiate the effects of social class on fat distribution from
those related to ethnicity (Brown et al. 1992).
From the reviewed literature it becomes quite apparent that, except only a few, all
studies have been conducted on the populations of the Western countries.
Unfortunately studies from the developing countries are found to be virtually lacking
(Cameron et al. 1992). Further, with respect to the population of India in particular,
many studies are now readily available which have particularly investigated the
association between fat distribution and the risk of various diseases (Nirmala Reddy
1998, Das Chaudhuri and Bose 2000). Only two normative studies may be well
quoted of which one has demonstrated the effects of age and SES on fat distribution
(estimated by the PCA) of the north Indian adolescents (Johnston et al. 1991) while
the other one has analysed the fat distribution of the Jat Sikh adolescent boys of
Punjab (Kapoor et al. 1998) aged 11-18 years by using the ratios of circumferences
and skinfolds. The importance of studying the possible effect of age and socio-
economic level on the relative fat patterning, particularly central fat distribution in
the adolescent population of the state of West Bengal could broaden our knowledge
on the relationship between the cultural environment of this important region of
India and the biological characteristics of the Indian population. Thus this paper is
aimed to study the effects of some ecological determinants on central fat distribution
of the Bengali adolescent boys from Calcutta.
2. MATERIAL AND METHODS

The data set comes from the comprehensive cross-sectional growth survey of 25
body dimensions, undertaken from the Indian Statistical Institute on the Bengali
boys residing in the city of Calcutta. The study was supervised by late S.R. Das,
late K.B. Pakrasi and P. Dasgupta during the period of 1982-83. The investigation
was carried out among the students from one of the oldest Missionary schools of
Calcutta, known as The Scottish Church Collegiate School (Pakrasi et al. 1988).
Altogether 856 boys aged 7-16 years voluntarily participated in the study. The socio-
economic characteristics of the students are predominantly urban middle class
whereas the two extremes i.e the lower and upper class represents in a smaller
frequency.
Ethnically the Bengalis are reported to be a heterogeneous group who claimed to
have originated by the admixture of several major racial strains like the Caucasoids,
Proto-Australoids and Mongoloids. According to Risley (1891, 1908) the Bengalis
of the lower Bengal belong to the Mongolo-Dravidian type. The subjects of the
present study were all Hindus affiliated with different caste groups like the Brahmins,
Vaidyas, Kayasthas and other different scheduled castes. Two major data sets were
94 CHAPTER 8
collected from the participants. The first set was concerned with the anthropometry
of each subject and the second with the socio-economic and demographic
backgrounds. The subjects were distributed in SES following the classification
provided by the National Classification of Occupations of the fathers (Central
Statistical Organisation 1962).
Subjects wearing only short pant, were measured according to the protocol of the
IBP (Weiner and Lourie 1969). From the set of anthropometric measurements taken
from each subject, four skinfolds (triceps, biceps, subscapular and suprailiac) and calf
circumference were studied in the present paper. However, due to various
organisational constraints, it was not possible to take 5 measurements from all
subjects. Skinfolds were measured by a Lange skinfold calliper (mm) whereas calf
circumference was taken by a steel tape (GPM make). The measurements were taken
by one anthropometrist all through the study on and around the birth dates of the
subjects with 3 days of tolerance between 8 am and 3 pm in the school premises.
The technical errors of the five measurements were calculated from the duplicate
observations (by the formula, TEM = ~ d 2 /2N ) and the values obtained are in the
range of other studies (Ulijaszek and Kerr 1999). The corresponding values are shown
in Table 1.
Table 1. Technical error afthe measurements (TEM)
Measurements n TEM
Triceps skinfold thickness (mm) 18 0.83
Subscapular skinfold thickness (mm) 18 0.29
Biceps skinfold thickness (mm) 11 0.05
Suprailiac skinfold thickness (mm) 11 0.05
Calf circumference (cm) 6 0.08
Expenditure per month in each nuclear family was recorded by interview (accuracy
of the items 50 Rupees) in order to find, as far as possible, an objective correlate of
socio-economic status (SES). Expenditure ranged from 300 to 10,000 Rupees per
month, with a median value of 1,200 Rupees per month and family. As shown in
Table 2, the whole sample was distributed in three SES levels (SESl, SES2 and
SES3) with the help of two cut off points: 1,000 Rupees (the cut off points for the
30th percentile) and 1,500 Rupees (the cut off points for the 70th percentile). The
asymmetry of the chosen cut off points around the mean was preferred instead of
other symmetric cut-off points due to the actual positive skewed distribution of the
expenditure. This procedure could afford comparative subsample sizes of the SES
levels. Moreover, 1,000 to 1,500 Rupees were also meaningful boundaries of this
sample from Calcutta.
In two previous analyses performed on this data set (Dasgupta, unpublished), the
results (2-way ANOV A) revealed statistically significant effect of age and per capita
expenditure level on four of the five traits (three skinfolds and calf circumference)
used in this paper. In addition, the chi-square test revealed significant association
between education and occupation of father with the per capita level of expenditure of
the family. By analysing the dietary data a qualitative difference between the subjects
of the expenditure groups in terms of the consumption of the protein intakes has
been found. The subjects in the higher expenditure group consumed greater frequency
of protein than the subjects belonging in the lower expenditure category.
2.1 Data analysis
Principal Component Analysis (PCA) was used to obtain the pattern of fat
distribution. To correct for overall body fat, ratios of relative fat distribution were
included in the analysis (Hattori et al. 1987). The chosen indices have the ability of
maximising the contrast between trunk and extremity fat (Rosique et at. 1994). All
the measures were used in millimetres and transformed with decimal logarithms.
This transformation was able to normalise the distribution. Two consecutive PCA's
were performed in order to test the ability of calf circumference to describe fat
patterning when included in the analysis like skinfolds. Although four skinfolds were
employed in the PCA the performance of the same PCA with the addition of calf
circumference was also analysed.
PCA type-a: the PCA based on only four skinfolds.
PCA type-b: the PCA based on four skinfolds and calf circumference.
The PCA based on four skinfolds was performed using the following indices:
TRI = log triceps/(log subscapular + log suprailiac)
BIC = log biceps/(log subscapular + log suprailiac)
SUB = log subscapular/(log triceps + log biceps)
SUPRA = log suprailiac/(log triceps + log biceps)
The PCA based on four skinfolds and calf circumference was performed using the
following indices:
TRI = log triceps/(log subscapular + log suprailiac)
BIC = log biceps/(log subscapular + log suprailiac)
SUB = log subscapular/(log triceps + log biceps + log calf circumference)
SUPRA = log suprailiac/ (log triceps + log biceps + log calf circumference)
CAL = log calf circumference/(log subscapular + log suprailiac)
After the extraction of three principal components, the factor score for each
individual were obtained to identify patterns of fat distributions. The first factor was
used as an indicator of central body fat distribution. The effect of age on the stability
of the factors was studied by comparing the load of each skinfold index in the
extracted factors from the sample with those obtained in subsequent PCA by age.
The sample was subdivided in two groups, 7 to 12 years and 13 to 16 years. A
separate PCA was used in each age group. The stability of the components across
age was analysed by means of applying subsequent PCA by ages ranges 7 to 12
years and 13 to 16 years.
96 CHAPTER 8
2.2 Socio-economic Status (SES)
SES differences were investigated based on per month expenditure of the families as
a single socio-economic criteria. Comparison of fat distribution in three different
SES subgroups according to the family expenditure (Table 2) was undertaken by
comparing the fat factor score from both types of PCA-a and PCA-b. A two-way
MANOV A by age and SES has been used in order to test the effects and possible
interactions of age and SES.
Table 2. Distribution of the sample expenditure per month of the nuclear families. The
final sample size and the % of lost sample due to missing values when applying two
different types of peA (a and b) are also shown.
SES expenditure per month sample final sample size

(Rupees) size PCA type a PCA type b
1 less than 1,000 186 172 166
2 from 1,000 to 1,500 297 264 254
3 more than 1,500 342 311 302
total sample 825 747 722
% of lost sample 9.5% 12.5%
3. RESULTS
3.1 Fat distribution
Results of the two sets of Principal Component Analysis (PC A type-a and PCA
type-b) performed on the total sample are shown in Table 3. Three extracted
components from type-a (4 skinfolds) and type-b (4 skinfolds and calf circumference)
have explained 99.3% and 93.3% of the variances respectively. In both PCA, the
components showed relationship of shape among variables but not on size. The
skinfolds are associated differently depending on the components and the PCA type (a
or b). However, only correlation values greater than 0.3 in the components have
been considered to be biologically meaningful.
The first component showed a very high correlation with all the indices of both
types of PCA's. The components show a sharp contrast between the limb sites
(triceps, biceps and calf circumference) and the trunk sites (subscapular and
suprailiac). It may however be noted that the corresponding signs of the indices in
two PCA types differ. While in type-a the trunk sites have loaded positively, in
type-b they manifest negative loading. The first component thus have explained
68.8% and 76.6% of the variance and their respective eigenvalues are much above
1.0 (3.1 to 3.4). The component can be designated in various ways, like trunk-
extremity contrast or central-peripheral component. It may further be noted that in
both PCA types suprailiac indices have shown the highest correlational value
followed by calf indices (which is more discernible in the PCA type-b). Among the
extremity sites calf has shown the highest loading (in type-b) followed by triceps
reflecting their greater role than biceps in making the contrast. In the first
component of the PCA type-b, the biceps site has shown relatively lower correlation
of all sites in comparison to the second component.
Table 3. Correlations of the first three principal components extracted with fat
distribution ratios in the total sample (n = 787)
skinfold PCA type a PCA type b

ratio (1) (4 skinfolds) (4 skinfolds + calf circumference)
1 2 3 1 2 3
SUPRA 0.89 -0.34 (-0.30)(2) -0.91 (0.20) (-0.01)
BIC -0.88 -0.36 (0.30) 0.59 0.76 (-0.26)
SUB 0.87 0.37 0.33 -0.89 (0.12) (0.17)
TRI -0.87 0.39 (-0.29) 0.82 (0.25) 0.50
CAL 0.89 -0.42 (-0.13)
eigenvalue 3.1 0.5 0.4 3.4 0.9 0.4
% variance 76.6 13.5 9.2 68.8 17.2 7.3
(1) In PCA type a: SUPRA = log suprailiac/(log triceps + log biceps), BIC = log biceps/(log
subscapular + log suprailiac), SUB = log subscapular/(log triceps + log biceps), TRI = log
triceps/(log subscapular + log suprailiac).
In PCA type b: SUPRA = log suprailiac/(log triceps + log biceps + log calf circumference), BIC =
log biceps/(log subscapular + log suprailiac), SUB = log subscapular/(log triceps + log biceps + log
calf circumference), BIC = log biceps/(log subscapular + log suprailiac), SUB = log
subscapular/(log triceps + log biceps + log calf circumference), TRI = log triceps/(log subscapular
+ log suprailiac), CAL = log calf circumference/(log subscapular + log suprailiac).
(2) The coefficients with the lowest contribution to the biological meaning of the components
(coefficients :-; 0.30) are shown in parentheses.
In PCA type-a, biceps and triceps site get almost equal magnitude of loading
with the component 1, but due to the inclusion of calf circumference in the PCA
type-b, in the extremity contrast the contribution of biceps has been minimised.
Thus it becomes apparent that to get the trunk-extremity contrast the role of calf and
triceps should be the primary choice.
The second component of the PCA type-a shows a clear contrast between the
dorsal (subscapular and triceps) and the ventral sites (biceps and suprailiac) which can
also be designated by anterior-posterior component. On the other hand, the second
component of PCA type-b has shown a clear contrast between upper extremity site
(biceps) and lower extremity site (calf circumference). This component has explained
13.5 to 68.5% percent of the variance and their respective eigenvalues are below 1
(0.5 to 0.9). It may be noted that in making the contrast of this component biceps
played a greater role than triceps index (in type-b).
The third component of type-a is correlated with subscapular fat whereas in the
type-b analysis, it is correlated mainly with the triceps fat. It has explained 7.3 to
9.2% of the variance and its eigenvalues are also much lower than 1 (0.4). The third
component has no homogenous meaning across PCA types.
98 CHAPTER 8
With respect to the stability of the components the first component of the PCA
type-a showed a poor stability relative to the PCA of the total sample and with
respect to the patterns of the signs (Table 4). In spite of this, the first component of
the PCA type-a is still a good indicator of the centripetal-peripheral fat pattern and is
consistent across ages. The first component of the PCA type-b showed fair stability
in the pattern of signs but in a lesser degree in the relative strength of the loading.
Table 4. Correlations of the principal components with fat distribution ratios by age
skinfold PCA type a PCA type b

ratio (1) (4 skinfolds) (4 skirifolds + calf circumference)
2 3 2 3
age range 7 to 12
SUPRA -0.86 -0.41 (-0.30) (2) -0.92 (0.09) (0.07)
BIC 0.83 -0.38 0.40 0.50 0.84 (-0.19)
SUB -0.81 0.46 0.36 -0.88 (0.16) (0.13)
TRI 0.84 0.39 -0.36 0.82 (0.13) 0.55
CAL 0.91 -0.33 (-0.19)
eigenvalue 2.8 0.7 0.5 3.4 0.9 0.3
% variance 69.6 17.0 12.8 67.7 17.4 8.0
age range 13 to 16
SUPRA -0.89 0.36 (-0.27) 0.89 (0.28) (-0.19)
BIC 0.89 0.35 (0.29) 0.59 0.72 (-0.29)
SUB -0.89 -0.36 (0.28) -0.89 (0.09) (0.29)
TRI 0.89 -0.35 (-0.29) 0.76 0.43 0.44
CAL 0.85 -0.50 (-0.08)
eigenvalue 3.2 0.5 0.3 3.2 1.0 0.4
% variance 78.9 12.6 7.9 64.7 20.9 8.1
(1) In PCA type a: SUPRA = log suprailiac/(log triceps + log biceps), BIC = log biceps/(log
subscapular + log suprailiac), SUB = log subscapular/(log triceps + log biceps), TRI = log
triceps/(log subscapular + log suprailiac).
In PCA type b: SUPRA = log suprailiac/(log triceps + log biceps + log calf circumference), BIC =
log biceps/(log subscapular + log suprailiac), SUB = log subscapular/(log triceps + log biceps + log
calf circumference), BIC = log biceps/(log subscapular + log suprailiac), SUB = log
subscapular/(log triceps + log biceps + log calf circumference), TRI = log triceps/(log subscapular
+ log suprailiac), CAL = log calf circumference/(log subscapular + log suprailiac).
(2) The coefficients with the lowest contribution to the biological meaning of the components
(coefficients ~ 0.30) are shown in parentheses.
The other components of the PCA type-a have poor stability across ages with
respect to the pattern of signs and strength of the coefficients. However, the meaning
of the second component is stable with respect to the PCA for the total sample
(Table 3), i.e., anterior-posterior or dorsal-ventral fat pattern. The second and third
components of the PCA type-b are stable across ages (Table 3 and 4) with respect to
the PCA of the total samples when comparing signs of the loading and also relative
loading themselves, except for triceps ratio in the second component from 13 to 16
years than expected in all other PCA's. In summary, only the first component of the
PCA type-b can be viewed as consistent for describing fat patterning during the
growth period due to the stability across age in this Indian male sample.
3.2 Socio-economic effects
Mean component scores obtained in both PCA types in relation to SES levels
(family expenditure) are shown in Table 5. In general SES2 seems to have a fat
distribution less centralised than SES levels in the extremes, SES 1 and SES3, as
shown by the mean values of the first components scores of both PCA's. SESI
seems to have more fat backsided distributed (subscapular and triceps) than the other
SES levels as shown by the mean values of the second component scores of PCA
type-a. SES 1 with respect to the other SES levels showed also lower values of the
lower extremity as shown by the mean score obtained in the second component of
the PCA type-b. However the posterior side of the upper extremity had also a
preferential distribution of fat in SES 1 (mean score of the third component in PCA
type-b). Most of these differences were small in magnitude and statistically not
significant.
Table 5. Descriptive statistics of the component scores by SES
SES n component scores

(1)
2 3
mean SD mean SD mean SD
PCA type a 1 172 0.06 0.93 0.15 1.07 0.13 1.05
2 264 -0.10 1.07 0.01 1.00 -0.02 0.99
3 311 0.05 0.97 -0.09 0.95 -0.06 0.98
PCA type b 1 166 -0.05 0.95 -0.08 1.01 0.08 1.01
2 254 0.13 1.03 0.01 1.05 -0.03 1.04
3 302 -0.08 1.00 0.04 0.95 -0.02 0.96
(1) SESI (less than 1,000 Rupees), SES2 (from 1,000 to 1,500 Rupees) and SES3 (more than 1,500
Rupees)
Table 6. Univariate contribution of the components to the multivariate difference found in

the MANOVA(l) by SES
component F (df" 2,719) P Scheffe test (a = 0.05)

PCA type a 1 1.77 0.171
2 4.00 0.019 SESI =I- SES3
3 2.19 0.112
PCA type b 1 3.26 0.039 SES2 =I- SES3 and SES2 =I- SES 1
2 0.80 0.449
3 0.75 0.474
(I) Multivariate difference: Wilks' lambda - 0.96, F- 2.25 (d.f.: 12,1428; p < 0.01), and Box's M =
=
134.79, F 3.17 (d.f.: 42,984524; p < 0.001).
100 CHAPTER 8
The SES differences were analysed by multivariate design. The vectors of the
components of both PCA types (a and b) were included in a unique MANOVA by
SES (Table 6). The analysis showed significant multivariate differences among the
three subgroups compared (SES1, SES2 and SES3) because the MANOVA yielded a
Wilks' lambda = 0.96 which corresponded to an equivalent F (d.f.: 12, 1428) = 2.25
fairly significant (p < 0.01). However there was no homogeneity of the variances
among the subgroups as shown by the value afforded by Box's M = 134.79 which
corresponded to an equivalent F (d.f.: 42, 984524) = 3.17 (p < 0.001). Only the
second component of the PCA type-a and the first component of the PCA type-b
contributed with significant differences (p < 0.05) to the overall SES differences
found by Wilks' lambda. Moreover an independent multirange comparison test of
Scheffe (= 0.05) was performed (Table 6) with harmonic means due to the size
differences of the subgroups compared, in order to search the subgroups more affected
by SES effects on fat distribution. SES differences in the second component of PCA
type-a were due mainly to the differences between SES 1 and SES3 because the
preferential distribution of fat in the posterior side of the upper body in the lower
SES. SES differences in the first component of PCA type-b were due mainly to
differences between SES2 and the other SES levels, SES 1 and SES3 (Table 6)
because the less central fat distribution of the medium SES level.
Table 7. Results of the two-way MANOVA by age and SES. The component showing
significant value of F contributed to explain the multivariate differences of fat
distribution found by age and SES. However, the significant interaction SES x AGE was
mainly explained by the contribution of the first component of the peA type b, because
of its high F-value, although not significant.
MANOVA Wilks' contribution of the components (1)

lambda
PCA type a PCA type b

1 2 3 2 3
AGE (2) 0.79 149.21 16.30 0.27 61.75 76.67 1.98
F :: 32.45, p < .001 p<.OOl p < .001 p<.OOl p < .001
SES (2) 0.96 1.13 5.30 2.63 2.11 2.05 0.33
F:: 2.31, p < .01
SES x 0.97 0.21 1.09 0.68 2.70 1.59 0.96
AGE (2)
F :: 1.86, p < .05
(1) = degrees of freedom of the F-test, for AGE:
1,716; SES: 2,716; SES x AGE: 2,716
(2) = degrees of freedom for each F associated with the corresponding Wilks' lambda, AGE: 6,711;
SES: 12,1422; SES x AGE: 12,1422.
When taking age into account by a two-way MANOVA by SES and age, only
the differences among SES levels in central fat disappeared while the differences in
the antero-posterior pattern of fat distribution remained (Table 7). Age had a greater
contribution than SES to explain the variability in fat distribution as the respective
values of Wilks' lambda showed. The first two components of both PCA types had
significant contributions to explain age differences in fat distribution (Table 7). In

fact, central fat (Figure la and Ib), dorsal fat (posterior upper body fat) (Figure lc)
and lower versus upper extremity fat (Figure Id) increased with age in all SES
levels. However when Bengali boys are l3-16 years old SES differences in the
secondary fat pattern are more evident than when they are 7-12 years old (Figure 1).
SES differences in fat distribution were due mainly to the contribution of the second
component of PC A type-a (posterior upper body fat) and central fat lost the relevance
in distinguishing SES levels.
0.8 Factor 1 1.6

(a) (b) 1-(Factor 1)
0.6 1.4
0.4
12
0.2
o • SES1 • SES1
0.8
-0.2 o SES2 o SES2
• SES3 0.6 • SES3
-0.4
-0.6 7-12 13 -16 0.4 7-12 13 -16

years years years years
0.8 0.8
(c) Factor 2 (d) Factor 2
0.6 0.6
~
0.4 0.4
0.2 0.2
• SES1
o o o SES2
• SES1 -0.2 • SES3
-0.2
e-- SES2
-0.4 • SES3 -0.4
-0.6 7 -12 13 -16 -0.6 7 -12 13 -16
years years years years
Figure 1. (a) and (c) show the plot of the component scores by age and SES of the first two
fat factors in peA type-a, (b) and (d) show the same plot in peA type-b. The symmetrical
plot of the first factor (I-F) of the peA type-b, was preferred in the graphic because it has
a biological easier interpretation. The other factors do not need transformation.
There were significant interactions age x SES (Wilks' lambda = 0.97) which
corresponded to a significant (p < 0.05) equivalent F (d.f.: 12, 1422) = 1.86. This
two-way interaction was due mainly to the first component of the PCA type-b
because it afforded the highest univariate F-value with respect to the other
components (although not significant F (d.f.: 2, 716) = 2.70, p > 0.05). When
plotting mean component scores by age groups in each SES, the interaction can be
shown also graphically (Figure 1b). Figure 1 shows only the components which
102 CHAPTER 8
contributed significantly to explain the differences found in the two-way MANOVA

(Table 7), the first and second components. These are also the more meaningful
components in the study in both PCA's (type a and b) because contributed
significantly to distinguish SES levels. The complementary plot of the first factor
(I-F) of the PCA type-b, was preferred in the Figure because only with such
transformation its meaning is comparable to the first component of the PCA type-a
and thus, it has an easier biological interpretation visually.
4. DISCUSSION
For the last several decades studies on fat distribution (Gam 1955, Norgan 1991) of
the children and adults have gained a considerable importance because of its
significant association with the elevated risk of many diseases (Bjorntorp 1987,
Knowler et al. 1981). The study of the effect of age, sex and SES on fat distribution
in general populations can help to clarify some aspects of the relationship between
fat and risk for some diseases. Differential health risks associated with different types
of body fat distribution has justified the relevancy of studying the stability of the
adipose tissue distribution in general population (Katzmarzyk et al. 1999, Eveleth
and Tanner 1990) and also in populations with high associated risk of disease
(Bouchard et al. 1990, Brenner et al. 1994).
The superiority of the PCA is well known as it utilises all of the information
available in a covariance matrix (Johnston et al. 1991). However, in order to extract
the pattern of fat distribution, researchers have suggested several approaches for
performing the PCA, namely, (i) directly from the skinfold thickness measurements
(Mueller and Reid 1979, Ramirez and Mueller 1980), (ii) from the ratios of the
skinfolds (Cameron et al. 1992, Rebato et al. 1998), (iii) from the residuals of the
regressions of each skinfold on the mean of the log skinfold thickness for an
individual (Healy and Tanner 1981). Hattori et at. (1987) have however found no
major differences in the results by using these approaches. Such a conclusion
becomes more evident when the components extracted in the present study are
compared with those of the Indian study of Johnston et al. (1991) which used the
Healy and Tanner (1981) method.
Although the results of the PCA depends on the variables entered (Baumgartner et
al. 1990, Kaplowitz et al. 1987), the first component of the fat distribution of the
Bengali Boys can unequivocally be identified as the trunk-extremity component as
revealed from the two types of PCA developed in the present research. Since this
component has been obtained from all PCA made so far, it has therefore been coined
as the universal component (Johnston 1992) specific to Homo sapiens (Cameron et
al. 1992). Interchangeably, the component has been called by various names like,
proximal-distal, centripetal, central-peripheral, bipolar, and android-gynoid
continuum, and usually captured most of the variance in comparison to the other
patterns.
The trunk-extremity component of the PCA type-b appears to be more stable and
the contrasts seem to have become more intense due to the inclusion of calf
circumference which presumably has been able to replace the situation when data on
calf skinfold is not available. Thus calf circumference in this study has played a
significant role describing the index of fat of the lower extremity and the contrasts
with the other sides of fat distribution with acceptable performance (68.8%) of the
variance explained by the PCA type-b because the reduction of total variability
explained by the analysis was only 7.8% with respect to PCA type-a. Mueller et al.
(1989) have earlier documented the importance of the circumferential measurements
in extracting the components of fat patterning during the growing years. Further, the
introduction of the leg site has been found to be indispensable for developing the
trunk-extremity contrast (Mueller and Stallones 1981, Mueller and Wohleb 1981). It
is also interesting to note that the trunk-extremity component correlates highest with
the suprailiac index in all PCA's performed. Hence the index may be a sensitive
indicator of this dimorphic fat pattern and should be included as a standard index in
other studies. Coincidentally, among the Cuban infants (Dfaz et al. 1994), the best
association with the first component has been shown by the suprailiac/medial calf
ratio because these two skinfold sites are good indicators to describe the trunk-
extremity contrast.
4. J Stability and age changes in the first component

The first component of the PCA type-b has shown a remarkable stability across
ages. Recently several authors (Baumgartner et al. 1986, Rebato et al. 1998) have
shown that the trunk extremity component remains unchanged by sex, ethnicity and
educational level during growth. In the present study, the Bengali boys, during the
teenage years, and independently of SES, experienced the increased centralisation of
fat universally described in all populations (Rolland-Cachera et al. 1990, Johnston
1992, Garn and Haskell 1959, Baumgartner et al. 1990, Mueller 1982). Other studies
indicate that the adolescence is a period when redistribution of fat from the leg to the
trunk accelerates and this acceleration is more discernible in males than in females
(Baker et al. 1958, Stolz and Stolz 1951, Fry et aI., 1965). The mean scores of the
first component appears to have enhanced from 13.0 years onwards which is the age
at PHV of this sample (de Onis et al. 2001). Recently, Malina et al. (1999) have
stated that increment of this component is more apparent after the PHV than before.
It has recently been proposed that the hormones which are responsible for
influencing the amount and regional distribution of adipose tissue during puberty are
cortisol, insulin, growth hormone and sex steroids (Roemmich and Rogol 1999).
The relative instability of the trunk-extremity component of the PCA type-a may be
related to the variation in changes in individual skinfold thickness sites during male
adolescence (Malina et al. 1999). After the puberty, the adult pattern of fat
distribution becomes established, subscapular fat increase more rapidly than triceps
and in general trunk skinfold increases more rapidly than the extremity skinfolds.
Further at the end of the growth period, the centralisation has been found to be
greater in boys than in girls (Susanne et al. 2000). Thus two processes are found to
be operating: an increase in trunk fat and a decrease in upper arm fat due to this
redistribution of fat to the trunk after puberty (Rolland-Cachera et al. 1990).
4.2 Reversals in the signs o/the coefficients

With respect to the signs of the coefficients associated with the first component in
particular, popUlations display remarkable variation. For example in two Indian
studies, carried out on the adolescents and on the adults (Johnston et al. 1991, Indech
104 CHAPTER 8
et al. 1991) and one study on the U.S. children aged 12 years (Deutsch et al. 1985),
the authors found negative loading for the extremity sites and positive for the trunk
sites. On the other hand, Norgan (1987), in his study on the Lufa males obtained the
reverse, i.e., positive for the extremity sites and negative for the trunk sites which
resembles the present results for PCA type-b. Further reversals of the signs of the
first component within a sample is observed to be not unusual (Norgan 1987, Dfaz
et at. 1994). Mueller and Reid (1979) reported reversals in the signs of the first
component between the children and the adolescents. We have also found such
reversals in PCA type-a but without changes in the meaning of the components.
4.3 The second component

This component, unlike the first, varies widely among populations. Some
studies have identified it as the upper-lower trunk (Mueller and Reid 1979, Deutsch
et al. 1985, Hattori et al. 1987), some as the dorsal-ventral (Cameron et al. 1992)
while others as the upper-lower extremity (Mueller et at. 1986). It will be interesting
to mention that in this study with two different types of PCA (a and b) we have
found both the dorsal-ventral component (PCA type-a), consistent with the findings
of Cameron et al. (1992) in South African children, as well as the upper-lower
extremity (PCA type-b) as stated by Malina and Bouchard (1988).
Although with respect to the occurrence of the upper-lower extremity contrast
(PCA type-b) the role of biceps skinfold has been stated to be important (Mueller
and Wohleb 1981), but recently, among the Argentinean boys aged 10-15 years,
Demarchi and Marcellino (1999), have reported the contrast for this component
between the triceps and calf sites. The results of the present study for this component
reported the contrast, instead of triceps, between biceps and calf site probably because
the use of calf circumference as a marker of leg site.
The non-occurrence of the upper-lower trunk contrast as the second component
may find its valid clarifications. Generally this component has been reported to be
more prevalent among the adult samples where the measures are only circumferences
and the subjects would preferably be obese (Johnston 1992, Susanne et al. 2000).
The present sample, however, does not fulfil any of these criteria.
Baumgartner et al. (1986, 1990) speculated that the second and the subsequent
components did not have any biological meaning but that may be due to artefacts of
the PCA methodology. Cameron et al. (1992), have argued this view and showed
that the second component has consistently emerged in various populations. Our
results thus supports the view of Cameron et al. (1992).
To conclude therefore, we have obtained on the Bengali boys three patterns of fat
distribution from both types of PCA: the trunk-extremity, the dorsal-ventral and the
upper-lower extremity pattern. While the first component is remarkably stable the
other two are fairly specific with respect to the skinfold sites chosen. While the
universal occurrence of the first one has been already stressed the second one has
drawn the attention from the researchers (Cameron et at. 1992) because it could also
be more universal than thought in the past years. In fact the first two components
are of prime importance as they explained more than 90 % of the total variance.
4.4 Effect of SES on fat patterning

It has recently been shown that during the growing years the effect of SES on
centralisation is more pronounced among the girls than among the boys (Rebato et
al. 1998, Johnston et al. 1991). Thus sex specific effects of social class as stated by
Malina (1996) appears to be applicable in the context of the present study.
A greater centralisation in the two extreme SES groups, as noticed in the present
study, has been reported to exist more markedly in the upper and lower social classes
of a previous Indian study (Johnston et al. 1991). Further among the Mexican-
American and Puerto-Rican men of the both extreme SES groups, the centralisation
has occurred in a greater magnitude than in the middle SES groups (Georges et al.
1991). Moreover, with respect to the second component (PC A type-a), the results
show that the lower the social class is, the greater will be the dorso-ventral fat
deposition. Unfortunately no comparable result is available yet on this aspect. But
the upper-lower extremity component (PCA type-b) has remained unaffected due to
the influence of the SES and has made the trends unclear again. It may be noted that
among the adult males of Chandigarh, India, the effect of social class on the second
component (upper-lower trunk) has been reported to be significant (Indech et at.
1991). While greater centralisation in lower social class as reported from large
number of studies, has been interpreted by the hormonal theory (Bjorntorp, 1996),
yet there does not exist any such hypothesis to explain the mechanism involved with
greater centralisation in both SES extremes. However, at the end of the period
studied the centralisation tends to raise more in the low SES subsample than in the
other socio-economic levels only when calf circumference is considered (Figure Ib).
Thus probably at the end of the adolescence and in adults the SES influence on fat
distribution could be more consistent with an ecological model of greater
centralisation only in the low SES in West Bengal. Repeated occurrence of this trend
from the future growth studies in more contrasting environments and in other age
ranges, perhaps would clarify as well as hypothesise the causation of the sex
specificity and differential behaviour of the social groups on fat distribution during
growth.
Acknowledgements. The authors express their sincere gratefulness to late S.R.

Das, late K.B. Pakrasi and K.C. Malhotra for kindly initiating and supervising the
Project. The University of the Basque Country (UPV/EHU) has also partly supported
this research by the UPV 154.31O-EA049/99 project.
5. REFERENCES
Bailey, S.M., Gershoff, S.N., McGandy, R.B., Nandosita, A., and Tantiwongse, P., 1985, Subcutaneous
fat remodelling in Southeast Asian infants and children. American Journal of Physical
Anthropology,68,123-130.
Baker, P.T., Hunt, E.E., and Sen, T., 1958, The growth and interrelations of skinfo1ds and brachial
tissues in man. American Journal of Physical Anthropology, 16,39-58.
Baumgartner, R.N., Roche, A.F., Guo, S., Cameron Chumlea, W., and Ryan, A.S., 1990, Fat patterning
and centralised obesity in Mexican-American Children in the Hispanic Health and Nutrition
Examination Survey (HHANES 1982-1984). American Journal of Clinical Nutrition, 51, 936S-
943S.
106 CHAPTER 8
Baumgartner, R.N., Roche, A.F., Guo, S., Lohman, T., Boileau, R.A., and Slaughter, M.H., 1986,
Adipose tissue distribution: The stability of Principal Component by sex, ethnicity and maturation
stage. Human Biology, 58, 719-735.
Bjorntorp, P., 1987, Fat patterning and disease: A review. In Human Body Composition and Fat
distribution, edited by N.G. Norgan (Wageningen, Netherlands: Stichting Nederlands Instituut voor
de Voeding). Euro-Nut Report, 8, pp, 201-219.
Bjorntorp, P., 1988, Possible mechanisms of relating fat distribution and metabolism. In Fat distribution
during growth and later health outcomes, edited by C Bouchard and F.E. Johnston (New York:
Alan Riss) pp. 175-19L
Bjorntorp, P., 1996, The regulation of adipose tissue distribution in humans. International Journal of
Obesity, 20, 291-302.
Bogin, B., and Sullivan, T., 1986, Socioeconomic status, sex, age and ethnicity as determinants of body
fat distribution for Guatemalan children. American Journal of Physical Anthropology, 69, 527-535.
Bouchard, C, 1988, Inheritance of human fat distribution. In Fat distribution during growth and later
health outcomes, edited by C Bouchard and F. E. Johnston (New York: Alan R Liss) pp. 103-125.
Bouchard, C, 1992, Heredity and regional fat distribution during growth. In Human Growth: Basic and
clinical aspects, edited by M. Hernandez and 1. Argente. (Amsterdam: Elsevier Science Publishers,
B.V.) pp, 227-243.
Bouchard, C, Bray, G.A., and Hubbard, V.S., 1990, Basic and Clinical aspects of regional fat
distribution. American Journal of Clinical Nutrition, 52, 946-950.
Brenner, J.F., Casey, V.A., Dwyer, J.T., and Bailey, S.M., 1994, The measurement of body fat
distribution using somatotype photographs and computer assisted imaging techniques. Annals of
Brown, D.E., Severance, CJ., Sako, E.K, Chun, D.Y., Young, L.L, and Johnson, J.L., 1992, Growth of
native Hawaiian School children: II. Body Mass Index and skinfold measurements. American
Journal of Human Biology, 4, 433-445.
Cameron, N., Johnston, F.E., Kgamphe, J.S., and Lunz, R, 1992, Body fat patterning in rural South
African Black children. American Journal of Human Biology, 4, 353-364.
Central Statistical Organisation, 1962, Standard Industrial and Occupational Classifications. (New
Delhi: Department of Statistics, Government of India).
Cronk, C.E., Mukherjee, D., and Roche, A.F., 1983a, Changes in triceps and subscapular skinfold
thickness during adolescence. Human Biology, 55, 707-72L
Cronk, CE., Roche, A.F., Kent, R Eichorn, D., and McCammon, R.W., 1983b, Longitudinal trends in
subcutaneous fat thickness during adolescence. American Journal of Physical Anthropology, 61,
197-204.
Das Chaudhuri, A.B., and Bose, K, 2000, A review of studies on central body fat distribution among
Indians (Calcutta: Salmoli Publication).
de Onis, M., Dasgupta, P., Saha, S., Sengupta, D., and Bloessner, M., 2001, NCHS reference and the
growth of the Indian Adolescent boys. American Journal of Clinical Nutrition, (in press).
Demarchi, D.A., and Marcellino, AJ., 1999, Body composition and fat distribution in 10-15 year old
boys from Cordoba, Argentina. Acta Medica Auxologica, 31, 39-44.
Deutsch, M.I, Mueller, W.M., and Malina, RM., 1985, Androgyny in fat patterning associated with
obesity in adolescents and young adults. Annals of Human Biology, 12,275-286.
Diaz, M.E., Montero, M., Wong, I., Moreno, V., Toledo, E.M., and Moreno, R, 1994, Fat Distribution in
Cuban Infants. American Journal of Human Nutrition, 6, 687-692.
Eveleth, P.B., and Tanner, J.M., 1990, Worldwide variation in Human Growth (Cambridge: Cambridge
University Press).
Fiori, G., Facchini, F., Pettener, D., Rimondi, A., Battistini, N., and Bedogni, G., 2000, Relationships
between blood pressure, anthropometric characteristics and blood lipids in high and low altitude
popUlations from central Asia. Annals of Human Biology, 27, 19-28.
Fry, E.I, Chang, KS.F, Lee, M.M.C, and Ng, CK, 1965, The amount and distribution of subcutaneous
tissue in Southern Chinese children from Hong Kong. American Journal of Physical Anthropology,
23,69-80.
Gam, S.M., 1955, Relative fat patterning: an individual characteristic. Human Biology, 27, 75-89.
Gam, S.M., and Clark, D.C, 1976, Trends in fatness and the origins of obesity. Metabolism, 57, 443-
456.
Gam, S.M., and Haskell, J.A., 1959, Fat changes during adolescence. Science, 129, 1615-1616.
Georges, E., Mueller, W.H., and Wear, M.L., 1991, Body fat distribution: Association with
socioeconomic status in the Hispanic Health and Nutrition Examination Survey. American Journal
of Human Biology, 3, 489-50L
Georges, E., Mueller, W.H., and Wear, M.L., 1993, Body fat distribution in men and women of the
Hispanic health and nutrition examination survey of the United States: Association with behavioural
variables. Annals of Human Biology, 20, 275-29L
Gillum, F., 1987, The association of the ratio of waist to hip girth with blood pressure, serum cholesterol
and uric acid in children and youths aged 6-17 years. Journal of Chronic disease, 40, 413-420.
Haffner, S.M., Stern, M.P., Hazuda, H.P., Rosenthal, M., Knapp, J.A., and Malina, RM, 1986, Role of
obesity and fat distribution in non-insulin dependant diabetes mellitus in Mexican-American and
non- Hispanic Whites. Diabetes Care, 9, 153-16l.
Hammond, W.A., 1955, Measurement and interpretation of subcutaneous fat, with norms for children
and young adult males. British Journal of Preventive and Social Medicine, 9, 201-211.
Hattori, K., Becque, M.D., Katch, V.L., Rocchini, A.P., Boileau, RA, Siaughtere, M.H., and Lohman,
T.G., 1987, Fat patterning of adolescents. Annals of Human Biology, 14,23-28.
Healy, M.J.R, and Tanner, J.M., 1981, Size and Shape in relation to growth and form. Symposia of the
Zoological Society of London, 46, 19-32.
Indech, G.D., Sanjeev, I., and Johnston, F.E., 1991, Age, sex and socioeconomic correlates of fat
patterning among adults from Chandigarh zone of Northwest India. American Journal of Human
Biology, 18,463-470.
Johnston, F.E., 1988, Sex differences in fat patterning in children and Youth. In Fat Distribution During
Growth and Later Health Outcomes, edited by C. Bouchard and F.E. Johnston (New York: Alan R
Liss) pp, 85-102.
Johnston, F.E., 1992, Developmental aspects of fat patterning. In Human Growth: Basic and clinical
aspects, edited by M. Hernandez and J. Argente (Amsterdam: Elsevier Science Publishers, B.V.)
pp,217-226.
Johnston, F.E., Heath, B.H., and Shoup, RF., 1995, Patrones de distribucion de la grasa en cuatro
muestras de jovenes de 12-18 anos de edad. Revista Espanola de Antropologfa Biol6gica, 16, 69-
83.
Johnston, F.E., Sanjeev, 1., and Indech, G.D., 1991, Fatness and fat patterning in 12-17 year old youth
from the Chandigarh zone of Northwest India. American Journal of Human Biology, 3, 587-597.
Kaplowitz, H.J, Mueller, WH., Selwyn, B.1., Malina, RM., Bailey, D.A, and Mirwald, R.L., 1987,
Sensitivities, specificities and positive predictive values of simple indices of body fat distribution.
Kaplowitz, H.1., Wild, K.A., Mueller, W.H., Decker, M., and Tanner, J.M., 1988, Serial and parent-
child changes in components of body fat distribution and fatness in children from the London
Longitudinal Growth study, ages, two to eighteen years. Human Biology, 60, 739-758.
Kapoor, S., Patra, P.K. Sandhu, S., and Kapoor, A.K., 1998, Fatness and its distribution pattern among
Jat Sikhs. Journal of the Indian Anthropological Society, 33, 223-228.
Katzmarzyk P., Perusse, L., Malina, R., and Bouchard, e., 1999, Seven year stability of indicators of
obesity and adipose tissue distribution in the Canadian population. American Journal of Clinical
Nutrition, 69, 1123-1129.
Knowler, W.C., Pettitt, D.1., Savage, PJ., and Bennet, P.H., 1981, Diabetes in Pima Indian: contributions
of obesity and parental diabetes. American Journal of Epidemiology, 113, 144-156.
Larsson, B., Seidell, J., Svarsuud, K., Welin, L., Tibblin, G., and Bjorntorp, P., 1989, Obesity, adipose
tissue distribution and health in men. The study of men born in 1913. Appetite, 13,37-44.
Malina, R.M., 1996, Regional Body Composition: age, sex and ethnic variation. In Human Body
Composition, edited by A. Roche, S.B. Heymsfield and T.G. Lohman. (Champaign, IL: Human
Kinetics) pp, 217-255.
Malina, R.M., and Bouchard, e., 1988, Subcutaneous fat distribution during growth. In Fat Distribution
During Growth and Later Health Outcomes, edited bye. Bouchard and F.E. Johnston (New York:
Alan R. Liss) pp, 63-84.
Malina, R.M., Koziel, S., and Bielicki, T., 1999, Variation in Subcutaneous Adipose tissue distribution
associated with age, sex and maturation. American Journal of Human Biology, II, 189-200.
Malina, RM., Mueller, W.H., Bouchard, e., Shoup, R.F., and Lariviere, G., 1982, Fatness and fat
patterning among athletes at the Montreal Olympic Games, 1976. Medicine Science and Sports
Exercise, 14, 445-452.
Mueller, W.H, Marbella, A., Harrist, RB, Kaplowitz, H.A., Grunbaum, J.A., and Lebarthe, D.R., 1989,
Body circumferences as alternatives to skinfold measures of body fat distribution in children.
Annals of Human Biology, 16,49-506.
Mueller, W.H., 1982, The changes with age of the anatomical distribution of fat. Social Science in
Medicine, 16, 191-196.
Mueller, W.H., 1985, The biology of human fat patterning. In Human Body Composition and Fat
Distribution, edited by N. Norgan (London: Euro-Nut) pp, 159-174.
Mueller, W.H., 1986, Environmental sensitivity of different skinfold sites. Human Biology, 58, 499-506.
Mueller, W.H., 1988, Ethnic differences in fat distribution during growth. In Fat Distribution During
Growth and Later Health Outcomes, edited by e. Bouchard and F.E. Johnston (New York: Alan R
Liss) pp, 127-145.
Mueller, W.H., and Reid, R.M., 1979, A multivariate analysis of fatness and relative fat patterning.
American Journal of Physical Anthropology, 50, 199-208.
Mueller, W.H., and Wohleb, J.e., 1981, Anatomical distribution of subcutaneous fat and its description
by multivariate methods: How valid are principal components? American journal of Physical
anthropology, 54, 25 -35.
108 CHAPTER 8
Mueller, W.H., Deutsch. M.l., Malina, RM., Bailey, D.A., and Mirwald, R.L., 1986, Subcutaneous fat
topography: Age changes and Relationship to Cardiovascular fitness in Canadians. Human Biology,
58, 955-973.
Mueller, W.M., and Stallones, L., 1981, Anatomical distribution of subcutaneous fat: skinfold sites and
construction of indices. Human Biology, 53, 321-335.
Nirmala Reddy, B., 1998, Blood Pressure and adiposity: a comparative study of socioeconomically
diverse group of Andhra Pradesh, India. American Journal of Human Biology, 10, 5-21.
Norgan, N., 1987, Fat patterning in Papua New Guineans. Effects of age, sex and acculturation.
American Journal of Physical Anthropology, 74,385-392.
Norgan, N., 1991, Anthropometric assessment of body fat and fatness. In Anthropometric Assessment
of Nutritional Status, edited by Himes, W.H. (New York: WiJey-Liss Inc.) pp, 197-212.
Ramirez, M.E., and Mueller, W.H., 1980, The development of obesity fat patterning in Tokelau
children. Human Biology, 52, 675-687.
Rebato, E., Salces, 1., San Martin, L., and Rosique, J., 1998, Fat distribution in relation to sex and
socioeconomic status in children 4-19 years. American Journal of Human Biology, 19, 799-806.
Risley, H.H., 1891, Tribes and Castes of Bengal. Ethnographic Glossary I and 2. (Calcutta: Bengal
Secretariat Press).
Risley, H.H., 1908, The people of India. (London: Thacker, Sprinker and Co.).
Robson, lRK., Bazin, M., and Soderstrom, R, 1971, Ethnic differences in skinfold thickness. American
Journal of Clinical nutrition, 24, 864-868.
Roemmich, IN., and Rogol, A.D., 1999, Hormonal Changes during Puberty and their relationship to fat
distribution. American Journal of Human Biology, 11,209-224.
Rolland-Cachera, M.F., Bellisle, F., Deheeger, M., Pergquignot, F., and Sempe, M., 1990, Influence of
body fat distribution during childhood: and body fat distribution in adulthood: A two decade follow-
up study. International Journal of. Obesity, 14,473-481.
Rosique, l, Rebato, E., Gonzalez, A., and Pacheco, J.L., 1994, Somatotype related to centripetal fat
patterning of 8-to 19 -year-old Basque boys and girls. American Journal of Human Biology, 6, 171-
181.
Sarria, A., 1992, Methods for assessing fat patterning in children In Human Growth: Basic and Clinical
Aspects, edited by M. Hernandez and J Argente (Amsterdam: Elsevier Science Publishers, B.V.)
pp, 233-244.
Selby, J.V., Newman, B., Quesenberry, Ch.P., Fabsitz, RR, Carmelli, D., Meaney, F.J., and Slemenda,
Ch., 1990, Genetic and Behavioural influences on body fat distribution. International Journal of
Obesity, 14,593-602.
Solomons, N.W., and Kumanyika, S., 2000, Implications of racial distinctions for body composition and
its diagnostic assessment. American Journal of Clinical Nutrition, 71,1387-1389.
Stolz, H.R, and Stolz, L.M., 1951, Somatic development of Adolescent Boys. A study of the growth of
boys during the second decade of life. New York, Macmillan.
Susanne, c., Rebato, E., Hauspie, R.C., Vercauteren, M., Sa ices, I., San Martin, L., and Rosique, I.,
2000, A review of the relationship between nutrition and some growth and Development data. In
Puberty: Variability of changes and Complexity of factors, edited by Bodzsar, E.B., Susanne, C. and
Prokopec, M. (Budapest: Eotovos University Press).
Szathmary, E.J.E., and Holt, N., 1983, Hyperglycemia in Dogrib Indians of the Northwest Territories,
Canada: Association with age and a centripetal distribution of body fat. Human Biology, 55, 493-
515.
Ulijaszek, S., and Kerr, D.A., 1999, Anthroppometric Measurement error and the assessment of
nutritional status. British Journal of Nutrition, 82,165-177.
Wagner, D.R, and Heyward, V.H., 2000, Measures of body composition in blacks and whites: A
comparative review. American Journal of Clinical Nutrition, 71, 1392-1402.
Weiner, J.S., and Lourie, I.A., 1969, Human Biology: A guide to field methods. International Biological
Programme. Handbook, 9 (Oxford: Blackwell Scientific Publications).
CHAPTER 9
DENTOFACIAL MORPHOLOGY, GROWTH AND

GENETICS: A STUDY OF AUSTRALIAN ABORIGINES
T. BROWN, G. TOWNSEND
Dental School, The University of Adelaide, Adelaide, South Australia
1. THE YUENDUMU STUDY
In August 1951 an interdisciplinary team of scientists from The University of

Adelaide, South Australia visited Yuendumu for the first time. Yuendumu, located
about 290 km north-west of Alice Springs in the Northern Territory of Australia, as
shown in Figure 1, was a settlement established in 1946 by the Australian
Government to provide a less harsh environment with a plentiful supply of food and
water for the previously nomadic Aboriginal people of the region. The scientists
were led by the late Thomas Draper Campbell (1893-1967), a physical
anthropologist and dental academic who had worked with tribal Aborigines for many
years, describing their physical characteristics and dental conditions (Campbell,
Gray and Hackett 1936, Campbell 1939).
One of the team members was the late Murray James Barrett (1916-1975) who
was particularly interested in masticatory function and dental occlusion. He
recognised an unique opportunity for a long-term study of the Yuendumu
community stating "It was realised then that the community and conditions at
Yuendumu would offer excellent opportunities for continuing the dental studies. The
settlement was geographically isolated and there was every likelihood that the group
of Aborigines would stay there because of the plentiful supply of food and other
amenities available to them. The people had almost completely maintained their
pure aboriginal ancestry, they were tribally oriented in that they still held many of
the beliefs and practised many of the customs of former days, their near-primitive
way of life had not been influenced to any great extent by contact with Europeans at
the settlement, and they were friendly and co-operative. It was considered that
continued investigation of the group would provide new information about dental
features of anthropological interest and about the patterns of oral and dental diseases
under settlement conditions. Furthermore, repeated observation of the same subjects
at different ages would enable accumulation of information on facial growth and
dental development. Having these aims the field work was planned to continue for a
number of years with brief annual return visits to the settlement" (Barrett 1976).
Yuendumu was visited annually by dental teams resulting in the accumulation of
serial dental casts and observations on the oral health of Aboriginal adults and
109
© 200! Kluwer Academic Publishers.
110 CHAPTER 9
children enrolled in the study. Special attention was given to the changing
environment as the Aboriginal community gradually adopted a lifestyle markedly
different from their earlier hunter-gatherer existence (Campbell and Barrett 1953,
Barrett 1958).
-
NORTHERN
-<
.....J TERRITORY
-<~
E- O
rJ)
Z
;:J
-< -<
.....J
~1
rJ)
Yuendumu Reserve Z
tJ.J
tJ.J
tJ.J
r . Al ice Spdng'
~
0
~ ~
SOUTH AUSTRALIA
Figure 1. Map showing location of Yuendumu.
In 1961 the scope of the study was extended to include a wider range of
observations on the Yuendumu subjects including measurements of the face and
skull and selected general body features. Computer-based methods were developed to
DENTOFACIAL MORPHOLOGY, GROWTH AND GENETICS III
deal with the accumulation of study material which included serial somatometric
data, observations from cephalometric and hand-wrist radiographs, measurements
from dental casts and, of particular importance, genealogical records of Yuendumu
families (Barrett, Brown and Fanning 1965, Barrett, Brown and MacDonald 1965,
Barrett 1969). The overall objective of the researchers was to use the Yuendumu data
in correlated studies of morphological features, functional relations and patterns of
growth and development. Phase one was concerned primarily with descriptive and
comparative aspects of the physical and dental characteristics in order to provide
base-line references for later use. Phase two consisted of correlative analyses where
the dentition and craniofacial structures were considered in conjunction with general
body growth and skeletal maturation. Phase three, the most recent, was initiated by
Townsend (1976) to utilise the genealogical records. The relative geographic
isolation of the Yuendumu community and the practice of polygyny provided the
unique opportunity for genetic studies of variation in the metric and non-metric
dental characters (Townsend and Brown 1978, Townsend 1980, Townsend and
Brown 1981).
Annual trips to Yuendumu continued until 1971 resulting in the extensive
collection of records and observations, housed in the Dental School at The
University of Adelaide, that still provide material for study. To date over 300
publications have arisen from the project, including many by visiting researchers
from Sweden, Denmark, Finland, England, United States of America, Israel, Japan
and India. As the investigation continues, its uniqueness and many advantages are
being reinforced, particularly for the study of the nature and causes of variations in
dental maturation and general physical growth and development.
2. MORPHOLOGICAL VARIA nON

The most striking dentofacial features of Australian Aborigines include large primary
and permanent teeth, well-developed dental arches and a facial profile characterised
by alveolar prognathism, as shown in figures 2 and 3. In the Yuendumu group,
mesiodistal and buccolingual tooth diameters were, with few exceptions, greater than
those reported for other contemporary populations both for primary and permanent
dentitions (Barrett, Brown and MacDonald 1963a, Barrett, Brown and Luke 1963,
Barrett, Brown, Arato et al. 1964). However, there are wide variations in tooth size
among Aboriginal populations, the Yuendumu subjects displaying relatively small
tooth size when compared with earlier groups represented by skeletal material from
diverse regions of Australia (Brace 1980, Brown 1992). Sexual dimorphism in tooth
size within the Yuendumu Aborigines, which averaged about 3% and 4%
respectively for mesiodistal and buccolingual diameters, was particularly marked in
the canine teeth (Townsend and Brown 1979).
Emergence of the permanent teeth into good alignment for optimal occlusal
relationships and function depends primarily on the space made available by
exfoliation of the primary precursors. One measure of space availability, the leeway
space, is calculated as the difference between combined mesiodistal diameters of the
primary canines and molars and their permanent successors on one side of the dental
arch. We noted that the leeway space in Aboriginal children was amongst the
highest of several populations compared, 1.42 mm and 1.28 mm in the maxilla for
males and females and 2.82 mm and 3.25 mm in the mandible (Brown, Margetts
112 CHAPTER 9
and Townsend 1980a). This surplus space assists to minimise crowded dental
arches as the permanent canine and premolars emerge into occlusal alignment.
Figure 2. Models of the maxillary dental arches of an Australian Aboriginal (left) and
a white Australian male (right), highlighting marked differences in both size and arch
shape.
It is interesting that the correlations between corresponding dimensions of the

primary and permanent teeth were greater for combined mesiodistal diameters of
various tooth groups than for individual teeth. It is probable that this co-ordination
in tooth size, which tended to be greater in the Yuendumu subjects than in other
groups compared, is another important factor in the development of optimal occlusal
relationships (Brown, Margetts and Townsend 1980b).
As a consequence of their large teeth, the dental arches and palate in the
Yuendumu Aborigines are correspondingly large although there is considerable
variation between individuals at all ages. An interesting occlusal feature is alternate
intercuspation which is more marked in males. Subjects showing this character have
a disparity in breadths of the upper and lower arches which allows maximum
interdigitation of opposing teeth on either left or right side but not on both sides
together as is usually regarded to be the normal form of interarch tooth contact
(Barrett 1969, Brown, Abbott and Burgess 1987). A further consequence of the large
tooth diameters is the marked alveolar prognathism which increases as the
permanent teeth emerge and tends to be more pronounced in females of the group.
The midfacial prognathism and the apparent receding chin are noticeable from an
early age (Barrett, Brown and MacDonald 1963b).
DENTOFACIAL MORPHOLOGY, GROWTH AND GENETICS 113
Figure 3 A young Aboriginal male showing characteristic alveolar prognathism
3. TOOTH EMERGENCE AND DENTO-ALVEOLAR DEVELOPMENT

Attention was directed to the sequence and timing of tooth emergence, dento-
alveolar development and functional dental occlusion in the Yuendumu subjects.
The sequence of primary tooth emergence was reasonably consistent, the first
emerging teeth being the mandibular and maxillary central incisors followed by the
maxillary lateral incisors. The last teeth to emerge were the canines followed by the
second molars. However, the mandibular lateral incisor could either follow or
precede the first molar in emergence (Barrett and Brown 1966). With respect to the
timing of primary tooth emergence, there was a tendency for the Yuendumu children,
up to the age of 12 to 15 months, to be delayed compared with other groups.
However by age 18 months and later the Aboriginal children did not differ from
114 CHAPTER 9
other groups in the average number of teeth emerged. The relative delay in primary
tooth emergence may be associated with a prolonged period of breast feeding or it
may be a reflection of the severe illnesses suffered by many of the infants at
Yuendumu.
Emergence of the permanent teeth is conveniently considered in two phases:
phase one includes emergence of the first 12 teeth, that is all incisors and first
molars, while phase two includes emergence of the canines, premolars, second and
third molars. The average sequence of emergence in the Aboriginal children did not
differ greatly from that recorded for other groups but the sequence of the mandibular
second premolar and second molar was variable, either tooth taking priority,
suggesting that the emergence sequence of this tooth pair may be polymorphic
(Barrett, Brown and Cellier 1964).
Phase one was complete when 12 permanent teeth were present, a stage that
extended on. average from 8.5 to 9.9 years in boys and from 8.1 to 9.1 years in girls.
After this plateau stage, phase two commenced with emergence of the first premolar
or canine, whichever was the earlier. All teeth except third molars were present at
about 11.5 and 11.1 years in Aboriginal boys and girls respectively. The quiescent
plateau period, with no teeth emerging, tended to be of shorter duration in the
Aboriginal children compared with several other groups, l.47 years in boys and
0.99 years in girls (Brown 1978). Compared with Australian children of European
descent, the Aboriginal children tended to be advanced during phase two. Thus it
appears that tooth emergence in the Yuendumu children is a process that commences
relatively late but finishes earlier than in some European populations. The entire
process extended from about 1 year when the primary incisors emerge until about 16
years with emergence of the third molars. The low prevalence of minor tooth
crowding and other more severe malocclusions in this group is evidence that the
emergence process is relatively unimpeded.
The association between the paths of tooth emergence and the pattern of facial
growth has been documented by Bjork and Skieller (1972). In this respect
compensatory dento-alveolar growth is an important biological mechanism acting
during development of the dental arches and establishment of dental occlusion. The
pattern of craniofacial growth was studied in an Aboriginal boy between 7 and 18
years of age using cephalometric radiographs (Bjork, Brown and Skieller 1984). The
structural method for superimposing the serial radiographs as outlined by Bjork and
Skieller (1972) was used for the analysis. In this boy the entire dental arch migrated
anteriorly and occlusally during the growth period. Anterior migration was 8 mm in
the region of the first permanent molar in both mandibular and maxillary arches. The
anterior teeth migrated less, 6 mm in the mandible and 5 mm in the maxilla. This
anterior migration of the dental arch, which was considerably more than that
recorded for Danish subjects, was accompanied by a shortening of the dental arches
and increasing alveolar prognathism. As a consequence there was ample space for
emergence of the third molars, a condition not always present in the Caucasian
dentition.
The natural function of the dentition and dental occlusion of Australian
Aborigines was described by Barrett (1958) who emphasised the dynamic and
continuously changing nature of occlusal relationships in this group. According to
Barrett's concepts, the dentition underwent a wear-in and a wear-out phase. During
the wear-in phase, continuing wear on the occlusal tooth surfaces, conditioned by
vigorous mastication and abrasives incorporated in foodstuffs, brought about changes

in the shape and areas of contacting surfaces on opposing teeth. This process allowed
occlusal contacts that would not be possible in the absence of tooth wear. As the
wear continued, the occlusal plane developed a distinctive helicoidal appearance
(Richards and Brown 1986).
Occlusal wear progressed even further with increasing age until the second wear-
out phase became apparent. At this time there was excessive loss of tooth substance
and loss of vertical face height often accompanied by pulp exposure, pulp death and
periapical and alveolar pathology. In the fully functioning Aboriginal dentition,
tooth wear also occurred on the interproximal tooth surfaces leading to a gradual
reduction in mesiodistal crown diameters and a corresponding anterior migration of
the dentition.
4. GENERAL GROWTH AND SKELETAL MATURA nON

The collection of serial observations allowed us to detail the patterns of general body
growth and relate these to craniofacial growth and skeletal maturation. Initially, age
changes in stature and weight in the Aboriginal children were studied by a method
of multiple linear interpolations developed by Solow (1969). This form of analysis
provided estimates of the variables being studied at monthly intervals between the
earliest and the latest observations for each subject and produced centile curves for
stature and weight (Brown and Barrett 1971, 1972).
Our records did not include observations before five years of age but, with respect
to stature, after this age the Yuendumu children differed little from British children of
similar age with whom they were compared. Adult height was 173.0 cm on average
in males and 162.7 cm in females. The analysis of weight, however, confirmed
previous studies of Australian Aborigines and showed that the Yuendumu children
were lighter on average than British children of the same age. In boys the difference
in average weight between groups was about 2.6 kg at age 5 years, 3.7 kg at 10
years, 6.6 kg at 15 years and 3.0 kg at maturity. In the girls the differences in
average weight were 2.9 kg at age 5, 3.8 kg at age 10,8.6 kg at age 15 and 10.1 kg
at maturity. The similarity in stature but difference in weight between Aboriginal
and British children emphasised the linearity of body build in the former.
Growth in height was also studied in individual children by applying a non-
linear curve-fitting procedure with five constants (Preece and Baines 1978). This
analysis of longitudinal observations provided a more detailed description of
individual variations in the patterns of general growth and, by fitting both distance
and velocity curves, allowed the calculation of the magnitude and timing of peak
growth velocities at adolescence as well as several other useful indices of growth
(Brown and Townsend 1982). By computing mean constant curves, with each of the
five constants averaged over all children, the growth characteristics of the group can
be summarised. In the Yuendumu boys adolescent growth in stature peaked at 14.0
years when it reached 10.3 cm/year whereas in girls the peak of 8.4 cm/year was
reached at 11.9 years.
Although we have no records of Yuendumu infants, there is ample evidence that
Aboriginal children living under settlement conditions were retarded in their early
growth when assessed against North American or British standards for white
children (Moodie 1973). The growth retardation was invariably associated with high
116 CHAPTER 9
infant mortality, morbidity and sometimes protein-calorie malnutrition. The older

Yuendumu children, however, did not display any marked growth retardation which
we believe may be explained by "catch-up" growth in late childhood or adolescence.
A distressing record of infant mortality and morbidity at Yuendumu in the years
preceding and during our longitudinal study supports this view (Middleton and
Francis 1976).
Anthropometric surveys of people living at Yuendumu or in the same geographic
region were undertaken in the 1930s (Campbell, Gray and Hackett 1936) and in the
1950s by Abbie (1957). Data derived during these earlier expeditions to Central
Australia were used to disclose any secular changes in average stature and selected
dimensions of the head that occurred in successive generations (Brown 1976). There
were few differences between the mean values reported in the 1930s and 1950s, only
one difference exceeding 2%. In contrast, the average values derived from our
measurements in the 1960s were significantly greater than those of the 1930s. For
stature, the secular change amounted to 5.9 cm and 6.0 cm in males and females
respectively. For radius length the increase was 1.3 cm and 0.7 cm respectively.
Craniofacial dimensions also showed secular increases ranging from 0.2 cm for
bizygomatic diameter to 1.0 cm for head length. Bigonial diameter increased more
than bizygomatic diameter resulting in a change in head shape as well as size
between generations. As the secular increase in head length exceeded that in breadth
there was a reduction in the cephalic index between the 1930s and 1960s of 2.2 in
males and 1.5 in females. It is interesting that similar changes in head size and
shape were recorded in two other populations that also experienced a transition from
their earlier life styles with exposure to Westernised diet and food habits, Western
Apaches (Miller 1970) and Skolt Lapps (Lewin and Hedegard 1971), the latter
authors noticing that "the facial index is increasing, facial height having increased
more than bizygomatic breadth, and in the neurocranium there is a tendency towards
dol ichocephalisation."
Skeletal maturation in Yuendumu children, aged from 5 years to maturity, was
investigated by applying the atlas method (Greulich and Pyle 1959) to radiographs
of the right hand and wrist which included the distal extremities of the radius and
ulna (Brown and Grave 1976). In the Aboriginal boys, the skeletal ages lagged
behind the Greulich-Pyle standards for North American white children for each age
group. The retardation tended to decrease with age, ranging from 6 to 10 months
between ages 7 and 14 years but only from 1 to 3 months between ages 16 and 18
years. On average the skeletal retardation in the girls compared with the North
American standards was less than in boys and rarely exceeded 6 months. The
analysis of skeletal maturation indicated that use of the Greulich-Pyle atlas for
ratings of skeletal age in the Aboriginal children would require an adjustment to the
age scale according to the developmental delay. More detailed analysis of the timing
of specified ossification events was undertaken to locate clinical indicators which
would assist the prediction of the timing of maximum growth velocity at
adolescence (Brown, Barrett and Grave 1971, Grave and Brown 1976). Information
on these events and their chronological relationship to peak facial growth velocities
is clinically important when planning orthodontic treatment (Grave and Brown
1979).
Peak velocities in growth of selected measurements of the face and stature were
shown to occur within a few months of each other (Brown, Barrett and Grave 1971).
For example, in the Aboriginal boys, peak growth velocity in maxillary length,
mandibular length, upper face height and total face height occurred at 13.5, l3.8,
l3.0 and l3.8 years respectively. Relationships in the girls followed a similar
pattern. Correlations between the timing of peak adolescent growth in stature,
mandibular length and facial height were high, ranging from 0.78 to 0.84.
Of the skeletal events studied, it was found that initial ossification of the
pisiform, hook of the hamate and the ulna metacarpophalangeal sesamoid of the first
finger were clustered around the time of peak growth velocity in stature and facial
dimensions as shown in figure 4. Ossification of the pisiform and initial ossification
of the hamate occurred, on average, between 1.2 and 1.7 years before peak stature
velocity whereas late ossification of the hamate hook and the sesamoid occurred at
about the same time as peak growth velocity. For diagnosis and treatment planning
in orthodontics, the timing of these ossification events form a useful guide to a
child's general and craniofacial growth status, particularly if there is reason to
suspect that chronological and skeletal ages differ significantly (Grave and Brown
1979).
10
..
to
~8
e
(.;
...
1
'uQ
- 6
-Z,.
~4
'OJ 1. Pisiform
::t
2. Hamate-I
3. Hamate- 2
4. Sesamoid
"
9 10 12 13 14 15 Ifl
Age - years
Figure 4. Ossification events in relation to growth velocity at adolescence
5. GENETIC INFLUENCES ON THE DENTITION

The custom of polygyny amongst the Aborigines at Yuendumu, with a male often
having more than one wife, has provided an opportunity to examine associations for
118 CHAPTER 9
various dento-facial features between half-siblings, who have the same father but
different mothers, as well as full-siblings. This unusual population structure enables
insights to be gained into the influence of genetic factors on developing structures,
but also the role of maternal factors that may be operating pre- and post-natally. We
have concentrated mainly on analysing genetic influences on tooth-size variability in
the Yuendumu Aborigines, but we have also attempted to clarify the importance of
genetic effects on morphological crown features, such as Carabelli trait.
Heritability estimates indicate the proportion of observed variation that can be
attributed to genetic influences. Caution is needed in interpreting these values
because they only relate to the population under examination at a given time, and
they may vary for the same population over time, or between different populations
for the same feature. Our approach to estimating heritabilities for tooth size in the
Yuendumu population was based on analysis of variance, with the assumption that
tooth size is controlled by a multifactorial form of inheritance, with both genetic and
environmental components (Townsend and Brown 1978, Townsend, 1980).
Table I shows the results of partitioning tooth-size variation into additive
genetic, common environmental and unique environmental components. Narrow-
sense heritability estimates (V A) of around 60% were found for both deciduous and
permanent dentitions, but common environmental influences (VEe) or maternal
effects appeared to be more important in determining variation of deciduous teeth
(around 15%) than permanent (around 6%). This finding is consistent with the
known timing of development of the two dentitions: deciduous tooth crowns form
mainly pre-natally, whereas permanent teeth only commence to calcify at birth.
Table I. Genetic and environmental contributions to tooth size variability in

Australian Aborigines
Variance component (%)
Dentition Additive Common Unique

genetic VA environment VEe environment V EW
Deciduous
mesiodistal 50 19 32
buccolingual 66 12 23
both 58 15 27
Permanent
mesiodistal 63 4 33
buccolingual 66 8 26
both 64 6 30
It is interesting that our more recent studies of tooth size variability in white
Australian twins have yielded higher heritability estimates, most being over 80%
(Dempsey, Townsend and Martin 1999).
As mentioned earlier, care is needed in comparing heritability estimates derived
from different populations, but it is possible that the relatively low estimates for
Aborigines reflect a period of time when the environment was exerting considerable
influence on the dentition. As Corruccini (1999) has pointed out, heritability

estimates reflect interplay between both genetic and environmental variances.
Canalising environmental factors may tend to suppress genetic variance in pre-
industrial populations, whereas in western urbanised populations in which
environmental forces are relaxed, genetic variance may increase. It is interesting to
note that there is some evidence that the amount of occlusal variability in the
children at Yuendumu, a reflection of the frequency of so-called malocclusion,
increased within a single generation after settlement (Corrucini, Townsend and
Brown 1990). Such a rapid change is difficult to explain in genetic terms and may
relate to a reduction in maxillary and mandibular structures as diet and food habits
began to change.
Figure 5. Expression of the Carabelli trait in permanent and deciduous dentitions
The availability of serial dental models for many of the Aboriginal children has
enabled us to examine the expression of various morphological crown features, such
as Carabelli trait, in both the deciduous and permanent dentitions of the same
individuals, as shown in figure 5. Assuming that the genetic control of Carabelli
trait is similar in both dentitions, any differences in expression presumably result
from the influence of environmental factors during the period of crown formation.
Similarly, any differences in expression of dental traits between right and left sides of
individuals must reflect developmental disturbances, if it is assumed that genetic
effects on both sides are the same. We have noted that Carabelli trait tends to occur
120 CHAPTER 9
bilaterally with symmetrical expression in both dentitions. Although the trait was
found relatively often in the deciduous dentition but not the permanent, the reverse
situation was very rare (Townsend and Brown 1981). This suggests that the
genotype may be reflected more faithfully in the deciduous dentition. Reduced
penetrance in the permanent dentition could result from environmental influences
operating during its longer period of development. The deciduous dentition is often
described as being more conservative in evolutionary terms than the permanent, and
it would seem that more genetic studies of the former are warranted, particularly as
there is a move in clinical dentistry to deal with developmental disturbances in the
dentitions of growing children at younger ages.
6. IMPLICA nONS OF THE YUENDUMU STUDY

The year 2000 marked the 50th anniversary of the commencement of the Yuendumu
growth study. The records that were collected over a 20-year period are still
providing valuable information about the physical growth and development of a
group of Australian Aborigines who had moved from a traditional nomadic existence
to a settlement where their lifestyle, including their diet, was influenced by European
customs. The careful planning that took place at the commencement of the study,
and the meticulous care taken in collecting records and measurements, has proven to
be well worthwhile, as human biologists and dental researchers continue to learn
more about aspects of physical growth and development, particularly of the dento-
facial structures.
A central theme of the research to date has been emphasis on the importance of
understanding that the various components of the human masticatory system are
inter-related morphologically and functionally. An appreciation of this
interdependence is essential for practising dentists faced with treating patients in
whom one or more of the components may be compromised. The study has also
highlighted links between general body growth and dento-facial development. These
associations are of direct interest to orthodontists who often aim to time their
management of occlusal variations to coincide with the adolescent growth spurt.
Another important outcome of the research has been to provide researchers and
clinicians with a perspective of what might be considered "normal" and "abnormal"
in human dento-facial growth and development. The Yuendumu study has
demonstrated that certain features that many dentists would consider to be
"abnormal" and therefore in need of treatment, e.g., alternate intercuspation or tooth
wear, occur commonly in Aborigines and are not associated with diminished
function. Indeed, often these variations appear to enhance masticatory efficiency.
These findings serve to strengthen the view that a thorough understanding of the
extent and nature of variation in dental and facial structures, both within and between
different human populations, is essential to ensure the provision of appropriate
clinical management.
There are many difficulties associated with longitudinal growth studies including
the expense, retention of subjects, and ensuring that the records collected will remain
useful despite improvements in technology over time. Fortunately, these problems
and many others were minimised during the Yuendumu growth study, providing an
unique resource for continuing research.
Acknowledgements. The growth study was supported by grants from the National
Institute of Health, Bethesda, Maryland, the National Health and Medical Research
Council, Canberra, and the University of Adelaide. We are indebted to the
Aboriginal children and adults who participated in the study and to the late Rev.
and Mrs Tom Fleming, Baptist missionaries of Yuendumu, for their continuing
assistance during 20 years of field visits.
7. REFERENCES
Abbie, A.A., 1957, Metrical characters of a Central Australian tribe. Oceania, 27, 220-243.
Barrett, M.1., 1958, Dental observations on Australian Aborigines: continuously changing functional
occlusion. Australian Dental Journal, 3, 39-52.
Barrett, M.1., 1969, Functioning occlusion. Annals of the Australian College of Dental Surgeons, 2, 68-
80.
Barrett, M.1., 1976, Dental Observations on Australian Aborigines: Collected Papers and Reports 1953-
1973. (Faculty of Dentistry: The University of Adelaide, Adelaide).
Barrett, M.1., and Brown, T., 1966, Eruption of deciduous teeth in Australian Aborigines. Australian
Dental Journal, 11,43-50.
Barrett, M.1., Brown, T., and Cellier, K.M., 1964, Tooth eruption sequence in a tribe of Central
Australian Aborigines. American Journal of Physical Anthropology, 22, 79-89.
Barrett, M.J., Brown, T., and Fanning, E.A., 1965, A long-term study of the dental and craniofacial
characteristics of a tribe of Central Australian Aborigines. Australian Dental Journal, 10, 63-68.
Barrett, M.1., Brown, T., and Luke, J.I., 1963, Dental observations on Australian Aborigines -
mesiodistal crown diameters of deciduous teeth. Australian Dental Journal, 8, 299-302.
Barrett, M.J., Brown, T., and MacDonald, M.R., 1963a, Dental observations on Australian Aborigines:
mesiodistal crown diameters of permanent teeth. Australian Dental Journal, 8, 150-155.
Barrett, M.1., Brown, T., and MacDonald, M.R., 1963b, Dental observations on Australian Aborigines:
roentgenographic study of prognathism. Australian Dental Journal, 8, 418-427.
Barrett, M.1., Brown, T., and MacDonald, M.R., 1965, Size of dental arches in a tribe of Central
Australian Aborigines. Journal of Dental Research, 44, 912-920.
Barrett, M.1., Brown, T., Arato, G., and Ozols, LV., 1964, Dental observations on Australian
Aborigines: buccolingual crown diameters of deciduous and permanent teeth. Australian Dental
Journal, 9, 280-285.
Bj6rk, A., and Skieller, V., 1972, Facial development and tooth eruption. An implant study at the age of
puberty. American Journal of Orthodontics, 62, 339-383.
Bj6rk, A., Brown, T., and Skieller, V., 1984, Comparison of craniofacial growth in an Australian
Aboriginal and Danes, illustrated by longitudinal cephalometric analysis. European Orthodontic
Journal, 6,1-14.
Brace, C.L., 1980, Australian tooth-size clines and the death of a stereotype. Current Anthropology, 21,
141-164.
Brown, P., 1992, Post-Pleistocene change in Australian Aboriginal tooth size: Dental reduction or
relative expansion? In T Brown and S Molnar (eds.) Craniofacial Variation in Pacific Populations
(Adelaide: Anthropology and Genetics Laboratory, The University of Adelaide), p.33-51.
Brown, T., 1976, Head size increases in Australian Aborigines. An example of skeletal plasticity. In R
L Kirk and A G Thome (eds.) The Origin of the Australians (Canberra: Australian Institute of
Aboriginal Studies), pp. 195-209.
Brown, T., 1978, Tooth emergence in Australian Aboriginals. Annals of Human Biology, 5, 41-54.
Brown, T., and Barrett, M.1., 1971, Growth in Central Australian Aborigines: stature. Medical Journal
of Australia, 2, 29-33.
Brown, T., and Barrett, M.1., 1972, Growth in Central Australian Aborigines: weight. Medical Journal
of Australia, 2, 999-1002.
Brown, T., and Grave, K.C., 1976, Skeletal maturation in Australian Aborigines. Australian Paediatric
Journal, 12,24-30.
Brown, T., and Townsend, G.c., 1982, Adolescent growth in height of Australian Aboriginals analysed
by the Preece-Baines function: a longitudinal study. Annals of Human Biology, 9, 495-505
122 CHAPTER 9
Brown, T., Abbott, A., and Burgess, V.B., 1987, Longitudinal study of dental arch relationships in
Australian Aboriginals with reference to alternate intercuspation. American Journal of Physical
Anthropology, 72, 49-57.
Brown, T., Barrett, MJ., and Grave, K.C., 1971, Facial growth and skeletal maturation at adolescence.
Tandlregebladet, 75, 1221-1222.
Brown, T., Margetts, B., and Townsend, G.C., 1980a, Comparison of mesiodistal crown diameters of the
deciduous and permanent teeth in Australian Aboriginals. Australian Dental Journal, 25, 28-33
Brown, T., Margetts, B., and Townsend, G.c., 1980b, Correlations between crown diameters of the
deciduous and permanent teeth of Australian Aboriginals. Australian Dental Journal, 25, 219-223.
Campbell, T.D., 1939, Dental conditions of Aborigines compared with civilized groups. Mankind, 2,
228-229.
Campbell, T.D., and Barrett, MJ., 1953, Dental observations on Australian Aborigines: a changing
environment and food pattern. Australian Journal of Dentistry, 57,1-6.
Campbell, T.D., Gray, J.H., and Hackett, CJ., 1936, Physical anthropology of the Aborigines of central
Australia. Part 1. Anthropometry. Oceania, 7, 106-139.
Corruccini, RS., 1999, How Anthropology Informs the Orthodontic Diagnosis of Malocclusion's Causes
(Lewiston: The Edwin Mellen Press).
Corruccini, RS., Townsend, G.C., and Brown, T., 1990, Occlusal variation in Australian Aboriginals.
Dempsey, PJ., Townsend, G.c., and Martin, N.G., 1999, Insights into the genetic basis of human dental
variation from statistical modelling analyses. Perspectives in Human Biology, 4(3), 9-17.
Grave, K.C., and Brown, T., 1976, Skeletal ossification and the adolescent growth spurt. American
Journal of Orthodontics, 69, 611-619.
Grave, K.C., and Brown, T., 1979, Carpal radiographs in orthodontic treatment. American Journal of
Orthodontics, 75, 27-45.
Greulich, W.w., and Pyle, S.l., 1959, Radiographic Atlas of Skeletal Development of the Hand and
Wrist. 2nd ed. (Stanford: Stanford University Press).
Lewin, T., and Hedegard , B., 1971, Secular changes in craniofacial dimensions of adult Skolt Lapps:
studies on population and family levels within a genetic isolate group. Suomen
Hammaslaakariseuran Toimituksia, 67,171-183.
Middleton, MR, and Francis, S.H., 1976, Yuendumu and its Children. Life and Health on an Aboriginal
Settlement. (Canberra: Australian Government Publishing Service).
Miller, P.S., 1970, Secular changes among the Western Apache. American Journal of Physical
Moodie, P.M., 1973, Aboriginal Health. (Canberra: Australian National University Press), pp. 181-187.
Preece, M.A., and Baines, MJ., 1978, A new family of mathematical models describing the human
Richards, L.C., and Brown, T., 1986, Development of the helicoidal plane. Human Evolution, 1, 385-
398.
Solow, B., 1969, Automatic processing of growth data. Angle Orthodontist, 39,186-197.
Townsend, G.C., 1976, Tooth Size Variability in Australian Aboriginals: A Descriptive and Genetic
Study. Ph.D Thesis, The University of Adelaide.
Townsend, G.c., 1980, Heritability of deciduous tooth size in Australian Aboriginals. American Journal
of Physical Anthropology, 53, 297-300.
Townsend, G.c., and Brown, T., 1978, Inheritance of tooth size in Australian Aboriginals. American
Townsend, G.C., and Brown, T., 1978, Heritability of permanent tooth size. American Journal of
Physical Anthropology, 49, 497-504.
Townsend, G.C., and Brown, T., 1979, Tooth size characteristics of Australian Aborigines. Occasional
Papers in Human Biology, 1, 17-38 (Canberra: Australian Institute of Aboriginal Studies).
Townsend, G.C., and Brown, T., 1981, The Carabelli trait in Australian Aboriginal dentition. Archives
of Oral Biology, 26,809-814.
CHAPTER 10
RIDDLES IN HUMAN DEVELOPMENT PATTERNS: A

FEW FAMILY CASE STUDY
R.D. SINGH
Department of Sociology and Anthropology, University of Windsor, Windsor, Ontario,
Canada
1. INTRODUCTION
Growth of an organism in biological terms begins with the fertilised ovum, followed
by the 'process of self-multiplication of living substance' (Malina 1975). After the
initial stages in the proliferation of cells, the generalised cells get involved in the
process of differentiation and specialisation as different functional units. This later
activity continues and the initial embryo begins to develop the species specific
bodily characteristics leading towards the birth and maturity of the individual. This
process is usually referred as the developmental process. Tanner (1990) does not refer
to such distinctly defined views (meanings). He is comfortable with the use of the
word growth in a wider perspective. In general, the word growth has been
interchangeably used for both in most of the writings on the subject. In a very recent
discussion in tracing the evolution of human growth Bogin (1 999) appears to have a
similar approach. Expressions in Bogin's (1999) book are no exception although the
glossary does suggest separate meanings for the two words. The meanings are
overlapping and in reality the two go side by side without specific boundaries,
except where a particular case (a feature) in point is under discussion.
Irrespective of the fine nuances between the two terms as some would prefer to
have for the sake of definition or clarity on specific issues, the fact of the matter is
that the initial genetic components and the environment, pre-natal as well as post-
natal, both contribute towards the attainment of final bodily attributes of an
individual. It, obviously, involves the interaction between the genetic and the
environment which, in tum, controls and directs the whole process of development.
Beyond the normal control of mechanisms of growth and development, factors like
nutrition, diseases, temporal infections, cultural constraints, geographical factors
affecting food supplies, periodic or long term, and many other unknown, heavily
influence the net results in the developmental outcome.
Potentials of genetic components under normal circumstances have been
substantiated by the twin studies, but a potential may not be fully reached because of
malnutrition or childhood diseases of a particular nature, or some unknown factors in
operation.
Some positive aspects of the influence of nutritional and health environments,
however, have been reported with regards to the stature in various human
123
124 CHAPTER 10
populations in their native countries (Meredith 1976), as well as among the

immigrant children in healthier and better nutritive environments (Meredith 1963).
It is believed that in all such cases, the conducive environmental factors permitted
individuals to reach their genetically determined potential for this character. Takaishi
(1995) reported continued increase in average height at a slower rate in Japanese men
and women. Stein and Rowe (1989) observe that a 'levelling off of the secular trend
appears to be occurring among the higher socio-economic, urban population'.
Slowing down has also been indicated by Kromeyer-Hauschild and Jaeger (2000).
On the other side of the spectrum is the progressive decline in stature in some
population groups in India (Ganguly 1979), and South African blacks (Tobias
1985).
Basically, the exact methods of mechanisms involved and the interaction
between the genetic and the environmental forces has not yet been fully understood
and sorted out in all cases and various wider situations; particularly the influence on
the genetic mechanism is unknown where normal development is involved
Secular trends in a population are less complicated to explain where major events
around a population can easily be ascertained and understood in terms of
environmental factors. But they do not provide grounds to explain the influences on
genic compositions or the interaction between the two. Mostly it is the family
studies which provide better insights into the working of the basic patterns of
development in the background of heredity and the specific environmental pressures
which may have affected the direction of development. With these considerations in
mind three family histories are presented here with regard to stature only, where it
does not appear to follow a simple pattern of expression.
2. MATERIALS AND METHODS

The three families come from a village, Patmau, in Uttar Pradesh, India. The
information covers the time span from the year around 1940 to date. The village as a
whole, including the members of the three families, is involved in day to day
farming operations at various levels of economic status. Family A belongs to the
agricultural labour caste, Chamar, who worked for other farmers in the village in the
past but presently only occasionally, particularly for the past twenty years or so.
They have acquired their own piece of land to farm, yet day to day labour activity is
mostly related with agricultural activities or as labours in building activities, in and
around the village.
Family B worked on its own land and did not work for others for the first two
generations. In the third and fourth generations, professions like teaching in primary
school system, and working in a cotton factory hadlhave been also adopted.
Family C has the similar background as B but hired more often the labours to
work on the farm since the acreage is larger and the members in the family could not
effectively manage on their own in the first two generations. The present (adults)
generation is mostly involved in business (general merchandising) and only a few
members indulge directly in farm operations on occasions.
The dietary habits and patterns of consumption of food in these families, as well
as the village and the surroundings, are not significantly different from one another.
The difference is mainly noticed in the consumption of vegetables on a regular basis
and the dairy products (milk, butter, buttermilk) only occasionally. Since family A
RIDDLES IN HUMAN DEVELOPMENT PATTERNS 125
is financially weak, occasional insufficient supply of staple food items like, wheat,
gram, lentils, rice and potato, etc. puts them in a disadvantageous situation. Meat
proteins hardly form 0.5% of the diet, but they do eat meat once in a blue-moon.
Families Band C are vegetarians.
Table 1 Stature of individuals (in cm) in successive generations in the three families
Family Generation Individual Sibling's stature Spouse

number
male female
A 1 (1) 172.7 (2) 144.8
2 (3) 171.5 x
(4) 162.6 x
(5) 170.2 (6) 162.6
3 - (7) 157.5 x
(8) 166.4 x
(9) 166.4 x
(10) 167.4 x
4 (11) 160.0 x
B 1 (1) 172.7 (2) 162.6
2 (3) 157.5 (4) 162.6
3 (5) 165.1 (6) 162.6
(7) 170.2 (8) 162.6
4 (9) 166.4 (10) 160.0
(11 ) 166.4 (12) 161.3
(13) 166.4 (14) 160.0
(15) 157.5 x
5 (17) 158.1 x
(16) 160.0 x
C 1 (1) 166.4 (2) 158.1
2 (3) 154.9 (4) 167.6
(5) 160.0 x
3 (6) 168.9 (7) 160.0
(8) 166.4 (9) 160.7
(10) 157.5 x
(11) 157.5 (polio) x
4 (12) 166.4 x
(13) 165.1
(14) 165.7
( 15) 165.1
(17) (16) 154.9

non-adult
(18)
non-adult
There was hardly any difference in the dietary patterns and items in earlier
generations which grew up before the 1960's. Until then the diet was mostly based
126 CHAPTER 10
on cereals, lentils, occasional use of potatoes and some dairy products - mainly milk
and some butter and buttermilk in average and well-to-do families in the village.
Among the dairy products, only buttermilk was occasionally available to family A.
The oldest members in the families A and B only grew up during World War I,
and the dry seasons as well as the depression of the early 1930's. The second
generation grew up during the 1930's and witnessed the societal phases in the 1940's
and after. Since all the families (including C) were directly dependent on village
economy (farming and labour) they were more or less influenced along similar lines.
Therefore, I presume that food supply and the physical environment had similar
influence on the growing children and sub-adults in the population. With this
background, I proceed to look into the patterns of stature attainment in the three
families stated above.
The three families scrutinised here are patrilocal and patrilineal. As a result, the
information is available only for the males in the family and their female siblings.
The information on the family lines of the daughters of the families married outside
and away from the village could not be obtained. The spouses of the males within
the village do not willingly participate in such observations. Hence their stature was
assessed in comparison and in relation to the stature of another member in the family
wherever possible. This problem was more acute with newly arrived brides in the
family. This situation gives only a limited reliability on the recorded figures
presented here. The number of individual families is small, although the information
on four successive generations is made available. In family C, the fifth generation is
also available. Also, all members of each generation in each family are not alive.
Family A is represented by the members in the third and fourth generations, family
B by fourth and fifth generations and family C by second, third and fourth
generations.
The readings for the stature of individuals were available in inches, and are
presented in Table 1 after conversion to centimetre.
Family A and C represent four generations, while B goes to the fifth generation.
The individuals in each family are designated serially, from the start to end. The
male and female siblings are identified; the respective spouses are listed separately.
The cross mark(s) in this column indicate the non-availability of the information.
The data for the children of the female siblings in a family were not available except
in family A because the sibling resided in the village, along with her brothers.
3. DISCUSSION
The limited reliability of the figures (for stature) and the number of individuals
observed, is a drawback over which there was, and is, no control. And possibly no
standardised statistical procedures would have yielded any more reliable results than
the systematic observations on the trend in the increase and decrease of stature in
successive generations in each family.
Family A indicates a gradual decrease in stature of the males in each generation.
In family B, there are ups and downs. Here, the male 3 (son) drops down
considerably in stature against his father with 172.7 cm; even 5.1 cm shorter than
his mother with 162.6 cm. The third generation sons (5 and 7) gain back in stature
(165.1 and 170.2 cm, respectively). The younger male (7) gains back a considerably
higher stature of 170.2 cm than his older brother who is only 165.1 cm. In
RIDDLES IN HUMAN DEVELOPMENT PATTERNS 127
generation 4, the two sons (9 and 11) have a slight gain in stature over their father
(individual 5). However, in the same generation 4, the son (13) drops in stature by
3.8 cm as against his taller father (7). Thus, we notice here first a considerable drop
in the second generation, and a gain in the third generation. The generation 4 males,
however, do not indicate a considerable variation as in generation 2 and 3. The
relative increase of stature in the third generation may be partially accounted for
better conditions of living (in terms of availability of food resources), but this factor
of diet does not seem to have any influence on individuals in generation 4, where
there is somewhat better dietary environment. A question may be raised here as to
what happened to the genetic factors commencing in the tallest male individual (1)
in the first generation and expressed only in one individual in the third generation?
At the same time how does one explain the extreme loss in the second generation
male. Extreme environmental conditions have not been reported. On the other hand
members in the fourth generation have not reported any obvious gain back with
somewhat still better dietary resources.
Family C commences with average stature in the population, but a considerable
drop in stature similar to family B in the second generation is noticed. The third
generation gains in stature over their father (154.9 cm) but certainly not over their
mother (167.6 cm) except in one case (6). The males in generation 4, maintain more
or less (within 1.3 cm) the stature of the third generation, but slightly shorter than
their respective fathers. Undoubtedly, the nutritional environment for the generations
3 and 4 was better during the growth period than the generation 2. The dietary
environmental factors, here, may be interpreted to have lent their influence in stature.
But the introduction of a taller mother in the second generation must not be
overlooked. Her stature of 167.6 cm is considerable for a woman in that area, as well
as in the community. However, the lone female child (5) remains considerably
smaller as compared to her mother, though within the normal range of stature for
females; the allele(s) for the tall stature from the mother may have worked (if at all)
along with the improved nutritional environment for males only.
In family A where there is gradual decline in stature, individuals (as spouses) of
obviously higher or lower stature have not been added, and the nutritional
environmental factors have remained almost similar.
For specifically short stature, Family A is marked by individual 2, a female,
family B by individual 3, a male and family C by individual 3, also a male. The
spouses of these shorter individuals were taller and the fathers of the shorter males in
B and C were taller as well.
An overview of the data, thus, does not reveal a clear mode of inheritance and the
expression of stature attainment. If the dietary environment is to be considered as a
major factor, the consumption of some milk in family Band C in their third and
fourth generations may have played a role to attain the potential, which, in the
circumstances of the families, may be considered a responsible factor, since the
'stature is a dynamic phenotypic trait, one that is very responsive to the quality of
the environment for growth' (Bogin 1999). In reference to his discussion on the
inter-generational effect hypothesis, Bogin further comments that 'a child's height is
a historical record of both the individual and his or her parents'. The historical
record in the three families does point to the fact that average, or taller individuals
did exist among the parents of the individuals, who were either very short or average
and taller. Here, then, the arguments may be put forward in favour of the hereditary
128 CHAPTER 10
factors. Where the quality of environment for growth was better, the trait responded
in favour of taller genetic potential. But in the extreme cases of shortness, as in the
family Band C, the adverse environment for growth has not been reported. These
individuals do not even fall in between the stature of their parents. The question for
such a drop begs an explanation. Besides, family A registers a continuous downward
trend in more or less similar environment for growth.
Thus, the riddle in the expression of stature influenced by genetic or
environmental factors still remains unclear. It appears imperative that more refined
tools to examine the mode of working of the genetic factors, in a given environment
which influences it, are needed.
4. REFERENCES
Bogin, B., 1999, Evolutionary Perspectives on Human Growth. In Annual Review of Anthropology,
volume 28 edited by W.H. Durham (Palo Alto: Annual Reviews) pp. 109-310.
Bogin, B., 1999, Patterns of Human Growth (Cambridge: Cambridge University Press).
Ganguly, P., 1979, Progressive Decline in Stature in India: a study of sixty population groups. In
Physiological and Morphological Adaptation and Evolution, edited by M.A. Stini (Hague: Mouton),
pp.315-317.
Kromeyer - Hauschild, K., and Jaeger, U., 2000, Growth Studies in Jena, Germany: changes in sitting
height, biacromial and bicristal breadth in the past decenniums. American Journal of Human
Biology, 12,646-654.
Malina, R.M., 1975, Growth and Development: The First Twenty Years in Man (Minneapolis: Burgess).
Meredith, H.V., 1963, Changes in stature and body weight of North American boys during the last
eighty years. In Advances in Child Development and Behaviour, volume 1, edited by L.P. Lipsitt
and C.C. Speker (New York: Academic) pp. 63-114.
Meredith, H.V., 1976, Findings from Asia, Australia, Europe, and North America on secular changes in
mean height of children, youths, and young adults. American Journal of Physical Anthropology, 44,
315-326.
Stein, P.L., and Rowe, B.M., 2000, Physical Anthropology (New York: McGraw-Hili).
Takaishi, M., 1995, Growth standards for Japanese children - an overview with special reference to
secular changes in growth. In Essays on Auxology, edited by R. Hauspie, G. Lindgren and F.
Falkner (Welwyn Garden City, U.K. Castlemead) pp. 302-311.
Tanner, J.M., 1990, Foetus into Man: Physical Growth from Conception to Maturity (Cambridge,
Massachusetts: Harvard University Press).
Tobias, P.V., 1985, The negative secular trend. Journal of Human Evolution, 14,347-356.
CHAPTER 11
NUTRITION IN VENEZUELA AT THE END OF THE

MILLENNIUM
M. L6PEZ-BLANCO
Fundaci6n Bengoa. Caracas-Venezuela
1. VENEZUELA: A DEVELOPING SOCIETY IN THE MIDST OF A CRISIS.

TRENDS AND SECULAR CHANGES.
Industrialisation and socio-economic changes came about relatively late in
Venezuela: the 1930's are regarded as a turning point. The population grew from less
than 4 million to more than 19 million inhabitants in 1990, and is estimated in 25
million at the end of the millennium: it is still growing at a 2.5% rate, due to a
sharp decline in death rates (27.1%0 in 1935; 4.6%0 in the 1990's) and a still high
birth rate of 24.5%0 in the 1990's. The structure of the population is typical of a
developing society: 45.8% below 18 years of age and 10.8% above 50 years of age,
although trends indicate a decrease in the population under 18 and a great rise in the
adult population. On the other hand, urbanisation has been so intense that the rural
population decreased from 75% in the 30's to 14% at the end of the 20th Century
(Paez Celis, 1995).
Life expectancy at birth rose from 38 years in 1985 to 72 years in 1990 (70 years
for men and 75 years for women): more than 30 years have been obtained in life
expectancy in less than 60 years. Literacy increased from 40% in 1936 to 92% in
1990 and infant mortality decreased from 137.5%0 live births in 1935 to 21%0
presently, although it is still high in comparison to developed nations. Causes of
death have also changed: In 1935,40% of deaths were due to environmental causes,
mainly infections: malaria, tuberculosis, diarrhoea, while in 1990 these dropped to
12%, and Non Communicable Chronic Diseases (NCCD) such as cardiovascular
diseases, cancer and diabetes, are now the main causes of death (Paez Celis, 1995)
The start of the positive secular changes in growth and maturation coincides with
the moment that industrialisation began to improve socio-economic conditions. The
pattern of the increase: moderate in children, maximal during puberty and smaller in
adults, is similar to the trends found in developed countries and in accord with the
statement that the tendency towards a larger size is due to a more rapid maturation,
although part of it persists into adulthood. This secular trend has been higher in
males and greater for weight than for height. Venezuelans today are significantly
taller and heavier than in the 1930's (L6pez-Blanco, 1995).
130 CHAPTER 11
Van Wieringen states that "it is attractive to consider contemporary differences in

growth and maturation among population groups as differences in the stages of the
secular process" (van Wieringen, 1986). In fact, the faster secular changes in the
lower socio-economic groups have resulted in the obliteration of social and urban
gradients in some countries such as Sweden and Norway (Brundtland, 1980;
Lindgren, 1976). In Venezuela, however, significant social and urban-rural gradients
are still present. Furthermore, if adverse conditions continue, a slowing down of the
secular trend could be expected, due to the inhibiting effects of the economic crisis,
with the widening of the gap between the population groups (Lopez-Blanco, 1995).
2. THE GREAT SOCIAL DISPARITIES: DIFFERENCES IN SOCIAL STRATA.

STRUCTURAL AND CONJUNCTIONAL POVERTY.
These trends, however, do not reflect the great social disparities. In 1982, according
to the Graffar- Mendez Castellano method (that measures structural poverty), 5.5% of
the population belonged to Social Strata (SS) I and II, 14.1 % to SS III (the middle
stratum), 42.4% to SS IV and 38% to SS V, while in 1990, SS I and II had
increased to 8.2%, SS III and IV had decreased to 13.6% and 37.8% respectively, and
SS V had increased to 40.4% (Mendez-Castellano et aI., 1990).
Also a social gradient in growth and maturation has been found in height and
weight: boys and girls from the upper strata are significantly taller and heavier than
those of the middle stratum, and these last taller and heavier than those of the lowest
stratum. The greatest differences have been found during puberty between the upper
and the lowest strata children, with a maximum of 6 and 5 cm in boys and girls,
respectively; in weight, differences reached 7 kg, in boys and almost 6 kg in girls
(Lopez de Blanco et al. 1995).
This great social disparity is also present when one considers what is known as
conjunctional poverty, using the method of "The Line of Poverty" (OCEI, 1993)
that reflects income versus inflation. At the end of 1997, total poverty was 70%
(income equal or less than twice the cost of one "Food Basket") while extreme
poverty (when the family income only manages to cover one "Food Basket" or less),
reached 39%.
3. THE SITUATION IN FOOD AVAILABILITY AND INTAKE, NUTRITIONAL

STATUS AND SECULAR TREND: UNDERNUTRITION VERSUS "HIDDEN
HUNGER".
Food availability, in terms of energy (kcal/personJday) is referred to three levels: A
high reference value, an average reference value and a low reference value: defined for
Venezuela as 2600, 2450 and 2300 kcal/personJday. In Figure 1, a downward trend is
evident. When requirements are taken into account, in the follow-up between 1989
and 1999, one observes that availability from 1992 onwards was clearly insufficient,
reaching 2006 kcal/personJday: 87.2% in 1999. Protein availability follows a similar
pattern, although in the last three years it remained at a higher level (91-97%). With
regard to iron, the low availability levels up to 1992 are followed by high levels, due
to the fortification of maize and wheat flour. (INN and ULA, 1989-1990).
NUTRITION IN VENFZUELA 131
130
l
21
c
<U • Energy
E
~ 120
• Protein
.:; * Iron
0"'
~
B 110
00
c
...
:i:)
0
(.)
<.)
100
'"00
0)
.fl
c0)
90
~
0)
p...
80
1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
~ri~
Figure 1. Levels of sufficiency in energy, protein and iron availability 1989 - 1999.
Instituto Nacional de Nutrici6n - Universidad de Los Andes. Roja de Balance de Alimentos
(1989 - 1999). Data adjusted by Magaly Bracho and Magaly Rangel (INN)
130
• 1994
• 1995
~ 1996
120
o 1997
~ 1998
s:: 110 ~ 1999
~ 100
90
80
Stratum III Stratum IV Stratum V
Figure 2. Energy intake adequacy according to social strata in Caracas 1994 - 1999.
Fundacredesa - Estudio sobre condiciones de vida 1994 - 1999.
l32 CHAPTER 11
With regard to energy intake adequacy, according to our national requirements, it

rose slightly between 1990 and 1994 from 100.2% to 102.7%, yet declined to 98.3%
in 1997. Chronic calorie insufficiency is thus the main problem in our country;
also, we are greatly dependant on imported goods. When energy intake is analysed
according to social strata (Fundacredesa, 1998a; Fundacredesa, 1999b; Fundacredesa,
2000) a clear gradient is found in intake adequacy between strata III, IV and V. The
lowest percentage values correspond to stratum IV and V in 1998 and 1999. In fact,
1998 stands out as the most critical year with l3% of insufficiency in the lowest
stratum and, even in stratum III (middle class families), only 94% of adequacy was
reported (Figure 2).
16
15
~fj 14
~
13
12
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Period
Figure 3. Low weight-for-age prevalences in 0 - 2 year olds in Venezuela: Instituto
Nacional de Nutricon (INN) - Sistema de Vigilancia Alimentar{a Nutricional (SISVAN)
1990 - 1999. cut-off point: percentile 10 of NCHS/OMS
40
35
'E
cu
~
cu
30
Q..
25
20 90 91 92 93 94 95 96 97 98 90 91 92 93 94 95 96 97 98 99
2 - 6 years 7 - 14 years
Figure 4. Undernutrition prevalencest in children 2 - 14 years of age in Venezuela 1990 -
1999. Instituto Nacional de Nutricion (INN) - Sistema de Vigilancia Alimentar{a y
Nutricional (SISVAN) 1990 - 1999. t including low weight-far-height and low height-for-
age. Cut-off points: percentiles 10 and 3 of NCHS/OMS, respectively.
NUTRITION IN VENEZUELA 133
Anthropometric indicators, analysed by the Sistema de Vigilancia Alimentaria y

Nutricional (SISV AN) show that the nutritional status of children from 2 to 6 years
of age and from 7 to 10 years of age improved between 1990 and 1994, after which
physical status -in terms of height for age, weight for age and weight for height-
began to deteriorate, although in 1999 a slight improvement is observed (Figures 3
and 4). On the other hand, overall prevalence of undernutrition are not as high as
expected: 23.4% in young children, 25.6% in 7 to 14 year old, although prevalences
vary from low values of 10% in the most industrialised states to over 35% in the
poorer states (INN/SISV AN, 1998; INN/SISV AN, 1999).
It is important to point out that in several studies carried out in Venezuela where
undernutrition has presented low prevalence, "hidden hunger" was found. "Hidden
hunger" has been defined as the subclinical phase of undernutrition, whose indicators
include skinfolds, as well as biochemical variables, such as iron, calcium and
vitamin "A" determinations (Lopez de Blanco, 1999). In several studies carried out in
infants and small children in which nutritional status was in the normal range,
"hidden hunger" -especially anaemia and iron deficiency- was prevalent in 50% of the
samples (Fundacredesa, 1998a). Also, low birth weight prevalence have risen from
9.6% in 1989 to 14.4% in 1998 in Maternidad Concepcion Palacios, Caracas largest
maternity hospital (INN/SISVAN, 1998; INN/SISVAN, 1999). As shown in Figure
5, the tendency is most evident from 1997 onwards.
15
14
13
f 12
~ 11
10
9
1989 1990 1991 1992 1993 1994 1995 1996 1997 1998
Period
Figure 5. Low birth weight prevalence in Caracas 1989 - 1998: Instituto Nacional de
Nutrici6n (INN). Sistema de Vigilancia Alimentar{a Nutricional (SISVAN) 1989-/998.
A preliminary report indicates that the positive secular trend in height and weight
in the SISV AN data shows a tendency to reverse. However, data from an ongoing
study of Fundacredesa, clearly shows that the secular trend (at selected ages of
9.11,13 and 15 years of age in girls and boys) is still present, although higher in
height than in weight and most evident at ages 9 and 11 (Mendez de Perez, Landaeta
de Jimenez, Ledezma and Ortega, 1999). Also, the magnitude of this trend, analysed
between 1984 and 1995, (0.5-1 cm/decade) is lower than the one reported previously
134 CHAPTER 11
between 1936 and the 1980's, probably due to the socio-economic crisis of this
decade (LOpez-Blanco et at. 1989).
4. THE CONCEPT OF A DEMOGRAPHIC, EPIDEMIOLOGICAL AND

NUTRITIONAL TRANSITION: PREVALENCE OF UNDERNUTRITION AND
OVERNUTRITION.
As reported by SISVAN, in children under 15 years of age, overnutrition in terms of
weight for height seems to be as much of a problem as undernutrition. Furthermore,
cardiovascular disease is the first cause of death in Venezuela since 1953, with a
mortality rate of over 1501100.000. This means that Venezuela is in the midst of a
demographic, epidemiological and nutritional transition. In fact, the traditional
concept of undernutrition in poverty and overnutrition in wealth is not real anymore:
in one family -even of the lowest social level- one can find an undernourished infant,
an anaemic pregnant mother and an obese father with a high risk of NCCD. This is a
pattern found in many developing nations and must be taken into account in the
future, when policies and planning for food, nutrition and health are considered in
Venezuela at the start of the new millennium.
As a final statement, Tanner's concept on the growth of children reflecting the
material and moral conditions of a society (Tanner, 1986) is more commanding than
ever in Latin America and specifically in Venezuela.
5. REFERENCES
Brundtland, G. H., 1980, Height, weight and menarcheal age of Oslo school children during the last 60
years. Annals of Human Biology, 7, 307-322.
Fundacredesa, 1998a, Indicadores de Condiciones de Vida 1996-1997, Area Metropolitana de Caracas.
(Caracas: Fundacredesa)
Fundacredesa, 1998b, Estudio Impacto del Enriquecimiento de Harinas con Hierro y Vitamina A en la
Poblaci6n Venezolana. 1998 (Caracas: Fundacredesa)
Fundacredesa, 1999, Indicadores de Condiciones de Vida 1998, Area Metropolitana de Caracas
(Caracas: Fundacredesa).
Fundacredesa, 2000, Indicadores de Condiciones de Vida 1999, Area Metropolitana de Caracas.
(Caracas: Fundacredesa)
Instituto Nacional de Nutrici6n (INN)/Sistema de Vigilancia Alimentaria y Nutricional (SISV AN),
1998, Boletin Informativo 1994-91.
Instituto Nacional de Nutrici6n (INN)/Sistema de Vigilancia Alimentaria y Nutricional (SISV AN),
1999, Boletin Informativo.
Instituto Nacional de Nutrici6n (INN) and Universidad de los Andes (ULA) 1989-1999, Hojas de
Balance de Alimentos. (Merida: Talleres Gnificos Universitarios ULA).
Lindgren, G., 1976, Height, weight and menarche in Swedish urban school children in relation to
socioeconomic and regional factors. Annals of Human Biology, 3, 501-528.
L6pez de Blanco, M., Landaeta de Jimenez, M., Izaguirre de Espinoza, I., Macias de Tomei, c., 1995,
Crecimiento Fisico y Maduraci6n. In Estudio Nacional de Crecimiento y Desarrollo Humano de la
Republica de Venezuela: Proyecto Venezuela. (Vol 2), edited by H. Mendez Castellano. (Caracas:
Escuela Tecnica Popular Don Bosco), pp 695-773.
L6pez de Blanco. M., 1999, EI pediatra ante el Hambre Oculta. Anales Venezolanos de Nutrici6n,
12,129-136.
L6pez-Blanco. M., 1995, Growth as a Mirror of Conditions of a Developing Society: The Case of
Venezuela. In Essays on Auxology. Presented to James Mourilyan Tanner, edited by R Hauspie, G
Lindgren, Falkner (London: Castlemead Publications), pp 313-321.
L6pez-Blanco. M., Landaeta-limenez. M., and Mendez-Castellano. H., 1989, Secular trend in height
and weight: Carabobo, Venezuela. 1978-1987. In Auxology 88: Perspectives in the Science of
Growth and Development, edited by. J.M. Tanner. (London: Smith Gordon), pp 207-210.
NUTRITION IN VENFZUELA 135
Mendez-Castellano. H., L6pez-Blanco. M., Mendez, M., Fossi, M., Landaeta-Jimenez. M., and Bosch,
M., 1990, The social impact on child growth and development in Venezuela. In Malnutrition and
the Infant Brain: Neurology and Neurobiology. (New York: Wiley Liss Inc), pp 269-284.
Mendez de Perez, B., Landaeta de Jimenez. M., Ledezma, T. and Ortega de Mancera. A., 1999,
Tendencia secular en peso y talla entre 1984 y 1995 en niiios y j6venes venezolanos. Anales
Venezolanos de Nutrici6n, 12,117-122.
Oficinal Central de Estadfstica e Informatica (OCEI) 1993-1998, Mapa de la Pobreza: Basado en los
resultados del XII censo de poblaci6n y vivienda 1990.
Paez-Celis, J., 1995, Marco situacional de la Poblaci6n Venezuela. In Proyecto Venezuela 1981-
1987.(Vol. I), edited by H. Mendez Castellano (Caracas: Fundacredesa). pp 17-43
Tanner, J. M., 1986, Growth as a mirror of the condition of society: secular trends and class distinctions.
In Human Growth, a Multidisciplinary Review, edited by A. Demirjian and M. Brault-Dubuc.
(London and Philadelphia: Taylor and Francis), pp. 3-24
van Wieringen, 1. C., 1986, Secular Growth Changes. In Human Growth, a Comprehensive Treatise,
(Vol 3), edited by F. Falkner and J. M. Tanner. (New York and London: Plenum Press, 2nd edn),
pp. 307-331.
CHAPTER 12
THOUGHTS ON SECULAR TRENDS IN GROWTH AND

DEVELOPMENT
P.B. EVELETH
US National Institutes of Health, Maryland, USA
1. INTRODUCTION
Changes in body size, fatness, rate of growth, and timing of maturation have been
occurring over the past 150 years. We also are witnessing a decline in mortality, an
increase in life expectancy and a large increase in the proportion of the population
that is elderly in both industrialised and developing countries.
We assume that the secular increase in size and earlier maturation is a result of
improved environmental conditions within a population. These have led to improved
health status of the population as a whole. Similarly, secular decrease in size or
delayed maturation may come about because of negative conditions. But the reasons
remain only partially uncovered. Quite assuredly, many factors help children to more
closely approach their full genetic potential. Among these are improved nutrition,
more widespread health and medical care, higher education level of mother (maybe
the father, too), better housing and living conditions, better transportation, clean
water, improved sanitation, immunisation programs and population mobility, both
geographically to urban areas and socially upward. This can be linked, both in time,
and in conditions with the Industrial Revolution in Europe and the United States
when there were tremendous changes in society, some beneficial and some
detrimental. Early in the Industrial Revolution in Europe during the late 18th and
early 19th centuries, children were employed in factories and mines and working
families were densely packed in cities. Conditions were such that children were short,
birth weight was low and mortality was high (Tanner 1981, 1986). Only later, in the
early 20th century, when advances in science and technology improved standards of
living of ordinary people, did height begin to rise (Fogel 2000). Today many
developing countries are experiencing industrialisation with the shift from an
agrarian economy to a manufacturing and commercial one accompanied by a shift in
population from the country to the cities. Some already are witnessing an increase in
children's height and weight, and a shift to the right in distribution of Body Mass
Index (BMI).
137
138 CHAPTER 12
2. QUESTIONS ABOUT THE FUTURE

As conditions improve in developing countries and children become healthier and
grow larger in size, does this mean that with time all children in the world will grow
the same? Growth differences among social classes in some countries are narrowing.
Will population differences among countries disappear? Will there be no world-wide
variation in human growth?
We have some clues as to what could happen from two types of studies: (1)
children from refugee families who migrated to the US from Asia and (2) orphans
from Asia who have been adopted by European and American families. Generally the
immigrant studies show an improvement in growth in height and in growth rate.
Ray Yip and his colleagues at The Center for Disease Control and Prevention (CDC)
(Yip et aI., 1992) have looked at the growth of children, under 5, mostly from
Southeast Asian refugee families who were screened from 1980-89 under the US
Pediatric Nutrition Surveillance System. They reported that over that period the
prevalence of low height-for-age in Asiatic children decreased from 30% to 15% in
children from 2 to 5 years of age as seen in Figure 1, but Asians still had a higher
prevalence of low height than other ethnic groups in the samples (Hispanics, blacks,
whites). Similarly in Greulich's classical studies of Japanese immigrants to
California, first and second generation children were taller than sedentes but they
were not as tall as Californians (Greulich, 1957).
35
30 • Asiatic
·········+·····~·········+···········i··· • White
• Hispanic
25 )( Black
.....
§ 20 ..........~ ............ +- .......... + ........... ~ ....... ·····r··~·········· ················r ············r··········
u
~
~ : : : : : :
15
. . . . ~. . . . ~. mN··m
10
5 . . . . .:. . . . . ., . . . . ;············:·········I·············\····;·i······:···. . . . ". . . .

o L I_ _~_ _ _ _~_ _ - L_ _~_ _ _ _~_ _- L_ _~_ _ _ _~_ _- L_ _~_ _~
197919801981198219831984198519861987198819891990
Period
Figure 1. Prevalence of low height-for-age (ages 24 through 60 months)

SECULAR TRENDS IN GROWTII AND DEVELOPMENT 139
There are a number of adoption studies, all showing catch-up growth in height
and weight, the greater the stunting, the greater the acceleration of growth. In a study
of Indian girls adopted in Sweden at 1 month to 11 years of age, Proos (1993)
reported considerable improvement in growth. However, she noted shorter final adult
height in some cases in spite of rapid growth because menarche began sometimes as
early as 7-9 years of age, thereby limiting much additional height growth.
Some health workers in developing countries do believe that differences among
populations in child growth will disappear (Habicht et al. 1974). That indeed is the
basis for the World Health Organisation's (WHO) proposition to construct a new
international growth reference (Garza and de Onis 1999). This would be a single
standard for the world's children and replace the current WHO reference which is
based on the US National Center for Health Statistics data. It has been proposed that
seven geographic sites be selected from North and South America, Sub-Saharan
Africa, Europe, and East, South and West Asia. In order for new-borns to be selected
the parents must follow WHO feeding recommendations, i.e. exclusive breast feeding
for the first 4 to 6 months and partial breast-feeding until 12 months of age.
3. TALLER
It may be surprising to some but it appears that humans are not taller today than
they have ever been in history or prehistory.
According to long bone measurements of some fossil men reported by Styne and
McHenry (1993) and shown in Table 1 early European and West African males were
7 cm taller than modern Homo sapiens.
Table 1. Evolution of height (in em) in the Hominids (modified from Styne and McHenry
1993)
Species Males Females

Australopithecus afarensis 151 105
Australopithecus africanus 136 113
Homo habilis 157 100
Homo erectus 176 176
Homo sapiens, Neanderthals (West Africa) 170 157
Homo sapiens, Neanderthals (Europe) 167 160
Homo sapiens (early, West Africa) 184 169
Homo sapiens (early, Europe) 184 167
Homo sapiens (modern, USA) 177 163
Records of US Civil War soldiers have revealed that their average stature was
considerably lower than that of American males 100 years earlier, showing a secular
decrease in stature. Fogel (2000) proposes that the largest contributor to this decrease
was over-rapid urbanisation, combined with poor sanitation and diet. In the twentieth
century average heights in the US and in some European countries began to rise, as
shown in Table 2, probably due to improved nutrition as the food supply increased
(Fogel 1986, 1993, 2000).
140 CHAPTER 12
At other times and in other places conditions have been right for optimal growth
in height. It is quite likely that tallness has been cyclical throughout human history.
Table 2. Estimated average heights (in em) of men who reached maturity between 1750
and 1875 (modified from Fogel 1986, 1993)
Years Great Britain Sweden France

1750 - 1774 165.9 168.1
1775 - 1799 167.9 166.7 163.0
1800 - 1824 168.0 166.7 164.3
1825 - 1849 171.6 168.0 165.2
1850 - 1874 169.3 169.5 165.6
1950 - 1974 175.0 177.6 172.0
Taller populations may be healthier populations and more successful ones.

Tanner (1981, 1986) and others consider growth of children to reflect the conditions
of society. This was certainly the situation in the beginning of the Industrial
Revolution and still is in Third World countries and among some groups in
industrialised nations. For example, lower socio-economic status (SES) children
have lower heights on average than upper class, as shown in Table 3 from some
Polish data (Bielicki, 1986). These results were confirmed by two larger and more
recent Polish surveys showing that taller young men (19 years) were likely to
advance in educational and/or occupational status than their shorter peers (Bielicki,
2000).
Table 3. Height of 18-year-old sons of college-educated men from Warsaw and of 19-year-
old conscripts from other SES levels (adapted from Bielicki 1986)
Father's occupation Number of N Mean height,

siblings cm
Warsaw, academic education 1-2 47 178.2
Small town, unskilled manual workers 2-4 1009 172.9
Rural skilled manual workers 2-4 420 172.6
Rural unskilled worker 2-4 480 172.3
Peasant farmer 2-4 1272 171.9
However, the class difference in height apparently has disappeared now in Sweden
(Lindgren 1976) and perhaps in Norway (Brundtland et aI., 1980). Not only is the
assumption made that a positive secular trend reflects an improvement in social
conditions, the recommendation is that child growth be used to monitor conditions
in the population at large (Tanner 1981). It is assumed that this is due to
environmental effects but recently Bielicki (2000) presented one hypothesis (among
others) of a genetic influence.
There are public health and societal implications involving an increase in height.
Tall individuals may be healthier. Fogel et al. (1982) have shown that large size is
highly correlated with increased work capacity and labour productivity and also with
lower mortality. Waaler (1984), working with extensive Norwegian data, also
demonstrated reduction in mortality with greater height. All adults in Norway (except
SECULAR TRENDS IN GROWTH AND DEVELOPMENT 141
Oslo) were measured between 1963 and 1975. Males and females from 40 to 59
years, separated into five year age groups, showed decreasing relative mortality by
increasing height except for the very tall (> 185 cm for men and > 170 cm for
women).
Taller people have lower risk of myocardial infarction than shorter people
(Palmer et at. 1990, Barker et at. 1989). Tall women have fewer low birth weight
babies. On the negative side, Micozzi (1987) and other researchers have reported an
increase in cancer incidence at some sites and in cancer mortality with increasing
height (Table 4).
Table 4. Correlation of age-specific mean stature with age-adjusted breast cancer mortality
(adapted from Micozzi 1987)
Age Number of Pop. Correlation coefficient

6 23 0.49
8 29 0.56
10 31 0.53
12 30 0.56
14 28 0.62
16 24 0.77
18 15 0.72
4. FATTERIHEAVIER
The secular trend also is towards heavier and fatter people. There are clear public
health implications of the secular trend toward more overweight and obesity. The US
national surveys show that average weight, triceps and subscapular skinfold have
been rising in children and adolescents over the past 20 years.
Table 5. Estimated prevalence of obesity in US black and White children and adolescents:
based on triceps skinfolds ~ 85th percentiles HES 11 & III (from Malina 1993, as adapted
from Gortmaker et at. 1987)
Males Females
Age group/survey Black White Black White
6-11 years
RES cycle II 8.1 19.5 9.5 18.6
NHANES I 9.6 25.0 20.0 22.8
NHANESII 16.6 31.5 20.9 26.0
12 - 17 years
RES cycle II 7.5 16.7 12.8 16.6
NHANESI 8.9 17.6 20.4 24.5
NHANESII 12.7 19.5 25.1 25.6
Table 5 shows the increasing prevalence of obesity in black and white children from
the Health Examination Survey (HES), collected 1963-70, the National Health and
142 CHAPTER 12
Nutrition Examination Surveys (NHANES I), collected 1971-4, and NHANES II,
collected 1976-80, (Gortmaker et al. 1987, Kuczmarski 1993).
The trend is not limited to the US. In fact, Price et at. (1993) have suggested
there may be a world-wide epidemic of obesity. Studies documenting the
unfavourable health outcomes of overweight and obesity in children and adults are
increasing as rapidly as the problem itself. The well-recognised links to diabetes,
cardiovascular disease, gall bladder disease and some site-specific cancers leaves little
question as to the increased risks involved in an individual being heavier or fatter.
Furthermore, the population prevalence of these chronic diseases is increasing as
well. There are social implications of obesity since in Europe and the US, the
increase in prevalence is found among the lower, not upper, socio-economic groups.
5. PROPORTIONS
Body proportions change with secular increase in stature. In Japan and in China the
increase in stature was due almost entirely to an increase in leg length. Among
Japanese children relative leg length has increased in the past four decades (Ali 2000).
The major part of the increase in height of Norwegian recruits between 1921 and
1962 was due to increase in leg length (Udjus 1964). Similar results were seen in
Harvard students from the late 1900's to 1930 (Himes 1979). Previously we had
thought body proportions were immutable, but it appears that is not so. The reason
may be that the greatest influence of environmental improvement occurs during early
childhood when the legs are the fastest growing part of the body. Thus, the
relationship of sitting height to stature is altered.
6. SEXUAL MATURATION
Secular change also has been occurring in rate of maturation. This is shown by age
of menarche which has been getting earlier during the past 100 years, by some three
to four months per decade, in most European countries. The trend seems to be
slowing down in most developed countries or even stopped; but one cannot predict
when these changes will cease. Many environmental factors influence the mean age
of menarche. Some that have not been considered until recently are hormone
disrupters, especially man-made oestrogen mimics, such as DDT, polychlorinated
biphenyls (PCB), and organophosphates. It is thought that these may cause earlier
sexual maturation (Colborn et at. 1996, Holden 1997). Since menarche is brought
on by the heightened activity of the sex hormones one must consider the possible
influence of natural phyto-oestrogens and oestrogen mimics in the environment.
7. NATURAL RESOURCES
The secular trend will influence a nation's resources and economic development. On
the one hand, taller people require more calories for maintenance, while on the other
hand, taller people have higher productivity and less disease (WHO 1995).
Fogel in his Nobel Prize address (Fogel 1993) summarised some of his work on
secular trend in energy intake, productivity and economic development.
Significantly, he believes that the food supply in Europe during the eighteenth
SECUlAR TRENDS IN GROWTH AND DEVELOPMENT 143
century would not have been sufficient to maintain tall people and have much energy
left over to sustain work. Hence stature remained low as height data from Europe
show (see Table 2). The trend in increasing height came as nutrition improved
among all classes of people.
Other studies on work capacity and BMI in India and some African countries
indicate lower work capacity with lower BMI (Shetty and James 1993). In a sample
of llOO Rwanda women those with low BMI spent more time resting and less time
doing productive work (Figure 2) .
.£50
>-
45
.~ 40 ~ ...
~
'0
...
•
~
BMI < 17.1
BMI 17. 1 - 17.5
CIl
Q) 35 ... [Ill] BMI 17.6 - 18.6
;:-
~
:: 30
0 BMI 18.7 - 23.8
~
t
~
25
15
~ 20
.....
(\3
~ 15
8
.';:: 10
.....
c~
(.) 5
I-<
~
A... 0
laying down maintenance heavy work
Level of activity
Figure 2. Effect of BMl on physical activities of Rwandese women, 1982 (after Fram;ois
P., FA 0, 1990, unpublished data)
It is important to document the ongoing secular trend both positive and negative.
The world is changing at a rapid pace and many of these changes may influence child
growth, such as, the revolution in Yugoslavia, violence in Israel and Indonesia,
hurricanes in Mexico, floods in Bangladesh, AIDS in Africa, East Asia, Europe and
the Americas, newly emerging diseases, development of new drugs and access to
them, drought in Africa and Australia. Health workers, economists and politicians
now understand the value in recording anthropometric measurements of well-defined
samples at regular intervals to use in the evaluation of past policies and current
conditions. It is hoped that this will influence the formulating of national policies in
the future.
144 CHAPTER 12
REFERENCES
Ali A., 2000, Secular changes in relative leg length in post-war Japan. American Journal of Human
Biology, 12,405-416.
Barker, D.J.P., Osmund, c., Golding, J., Kuh, D., and Wadsworth, M.E.I., 1989, Growth in utero, blood
pressure in childhood and adult life and mortality from cardiovascular disease. British Medical
Journal Journal, 298, 564-567.
Bielicki T., 1986, Physical growth as a measure of economic well-being of populations: The twentieth
century. In Human Growth - A Comprehensive Treatise, 2nd ed., vol 3, edited by F. Falkner and
1.M. Tanner (Plenum Press, New York), pp 283-305.
Bielicki T., and Szklarska, A., 2000, Are social class differences in stature partly genetic? A hypothesis
revisited. American Journal of Human Biology, 12,97-101.
Brundtland, G.H., Leistol, K., Walloe, L., 1980, Height, weight and menarcheal age of Olso
schoolchildren during the last 60 years. Annals of Human Biology, 7, 307-322.
Colborn, T., Dumanoski, D., and Myers, J.P., 1996, Our Stolen Future (Penguin USA, New York).
Fogel, R.W., 1986, Physical growth as a measure of the economic well-being of a popUlation: The
eighteenth and nineteenth centuries. In Human Growth - A Comprehensive Treatise, 2nd ed., vol
3, edited by F. Falkner and J.M. Tanner (Plenum Press, New York), pp 263-281.
Fogel, RW., 1993, Economic growth, population theory, and physiology: The bearing of long-term
processes on the making of economic policy. Presentation as the Prize Lecture in Economic
Sciences in Memory of Alfred Nobel, December 9, 1993.
Fogel, RW., 2000, The Fourth Great Awakening & the Future of Egalitarianism (University of Chicago
Press, Chicago).
Fogel, RW., Engerman, S.L., and Trussell, J., 1982, Exploring the uses of data on height: the analysis of
long-term trends in nutrition, labor welfare and labor productivity. Social Science History, 6, 401.
Garza, C., and de Onis, M., 1999, The selection of reference data in the assessment of growth: the new
World Health Organization reference. In Human Growth in Context edited by F.E. Johnston, B.
Zemel, P.B. Eveleth (Smith-Gordon, London).
Gortmaker, S.L., Dietz, W.H. jr., Sobal, A.M., Wehler, C.A., 1987, Increasing pediatric obesity in the
United States. American Journal of Diseases of Children, 141, 535-540
Greulich, W.W., 1957, A comparison of the physical growth and development of American-born and
native Japanese children. American Journal of Physical Anthropology 15,489-515.
Habicht, J.-P., Martorell, R, Yarborough, C., Malina, RM., Klein, RE., 1974, Height and weight
standards for preschool children. How relevant are ethnic differences in growth potential? Lancet
i,611-615.
Himes, J.H., 1979, Secular changes in body proportions and composition. Monographs of the Society for
Child Development, 44, 28-58, 103-120.
Holden, C., 1997, Early puberty getting more common. Science 276, 537.
Kuczmarski, RJ., 1993, Trends in body composition for infants and children in the U.S. Critical
Reviews in Food Science and Nutrition 33 (4/5), 375-387.
Lindgren, G., 1976, Height, weight and menarche in Swedish urban school children in relation to socio-
economic and regional factors. Annals of Human Biology, 3, 501-528.
Malina RM., 1993, Ethnic variation in the prevalence of obesity in North American children and youth.
Critical Reviews in Food Science and Nutrition, 33, 389-396.
Micozzi, M., 1985, Nutrition, body size and breast cancer. Yearbook of Physical Anthropology, 28,
175-206.
Palmer, J.R., Rosenberg L., and Shapiro, S., 1990, Stature and the risk of myocardial infarction in
women. American Journal of Epidemiology, 132,27-32.
Price, A.B., Charles, M.A., Petit, D.J., Knowler, W.C., 1993, Obesity in Pima Indians: Large increases
among post-World War II birth cohorts. American Journal of Physical Anthropology, 92, 473-480.
Proos, L.A., 1993, Anthropometry in adolescence-secular trends, adoption ethnic and environmental
differences. Hormone Research 39 (suppl 3), 18-24.
Shetty, P.S., and James, W.P.T., 1993, Body Mass Index: A Measure of Chronic Energy Deficiency in
Adults. FAO, Rome.
Styne, D.M., and McHenry, H., 1993, The evolution of stature in humans. Hormone Research 39 (suppl
3),3-6.
Tanner, J.M., 1981, A History of the Study of Human Growth (Cambridge University Press,
Cambridge).
Tanner, J.M., 1986, Growth as a mirror of the conditions of society: Secular trends and class
distinctions. In Human Growth: A Multi disciplinary Review edited by A. Demirjian (Taylor &
Francis, London).
Udjus, L.G., 1964, Anthropometrical Changes in Norwegian Men in the 20th Century
(Universitetsforlaget, Olso).
SECULAR TRENDS IN GROWTH AND DEVELOPMENT 145
Waaler, H.T., 1984, Height, weight and mortality: the Norwegian experience. Acta Medica
Scandinavia, suppl. 679, 1-5.
WHO, 1995, Physical Status: The Use and Interpretation of Anthropometry. WHO Technical Report
Series 854, WHO, Geneva, p. 346.
Yip, R., Scanlon, K., Trowbridge, F., 1992, Improving growth status of Asian refugee children in the
United States. Journal of the American Medical Association, 267, 937-940.
CHAPTER 13
SECULAR TRENDS AND LONGTERM SERIAL

GROWTH STUDIES
A.F.ROCHE,B.TO~
Lifespan Health Research Center, Department of Community Health, Wright State

University School of Medicine, Kettering, Ohio USA
1. INTRODUCTION
We are delighted by the opportunity to contribute to this commemorative volume
that honours a leader responsible for major contributions to our communal
knowledge of child growth. Professor Das is one of relatively few investigators who
have initiated and conducted longterm serial studies of child growth; he may be the
only one among those few who made all the measurements himself. This
extraordinary data collection effort has already led to important papers dealing with
growth status, growth patterns, sibling comparisons and sexual dimorphism. His
data continue to be a great resource for future studies of secular trends and many other
aspects of child growth.
The term "secular trends" refers to differences among groups within a population
that are explained totally or largely by differences in birth dates. If the range of birth
dates is less than about ten years, any observed growth differences that are related to
birth dates are more likely to be due to sampling variations than to secular trends.
Exceptions to this statement may occur, however, when the environmental
influences on growth change very rapidly, as can occur during wars and famines.
Over the last two centuries, secular trends in child growth have generally taken the
form of increases; that is, children at a particular age are larger and more mature now
than in the past. Such secular trends are commonly referred to as "positive", but this
terminology is somewhat undesirable today because secular increases in weight,
skinfold thicknesses and body mass index, within populations that were already well-
nourished, have negative effects on health. Furthermore, in such populations, secular
increases in stature may not be associated with health benefits. Secular increases of
considerable magnitude have occurred in Europe and the United States, but the
occurrence and the size of secular trends are uncertain for many countries due to the
lack of national surveys and sampling differences among studies. Secular increases
are particularly well documented for the Czech Republic, Japan, the Netherlands, and
the United States where sampling and anthropometric methods have remained fixed
among repeated national surveys (Fredriks et al. 2000, Freedman et al. 2000, Kimura
1984, Lhotska et al. 1993, Troiano et al. 1995).
147
148 CHAPTER 13
Secular increases in size at a given age exceed those in skeletal maturity, as

shown by increases in adult stature. These increases in rates of skeletal maturation
are associated with decreases in the ages at which adult statures are reached. More
direct evidence that secular trends in skeletal maturation rates have been small come
from comparisons among large data sets from the United States and from inter-
generational comparisons within the Fels Longitudinal Study (Roche et al. 1975,
1978).
Secular trends have varied in size among countries and ethnic groups and they
have not been universal (Bodzsar and Susanne 1998, Roche 1979). Consequently,
during longterm serial growth studies and in the interpretation of data from such
studies, secular trends must be examined in the study groups, if the design allows
this, and in the population to which the study findings will be applied. Secular
trends are almost entirely due to environmental changes. Changes in the genetic
structure of populations over recent decades are unlikely to have had major influences
on child growth. The major exception to this is changes in a population's gene pool
due to migration. The effects of migration into or from the study groups can lead to
misleading conclusions about secular trends at the population level. Secular changes
in the target groups to which the data will be applied can reduce the utility of the
findings. Migration effects are particularly important if they differentially affect the
study groups and the target groups.
2. THE DESIGN OF SERIAL STUDIES

The influence of secular trends on longterm growth studies is considered in relation
to various study designs. In this paper, the emphasis is on longterm studies which
are defined arbitrarily as those that extend for at least ten years. Most such studies
have been conducted in developed countries, but there are notable exceptions
(Billewicz and McGregor 1982, Hauspie et al. 1982, Prentice et al. 1987, Rowland
et al. 1981). The concentration of these studies in developed countries is largely due
to these studies being expensive, difficult to execute, and needing complex statistical
analyses (Roche et al. 1999).
Some designs for serial growth studies include many examinations during
childhood that are continued, at longer intervals, during adulthood. These studies are
very difficult to maintain, but are particularly justifiable when health-related
variables are included. They allow descriptions of growth patterns and the
establishment of relationships between parameters of functions fitted to serial data
for individual children, or defined points on growth curves, to values in adulthood for
size or other variables (Guo et al. 1994, 1997, Kouchi et al. 1985a,b, Nieto et al.
1992, Rolland-Cachera et al. 1989).
Other studies have applied follow-up designs that relate status data for size at an
age to health in adulthood. Such analyses by Barker et al. (1989, 1993) have shown
that birth weight has some relationship to blood pressure, insulin resistance and the
probability of non-insulin dependent diabetes mellitus in adulthood. Must et al.
(1992) selected two groups of study participants: those with small values for body
mass index at two or more annual examinations during adolescence and those with
large body mass index values. The morbidity and mortality of these groups were
compared at a mean age of 73 years. Because the Barker and Must studies used data
from individuals born around 1920, and followed-up more than 50 years later, secular
SECULAR TRENDS AND GROWTH STUDIES 149
trends may have influenced the applicability of the results. Since 1920, prenatal and
neonatal care have improved in developed countries, and diseases during infancy and
childhood have become less common, but weight and body mass index have
increased over the last two decades, especially in adolescence and adulthood, despite
greater awareness of the health risks associated with overweight and obesity. Infants
born now with small birth weights, and adolescents with large weights or large body
mass index values, may not be subject to the health consequences indicated by these
retrospective data. There have been reductions in mortality rates for infants with
small birth weights and for adults with cardiovascular diseases.
Most longterm serial studies enrolled a single cohort. Well-known examples are
the Berkeley and Harvard studies in the United States (Jones et at. 1971, Stuart
1939). The participants in the Berkeley study were born in 1928 or 1929, and those
in the Harvard study were born from 1930 to 1939. Consequently, their growth may
have been affected by the Great Depression. Other notable examples are the Belgian,
English, French, Swedish and Swiss studies that began in 1954 and were co-
ordinated by the Centre International de I'Enfance (1980). Due to a lack of random
sampling, these studies cannot provide satisfactory cross-sectional reference data for
representative groups. Despite sampling limitations, cross-sectional reference data
have been developed from some of them (Karlberg et at. 1976, Prader et at. 1989,
Sempe et at. 1979). The utility of these reference data has been reduced, however, by
secular trends in the populations to which they were meant to be applied (Demoulin
1998, Lindgren 1998, Vercauteren et al. 1984). This statement should not be
interpreted as a criticism of these studies. They were designed to investigate patterns
of change and of relationships among variables; not to provide cross-sectional
reference data for representative populations. Still, they are useful and, in some
cases, unique sources of data.
A few longterm serial growth studies have enrolled participants over a large
number of years. The best, and perhaps only, examples are the United States studies
at the Child Research Council in Denver, Colorado, and the Fels Longitudinal Study
in Dayton, Ohio (McCammon 1970, Roche 1992). The Denver study enrolled 334
participants who were born from 1915 to 1966 and examined until 1968. In the Fels
Longitudinal Study, about 15 infants were enrolled at birth each year from 1929 to
the present except from 1975 through 1982 when financial constraints delayed
enrolments until the children were aged one to seven years. The Figure shows the
design of the Fels Longitudinal Study. The oblique lines indicating the nature of the
serial data for each annual group illustrate the opportunities for analysing age-to-age
correlations and growth patterns using these data. Both the Denver study and Fels
Longitudinal Study have been sources of reference data for growth status, despite the
admonitions stated in this paper, in part because they were considered better than
other data available at the time (Hamill et at. 1977, McCammon 1970).
The applicability of the data from the Denver study and the Fels Longitudinal
Study could be reduced by secular trends in the general population and/or within the
studies. Secular trends in the general population would reduce the applicability of
reference data for growth status, and increments or patterns of change, especially for
groups enrolled many years ago. There is a lack of information about secular trends
in the entire United States population prior to 1963. Since then, secular trends for
size at an age have been small except for weight and body mass index which have
increased markedly since 1980 for children aged six years and older (Troiano et al.
150 CHAPTER 13
1995). The absence of secular trends in stature, at least partly, reflects migration
from countries where statures are smaller.
70
60
50
~ 40
~
~
<: 30
20
10
o ~~~~~h~1G~~~~~~~~~~~~~~44~44~~4~
1929 1939 1949 1959 1969 1979 1989 1999

Year of birth
Figure 1. The design afthe Fels Longitudinal Study
Secular trends within the Fels Longitudinal Study may need to be considered
before data from the many annual groups are pooled for analyses. These trends have
been estimated by regressions on year of birth and (year of birth)2 and by
comparisons among groups defined by decade of birth. In the Fels Longitudinal
Study data, recumbent length and stature at an age became larger until the early
1950s and then became smaller, but these changes were slight and not significant
(Byard and Roche 1984). Similarly, Kouchi et al. (1985a, b) showed that parameters
of functions fitted to serial data for weight and recumbent length during infancy
changed only slightly among participants grouped by decade of year of birth. Bock
and Sykes (1989) confirmed these findings, but demonstrated inter-generational
increases in stature at an age in like-sex comparisons. Apparently, these intra-
SECULAR TRENDS AND GROwrn STUDIES 151
familial increases have been balanced by unintentional or coincidental enrolment of

somewhat shorter families during more recent years. Within the Fels Longitudinal
Study, there have been small secular trends in weight with girls becoming heavier,
and boys becoming lighter, until the mid-1960s, but weight and BMI for each sex
have increased since 1980 (Garn and French, unpublished data, Guo et aI.,
unpublished data, Siervogel et aI., unpublished data). The secular trends in head
circumference have been small (Roche et al. 1986).
The preceding text illustrates possible effects of secular trends in status, rates, and
patterns of growth on the applicability of findings from longterm serial growth
studies. As noted earlier, such studies may be inappropriate sources of reference data
for growth status. Ideally, such reference data should come from cross-sectional
studies of representative samples. Reference data for incremental growth can be
obtained from short-term serial studies. If schools are visited at intervals of 6 or 12
months, data can be obtained that will not be influenced by secular trends. Secular
trends within the population to which these data are applied, however, may make it
necessary to repeat the study after one or more decades. If there have been secular
changes in status values for the population to which the reference data for increments
will be applied, there must also be secular changes in increments. The possible
effects of secular trends on rates and patterns of growth should be evaluated, but few
reported data address this topic, and they necessarily come from serial studies. The
literature indicates that secular trends have not occurred in age-to-age correlations
(tracking) for weight, recumbent length and stature (Thomson 1955, Wright et at.
1994), or in sibling correlations for size at an age (Susanne 1980). Methods
developed decades ago in longterm serial studies for the prediction of adult stature
from childhood variables, and functions that describe serial data, still work
reasonably well (Bayley and Pinneau 1952, Jenss and Bailey 1937, Preece and Baines
1978). The continued utility of these methods is noteworthy; the description of
longterm changes and the prediction of future values lie at the heart of longterm
growth studies.
It is necessary to document the extent of secular trends within a longterm serial
study. Furthermore, if the analyses provide data for use by clinicians or public health
workers, it is essential to determine whether or not the study population matches the
current target population for status values. When status values differ between the
study population and the target population, interesting and useful information can
still be provided. It may be possible to select a subset of the study population that
has distributions of status values that are similar to those in the target population
and use those subsets for analyses of growth rates and patterns. Sample size
considerations may make it difficult to implement this approach, but typically there
will be at least 80% overlap between the study and the target groups.
Another approach is to design a longterm serial study so that secular trends
within the study will be reduced or absent. This is not easy, and the design will have
to be implemented before it is known if secular trends would be present in a more
traditionallongterm study. An excellent example of such an alternative design is the
Nijmegen (Netherlands) Growth Study that was conducted from 1970 through 1975
(Prahl-Anderson 1979). This study had a mixed-longitudinal design with six cohorts
that differed in birth dates and, consequently, differed in ages at enrolment and at
examinations. The age ranges at the examinations, which were made during a six-
year period, overlapped among the cohorts and provided data from 4 to 14 years.
152 CHAPTER 13
Prahl-Anderson and her colleagues believed that this design reduced possible
confounding due to examination effects and secular trends.
Examination effects, which are sometimes called testing effects or learning
effects, are changes in the recorded data due to the repetition of examinations.
Examination effects occur in behavioural variables and auditory thresholds and
probably explain why blood pressures are lower in serial studies than in cross-
sectional studies (Guo et al. 1998, Kallman and Jarvik 1959, Roche et al. 1983).
Examination effects are of limited interest in the present context because they do not
occur in anthropometric data nor, with the exception of blood pressure, in risk
factors for later disease that may be related to size during childhood. In theory, the
Nijmegen design could help address possible secular differences among cohorts, but
the cohorts in the Nijmegen design were born no more than six years apart.
C;onsequently, any inter-cohort differences are more likely to be due to sampling
variations than to secular trends. In fact, inter-cohort differences for anthropometric
variables were present for head circumference only; this increased significantly with
birth date in boys and in girls. It is also of interest that the anthropometric status
values from the Nijmegen Study exceeded those from a concurrent national survey in
the Netherlands (de Wijn 1976). This reinforces the view that status values from
serial studies should not be relied upon as reference data.
A design similar to that of the Nijmegen Study is logistically difficult, and
substantial financial support is needed from the commencement. Most longterm
serial growth studies have begun with modest support that increased gradually
(Roche et al. 1999). The Fels Longitudinal Study began simply and inexpensively
with a group of six infants, but became more complex and more expensive due to
enlargement of the sample, increases in the number of variables recorded, and the
need for more complex analyses as serial data accumulated.
An alternative design, sometimes called "shingling" or "laddering", is an
extension of the Nijmegen plan. In concept, if one were to enrol participants at
birth, 4, 8, 12 and 16 years and measure each participant serially for four years, one
would obtain a complete description of growth from birth to 20 years. This goal
might be met if sampling were consistent across groups, if the groups did not differ
in enrolment response rates and missed examinations, and if, on final analysis, the
pieces of the total curve join without disjunctions due to differences in levels or rates
of growth. A major disadvantage of this approach is that it requires large
expenditures from the beginning, and it is difficult to obtain such financial support.
There are also limitations to what can be learned. The combination of data from
cohorts differing in birth dates can provide reference percentiles for increments, but
they do not allow analyses of tracking or risk of a later outcome over periods longer
than those for which each cohort was measured. Clearly, decisions about the design
of a longterm serial study must be linked to the central aims of the study.
3. PLANNING A NEW STUDY

The preceding text might lead a reader to conclude that concern about the effects
of possible secular trends on the findings from serial growth studies are restricted to
studies that began decades ago. This is not so. Secular trends can affect current
studies and their effects should be considered during the design of future studies. A
new study of child growth in Nepal exemplifies the ways in which possible secular
trends can both strengthen and weaken a new serial investigation and the need to take
these into account during the design phase. While the primary factors responsible for
secular trends are generally understood (e.g., improvements in nutrition and control
of disease), secular trends remain dynamic processes that vary across populations.
Conducting serial studies of growth in developing countries offers an opportunity to
identify specific agents of secular change and to assess the magnitude of their effects.
In so doing, the varied processes of secular change may be more clearly understood.
A pilot study of the growth and development of children from the Jirel ethnic
group in eastern Nepal has recently been initiated. The Jiri Growth Study (Towne et
al. 1999,2000) is conducted in collaboration with the Jiri Helminth Project, a study
of genetic factors predisposing individuals to helminthic infection (Williams-
Blangero et al. 1999). The goal of the Jiri Growth Study is to quantify genetic and
environmental factors influencing variation among children in their growth and
development. The Jirel population is ideal for such a study, because most of the
approximately 4,000 Jirels are members of one very large extended pedigree.
Chronic infection with parasitic disease has particularly negative effects on
children's health. Gastrointestinal diseases, especially diarrhoea, malabsorption of
nutrients, and anaemia, lead to deficits in growth and development. In some cases,
the immune system is compromised, opening the door to other infections. Overall,
at any point during childhood, approximately 50% of Jirel children are infected with
at least one species of helminth, either roundworm, hookworm, or whipworm. Pilot
data from a sample of 431 children aged 3 to 18 years showed that helminthic
infections have deleterious consequences for the growth of Jirel children. Boys and
girls infected with roundworms or hookworms were significantly shorter than non-
infected children, and they were smaller in virtually all body size measures.
Besides various diseases, there are other environmental factors detrimental to
growth and development that are commonly placed in sociocultural, socio-economic,
or socio-demographic categories. These categories may serve as useful proxies for
measures of nutrition and hygiene, which are difficult to measure directly. Lack of
sanitation and poor hygienic practices contribute to the spread of infectious disease,
particularly geohelminthic gastrointestinal infections. Sociocultural factors including
availability of potable water, indoor plumbing, use of latrines, and food handling
practices that pertain to the spread of such diseases. Sociocultural and socio-
economic surveys of households participating in the Jiri Helminth Project show that
household income is strongly associated with helminthic infections (Subedi et al.
2000, Williams-Blangero et al. 1998). Household income is associated with a cluster
of characteristics that are risk factors for transmission of helminths. Among these are
presence or absence of a household latrine, and whether or not each individual in the
household wears shoes. Other characteristics associated with household income are
number of rooms in the house, and the type of house construction. Thus, household
income appears to serve as a useful overall measure of household cleanliness and
sanitation, which are important in the transmission of helminthic infection to
children who stay in and around the house most of the day. For roundworm,
hookworm, and whipworm, analysed separately, there were highly significant non-
random associations of infection status with household income, such that infected
Jirel children came predominately from low income households, while uninfected
children came predominately from high income households. These findings were
154 CHAPTER 13
even more pronounced when infection status was defined as being infected with more
than one species of helminth.
Clearly the growth of Jirel children is negatively impacted by parasitic infections.
It is likewise similarly impacted by other factors associated with poor household
environment (e.g., caloric and, especially, protein deficiencies). Measures of growth
and development, however, have significant genetic determinants. Ideally, both
environmental exposure and familial relationship data are analysed simultaneously to
evaluate the relative contributions of genes and environment to growth and
development. To evaluate the roles of roundworm infection and household
environment on the stature of Jirel children, the heritability of stature was estimated
in genetic models with and without inclusion of roundworm infection status (non-
infected vs. infected) and/or household income (low vs. high) as covariates. The best
statistical genetic model of the stature of Jirel children contained significant covariate
effects of both roundworm infection status and household income on the stature of
boys, such that infected boys are shorter than non-infected boys, and boys from
higher income households are taller than boys from lower income households.
Interestingly, these covariate effects were not significant for girls. Nonetheless, the
h2 of stature with both sexes considered was very high at approximately 90%.
Hopefully, the Jirel population, like the rest of the developing world, will see
decreases in the prevalence of disease and increases in personal wealth over the next
few decades. Such changes will contribute to increasing numbers of children being
healthier and growing larger than at present. These secular trends will not complicate
the objectives of the Jiri Growth Study, rather, by its very structure, the Jiri Growth
Study will offer an opportunity to quantify specific aspects of the environment that
result in secular trends in growth.
4. CONCLUSION
The preceding comments might lead a reader to conclude that longterm serial growth
studies are too problematic to either initiate or continue. This would be a gross error!
The major purposes of such studies are to describe growth patterns, predict future
growth, analyse the relative importance of determinants of growth and development,
and evaluate the significance of changes during childhood and adulthood. Without
such past studies, we would lack accurate descriptions of infant growth, the
pubescent growth spurt, and changes in the body mass index, and we would not have
methods to predict adult stature and assess skeletal maturity, to mention just a few of
their products. Further, such studies are needed in different populations, both normal
and abnormal, and should include measures other than those that describe size. In
particular, data should be collected for studying determinants of growth (e.g., familial
and genetic data), and for factors that increase the risk of diseases in adulthood. There
are clear guidelines for the conduct of these studies, and the necessary statistical
methods are readily available. More, not fewer, longterm serial growth studies are
needed.
5. REFERENCES
Barker, D.J.P., Hales, e.N., Fall, C.H.D., Osmond, C., Phipps, K, and Clark, P.M.S., 1993, Type 2 (non-
insulin-dependent) diabetes mellitus, hypertension and hyperlipidaemia (syndrome X): relation to
reduced fetal growth. Diabetologia, 36, 62-67.
Barker, DJ.P., Winter, P.D., Osmond, C., Margetts, P., and Simmonds, SJ. 1989, Weight in infancy and
death from ischaemic heart disease. Lancet, 2, 577-580.
Bayley, N., and Pinneau, S.R, 1952. Tables for predicting adult height from skeletal age: revised for
use with the Greulich-Pyle hand standards. Journal of Pediatrics, 40, 423-441. (Erratum corrected
in Journal of Pediatrics, 41, 371.)
Billewicz, W.z., and McGregor, LA., 1982, A birth-to-maturity longitudinal study of heights and
weights in two West Africian (Gambian) villages, 1951-1975. Annals of Human Biology, 9, 309-
320.
Bock, RD., and Sykes, RC., 1989, Evidence for continuing secular increase in height within families in
the United States. American Journal of Human Biology, I, 143-148.
Bodzsar, E.B., and Susanne, e., 1998. Secular Growth Changes In Europe, Budapest, Eotvos University
Press.
Byard, PJ., and Roche, A.F., 1984, Secular trend for recumbent length and stature in the Fels
Longitudinal Growth Study. In Human Growth and Development, edited by 1. Borms, R Hauspie,
A. Sand, C. Susanne, and M. Hebbelinck (New York: Plenum Press), 209-214.
Centre International de I'Enfance, 1980, Growth and Development of the Child. Courrier, 30, 1-260.
Demoulin, F., 1998. Secular trend in France. In Secular Growth Changes in Europe, Budapest, edited by
E. Bodzsar and e. Susanne (Budapest: EOtvos University Press), 109-134.
Fredriks, A.M., van Buuren, S., Burgmeijer, RJ.F., Meulmeester J.F., Beuker R.I., Brugman E., Roede
MJ., Verloove-Vanhorick S.P., and Wit, J.-M., 2000, Continuing positive secular growth change in
the Netherlands 1955-1997. Pediatric Research, 47,316-323.
Freedman, D.S., Khan, L.K., Serdula, M.K., Srinkvasan, S.R, and Berenson, G.S., 2000, Secular trends
in height among children during 2 decades: The Bogalusa Heart Study. Archives of Pediatric and
Adolescent Medicine, 154, 155-161.
Guo, S.S., Chi, E., Wisemandle, W., Chumlea, W.C., Roche, A.F., and Siervogel, RM., 1998, Serial
changes in blood pressure from childhood into young adulthood for females in relation to body mass
index and maturational age. American Journal of Human Biology, 10, 589-598.
Guo, S.S., Chumlea, W.e., Roche, A.F., Gardner, J.D., and Siervogel, R.M., 1994, The predictive value
of childhood body mass index values for overweight at age 35 y. American Journal of Clinical
Nutrition, 59, 810-819.
Guo, S.S., Chumlea, W.e., Roche, A.F., and Siervogel, RM., 1997, Age- and maturity- related changes
in body composition during adolescence into adulthood: The Fels Longitudinal Study. International
Journal of Obesity, 21,1167-1175.
Hamill, P.V.V., Drizd, T.A., Johnson, C.L., Reed, RB., Roche, A.F., 1977, NCHS Growth Curves for
Children. Birth-18 years, United States. Vital and Health Statistics, National Center For Health
Statistics, Series II, No. 165, pp. 74,1977.
Hauspie, RC., Das, S.R, Preece, M.A., Tanner, J.M., 1982, Degree of resemblance of the pattern of
growth among sibs in families of West Bengal (India). Annals of Human Biology, 9, 171-174.
Jenss, RM., and Bayley, N., 1937, A mathematical method for studying the growth of a child. Human
Biology, 9, 556-563.
Jones, M.C., Bayley, N., MacFarlane, J.W., Honzik, M.P., 1971. The Course Of Human Development
(Waltham, MA: Xerox College Publishing).
Kallman, FJ., and Jarvik, L.F., 1959, Individual differences in constitution and genetic background. In
Handbook of Aging and the Individual, edited by J.E. Birren (Chicago: University Press), 216-263.
Karlberg, P., Taranger, J., Engstrom, I., Karlberg, J., Landstrom, T., Lichtenstein, H., Lindstrom, B.,
and Svennberg-Redegen, I., 1976, Physical growth from birth to 16 years and longitudinal outcome
of the study during the same age period. The Somatic Development of Children in a Swedish Urban
Community. A Prospective Longitudinal Study, Acta Paediatrica Scandinavica, Supp1258, 7-70.
Kimura, K., 1984, Studies of growth and development in Japan. Yearbook of Physical Anthropology,
27, 179-214.
Kouchi, M., Mukherjee, D., and Roche, A.F., 1985a, Curve fitting for growth in weight during infancy
with relationships to adult status, and familial associations of the estimated parameters. Human
Biology, 57,245-265.
Kouchi, M., Roche, A.F., and MukheIjee, D., 1985b, Growth in recumbent length during infancy with
relationships to adult status and familial associations of the estimated parameters. Human Biology,
57,449-472.
156 CHAPTER 13
Lhotska, L., Blaha, P., Vignerova, Roth, Z., and Prokopec, M.V.V., 1993, Celostatnf Antropologicky
Vyzkum Deti A Mladeze 1991 (Ceske Zeme): Antropometricke charakteristiky (Praha: National
Institute of Public Health).
Lindgren, G., 1998, Secular growth changes in Sweden. In Secular Growth Changes in Europe, edited
by E. Bodzsar and C. Susanne (Budapest: E5Mls University Press), 319-333.
McCammon, R.W., 1970, Human Growth and Development (Springfield, IL: Charles C Thomas).
Must, A., Jacques, P.F., Dallal, G.E., Bajema, e.1., and Dietz, W.H., 1992, Long-term morbidity and
mortality of overweight adolescents. A follow-up of the Harvard Growth Study of 1922 to 1935,
New England Journal of Medicine, 327, 1350-1355.
Nieto, F.J., Szklo, M., and Comstock, G.W., 1992, Childhood weight and growth rate as predictors of
adult mortality. American Journal of Epidemiology, 136,201-213.
Prader, A., Largo, R.H., Molinari, L., and Issler, C., 1989, Physical growth of Swiss children from birth
to 20 years of age. First Zurich Longitudinal Study of Growth and Development. Helvetica
Paediatrica Acta, S52, 1-125
Prahl-Anderson, B., 1979, The conduct of the Nymegen Growth Study. In A Mixed-Longitudinal
Interdisciplinary Study of Growth and Development edited by B. Prahl-Anderson, e.J. Kowalski,
and P. Heydendael (New York: Academic Press), 61-70.
Preece, M.A., and Baines, MJ., 1978, A new family of mathematical models describing the human
Prentice, A.M., Cole, T.J., and Foord, F.A., Lamb, W.H., and Whitehead, R.G., 1987, Increased
birthweight after prenatal dietary supplementation of rural African women. American Journal of
Clinical Nutrition, 46, 912-925.
Roche, A.F., 1979, Secular Trends in Human Growth, Maturation and Development. Monographs of the
Society for Research in Child Development (edited), 44, 1-120.
Roche, A.F., 1992, Growth, Maturation and Body Composition: The Fels Longitudinal Study 1929-1991
(Cambridge, UK: Cambridge University Press), 345.
Roche, A.F., Guo, S.S., and Towne, B., 1999, The establishment and maintenance of longitudinal growth
studies: Experiences from Melbourne and Fels. In Human Growth in Context, edited by F. Johnston,
B. Zemel, and P.B. Eveleth. (London, UK: Smith-Gordon), 101-120.
Roche, A.F., Mukherjee, D., Chumlea, W.e., and Champney, T.F., 1983, Examination effects in
audiometric testing of children. Scandinavian Audiology, 12,251-256.
Roche, A.F., Mukherjee, D., Guo, S., and Moore, W.M., 1986. Head circumference reference data:
birth to 18 years. Pediatrics, 79, 706-712.
Roche, A.F., Roberts, J., and Hamill, P.V.V., 1978, Skeletal maturity of youths 12-17 years: Racial,
geographic area, and socioeconomic differentials, United States. Vital and Health Statistics,
National Center for Health Statistics, Series 11, No. 167,98.
Roche, A.F., Wainer, H., and Thissen, D., 1975, Skeletal Maturity: The knee joint as a biological
indicator (Plenum, New York), 374.
Rolland-Cachera, M.-F., Bellisle, F., and Sempe, M., 1989, The prediction in boys and girls of the
weightlheight2 index and various skinfold measurements in adults: A two-decade follow-up study.
International Journal of Obesity, 13, 305-311.
Rowland, M.G.M., Paul, A., Prentice, A.M., MUlier, E., Hutton, M., Barrell, R.A.E., and Whitehead,
R.G., 1981, Seasonality and the growth of infants in a Gambian village. In Seasonal Dimensions to
Rural Poverty, edited by R. Chambers, R. Longhurst, and A. Pacey (London: Francis Pinter), 164-
175.
Sempe, M., Pedron, G., and Roy-Pernot, M.P., 1979, Auxologie methode et sequences (Paris:
Laboratorie TMraplix).
Siervogel, R.M., Roche, A.F., and Guo, S., Mukherjee, D., and Chumlea, W.C., 1991, Patterns of
change in weightlstature 2 from 2 to 18 years: Findings from long-term serial data for children in the
Fels Longitudinal Growth Study. International Journal of Obesity, 15,479-485.
Stuart, H.C., 1939, Studies from the Center for Research in Child Development, School of Public Health,
Harvard University. I. The Center, the group under observation, sources of information, and studies
in progress. Monographs of the Society for Research in Child Development, 4, No.1.
Subedi, J., Subedi, S., Sidky, H., Singh, R., Blangero, J., and Williams-Blangero, S., 2000, Health and
health care in liri. Contributions to Nepalese Studies. April-May, in press.
Susanne, C., 1980, Developmental genetics of man. In Human Physical Growth and Maturation:
Methodologies and Factors, edited by F.E. Johnston, A.F. Roche, and C. Susanne (New York:
Plenum),221-242.
Thomson, J., 1955, Observations on weight gain in infants. Archives of Diseases in Childhood, 30, 322-
327
Towne, B., Blangero, J., Williams-Blangero, S., 1999, Growth and body composition of Jirel children.
American Journal of Human Biology, 11,131.
Towne, B., Blangero, J., Williams-Blangero, S., 2000, Genetic and environmental Influences on the
growth status of Nepali children. American Journal of Human Biology, 12,278.
SECULAR TRENDS AND GROWTII STUDIES 157
Troiano, R.P., Flegel, K.M., Kuczmarski, R.I., Campbell, S.M., and Johnson, C.L., 1995, Prevalence and
trends in overweight for children and adolescents: The National Health and Nutrition Examination
Surveys, 1963-1991. Archives of Pediatrics and Adolescent Medicine, 149, 1085-1091.
Vercauteren, M., Susanne, c., and Orban, R., 1984, Secular evolution in Brussels between 1960 and
1980. In Human Growth and Development, edited by J. Borms, R. Hauspie, A. Sand, C. Susanne,
and M. Hebbelinck (New York: Plenum Press), 215-223.
Wijn, J.F., de, 1976, Zesde orientering omtrent de voedingstoestand van achtjarige schoolkinderen in
Nederland (1973174). Tijdschrift Kindergeneesk, 44, 18-32.
Williams-Blangero, S., Subedi, J., Upadhayay, R.P., Maural, D.R., Khanda, K., lirel, S., Robinson, E.S.,
and Blangero, J., 1998, Attitude towards helminthic infection in the Jirel population of eastern
Nepal. Social Science and Medicine, 47, 371-379.
Williams-Blangero, S., Subedi, 1., Upadhayay, R.P., Maural, D., Rai, D.R., Jha, B., Robinson, E.S., and
Blangero, J., 1999, Genetic analysis of susceptibility to infection with Ascaris lumbricoides.
American Journal of Tropical Medicine and Hygiene, 60, 921-926.
Wright, C.M., Matthews, J.N.S., Waterston, A., and Aynsley-Green, A., 1994, What is a normal rate of
weight gain in infancy? Acta Paediatrica, 83, 351-356.
CHAPTER 14
SECULAR TRENDS IN BODY HEIGHT - INDICATOR

OF GENERAL IMPROVEMENT IN LIVING
CONDITIONS OR OF A CHANGE IN SPECIFIC
FACTORS?
M.HENNEBERG
Wood Jones Chair of Anthropological and Comparative Anatomy, University of Adelaide

Medical School, Adelaide, Australia
1. INTRODUCTION
Body size of an animal is a result of its adaptations. These consist of long-term

genetic effects and short-term adjustments during individual ontogeny. To what
extent is the variability of body size in a human population dependent on the genetic
background, and to what on direct ontogenetic adjustments is indicated, although
imperfectly, by the heritability coefficients (Falconer 1980). These are invariably
high for body height (Cavalli-Sforza and Bodmer 1971, Mueller 1976, Roberts et ai.
1978, Susanne 1971). Typically more than 90% of variation in height is due to the
variation of the underlying genetic material. The rest is due to direct environmental
influences upon individual bodies and to measurement errors including diurnal, or
seasonal fluctuations.
During the hominid evolution body size generally increased until the Late
Pleistocene (Frayer 1984, Styne and McHenry 1993, Mathers and Henneberg 1995,
De Miguel and Henneberg 1999). The increase from Pliocene to Late Pleistocene
was about 450 mm (35%) in height and some 30 kg (nearly 100%) in weight.
Reasons are not quite clear, but most probably were related to changing ecological
niche of hominids that provided access to richer foods and favoured stronger bodies
for big-game hunting.
During the Holocene, human body size decreased on all continents (Frayer, 1984,
Jacobs 1985, Henneberg 1988, Brown 1992, Henneberg and Steyn 1993). Sex-
averaged body height declined from about 1750 mm 30 Ka ago to some 1570 mm in
the middle ages (about 10%), while weight declined from nearly 80 kg to 55 kg
(30%) during the same period (Mathers and Henneberg 1995). This decrease was once
again related to changing subsistence patterns as big game became scarcer and many
populations switched to agriculture in which food supply was limited while great
body strength was not necessarily at premium any more. These changes occurred to a
varying degree in various human groups being superimposed on body size
159
160 CHAPTER 14
adaptations to natural environmental factors, especially those related to

thermoregulation. All these millennial changes seem to be of the genetic adaptation
kind that could hardly occur within several decades. Were human stature indicative
solely of the living conditions in terms of the socio-economic stratification one
would have to be forced to conclude that our Upper Pleistocene ancestors, with their
stature matching that in present-day First World countries, were as well-off as
modern Europeans (Styne and McHenry 1993). This can hardly be supported by other
indicators of biological status based on age-specific mortality that indicate much
poorer living conditions (Henneberg 1976, 1992).
In the recent hundred years body height increases were reported for many human
populations (Roche 1979, Bogin 1988, Malina 1990, Prokopec 1984, Sobral 1990,
Jamison 1990, Rao et al. 1993, Spurgeon et al. 1994, Gerver et al. 1994, Gutierrez-
Muniz et al. 1994, Roberts 1994, Huang and Malina 1995, Weber et al. 1995,
Hauspie et at. 1996, 1997, Eiben 1996, Loesch et at. 2000). The increases were in
most cases the greatest when statures of older children and adolescents were compared
by age group. This is the effect of the acceleration of physical growth and maturation
that does not necessarily translate into same magnitude increase in adult stature.
It is significant that a number of authors reported finding no increase in adult
stature in certain populations (Tobias 1975, 1985, 1990, Henneberg and van den
Berg 1990, Dettwyler 1992, Ulijaszek 1993, Leatherman et al. 1995, Prince 1995,
Louw and Henneberg 1997, Pretty et al. 1998). Since the positivist attitude to
science does not attach value to a finding of non-existence of a phenomenon, it can
be suspected that many more researchers found no body height increases in the recent
past but simply neglected to report them (publication bias). It seems therefore that
stature increases in the last century were not universal nor uniformly related to
improvements in socio-economic situation.
Short-term changes occurring between consecutive generations are called "secular
trends" from Latin saeculum for a generation. Since secular increases in body size,
where present, coincided with improvements in the living conditions that occurred
during the last century, this coincidence has been considered indicative of causal
relationship. Hence, secular increases were explained by improvements in living
conditions while non-existence of secular trend was, without further proof,
interpreted as an indication of the lack of improvement in socio-economic situation
(Tobias 1975, 1985, 1990, Oyedeji et al. 1996). This kind of argument has been
strengthened by observations that individuals belonging to higher socio-economic
status groups are taller than members of "lower classes" (Bielicki and WeI on 1982,
Bogin 1988, Rosenbaum et al. 1985). No quantitative comparison of the magnitude
of secular increases with the amount of statural difference between socio-economic
groups was made. The inference that socio-economic improvement is the cause of
secular increases in stature has been based on two categorical statements: (1) higher
socio-economic status people are on the average taller that people of poorer socio-
economic situation and (2) stature increased in some populations experiencing socio-
economic improvements. The lack of secular increase in some populations
experiencing socio-economic improvements was ignored together with the presence
of secular increases where no socio-economic improvement was apparent.
It is argued here that the overall socio-economic improvement per se is not the
cause of secular trends in stature and that specific causes of this phenomenon must
be sought. Once these are found, changes in stature may become sensitive indicators
SECULAR TRENDS INDICATE SPECIFIC FACTORS 161
of changing specific elements of the environment while health policies may ensure
optimal conditions for child growth by setting standards for those specific elements
in addition to the obviously needed general socio-economic improvement.
2. SOCIO-ECONOMIC GROUPS WITHIN AND BETWEEN COUNTRIES

Different socio-economic groups show same trends within a country while same
socio-economic groups have different trends in different countries
Although positivist attitudes argue against publication of data indicating lack of
secular change, the specific situation of South Africa made it significant to report the
lack of secular increase in the stature of people oppressed by apartheid. Data
presented by Tobias (1975,1985) indicated that the stature of Black South Africans
did not change during the last century. This has been interpreted by the author as an
indicator of the lack of socio-economic progress among Blacks caused by restrictive
and biased policies of apartheid. The same policies gave Whites privileged access to
jobs, health services and innovative technologies. Logically it follows that White
stature should exhibit a strong secular trend, paralleling, if not exceeding, that
observed in the First World countries. Our initial observations based on a variety of
data from 19-20th century skeletal samples and surveys of students and military
servicemen did not support this prediction (Henneberg and van den Berg 1990). For
the period 1880-1980 slopes of regressions of stature on birthdate did not differ
significantly between Black males (2.4 mmldecade) Khoisan males (4.8 mmldecade)
White males (5.9 mmldecade) and White females (4.1 mmldecade) (Figure 1).
Figure 1. Secular trends in body height of White and Black South African males and
Khoisan males. Data from Tobias (1975, 1985), Henneberg and van den Berg (1990) and
Louw and Henneberg (1997). Note that slopes of the regression lines do not differ
significantly
162 CHAPTER 14
All slopes were, however, significantly different from zero indicating positive,
but very slow increase of stature. During a comparable period, stature of Dutch
people, from whom the majority of South African Whites are descended, increased at
a rate of about 15 mm1decade (van Wieringen 1986).
It is interesting to note that adult South Australian Aborigines recently surveyed
by ourselves (Pretty et at. 1998) display lack of strong secular increase of stature
similar to oppressed South African Blacks (Figure 2). This occurs despite
documented substantial welfare provided to Aborigines by the government during the
second half of the 20th century. Among some Australian children of European
ancestry a secular increase comparable to those reported for Europe and the US has
been found (Loesch et at. 2000), but this acceleration of physical development did
not translate into substantial increases of adult stature. Data on adult statures of
Australians, though, are scarce. Stature of 5000 adult (18 - 65 years) females
surveyed in 1926 was 1611 mm (Lancaster 1957) while the stature of another 4327
women of comparable age surveyed in 1995 by the Australian Bureau of Statistics
(1995) was 1624 mm. This indicates a rate of about 2 mm/decade. A study of
Australian military men indicated the secular increase of stature at a rate of about 4
mm/decade (Soar 1999). These rates are similar to the weak increase found among
South Africans and to the increases among Aboriginal Australians. The socio-
economic history of Australia is similar to that of Europe and the United States, but
the increase in adult stature seems to be less pronounced.
1900
e-- Aborigines
1850 o Blacks o
o
1800 o
~
o
1750 o o
E
E o~
1700 o -u co
19
....
A
..<::
01)
0 &ctJ 0 i oD
'Q) o 0 I:lll DCO@
::r:: 1650 o 0 0
00
1600
1550
1500
1890 1910 1930 1950 1970 1990
Year
Figure 2. Comparison of secular trends in body height of South African Blacks (sample
means like in Fig. 1) and South Australian Aborigines (individual heights by birthdate
adjusted for loss of stature with age, own anthropometric observations, Pretty et at. 1998)
In contrast to South Africa, in Poland policies of communist governments since

the end of the World War II were aimed at obliteration of socio-economic differences,
while the economy of the country, ruined during the War developed poorly under
Soviet-style policies. Yet the secular trend in stature was strong, exceeding 10
mm/decade (Piontek 1971, Wolanski 1978, Bielicki and Welon 1982). Traditionally,
Polish medical students came from the upper socio-economic strata of the society. In
a sample of 300 students observed by ourselves in 1983 (Henneberg et al. 1985)
64% of parents had full university education. Similar situation obtained in the 1930
study of 153 medical students by Wrzosek (1931). This contrasts with about 5% of
university-educated individuals in the country. Post-war policies clearly favoured the
upliftment of people of lower socio-economic backgrounds at the expense of
economic well-being of the upper and middle classes. Despite this alteration of the
rate of change of socio-economic conditions for various socio-economic groups, the
rate of secular trend among the medical students (12 mm/decade) was the same as
among conscripts representing the total population (12 mm/decade) (Figure 3).
Portugal has been considered one of the least socio-economically developed
countries of Western Europe. Sobral (1990) presented data on body height changes in
four provinces of Portugal (1930-1980). For each province data were analysed
separately for rural and for urban conscripts. In all provinces, and at all dates
examined, rural people were shorter than their urban counterparts, yet in all cases
trends had similar rates of about 10 mm/decade (Figure 4). Parallelism, once again,
though socio-economic conditions and their temporal changes different from those of
Poland.
1900
1850 I a-- Polish conscripts

0 Polish high SES
1800
S 1750
S
.._r 1700
,..t:i
b/)
.(j)
::r:: 1650
1600
1550
1500
1910 1930 1950 1970 1990
Year
Figure 3. Secular trends of body height in Polish males. High socia-economic status is
represented by medical students, data from Henneberg and van den Berg (1990).
164 CHAPTER 14
1700
e-- Setubal, urban
1690 .• _.& .. - Setubal, rural
e-- Beja, urban
1680
- - - - E}- - - - Beja, rural
§ 1670 --+-- Lisbon, urban

. . . . +- - - Lisbon, rural
.fo
·0
1660 * - - Evora, urban
::r:
--- -)(- -- - Evora, rural
1650 x
0
1640 X-
v ~ ~
0
+' :,- -
1630 .-X
B
)(-
1620
1930 1940 1950 1960 1970 1980
Year
Figure 4. Body height of Portuguese conscripts by date of examination, province and

rural/urban residence. Datafrom Sobral (1990).
In a series of data on Japanese university students and conscripts covering the

period 1900-1963 Kimura (1967) found the same situation as that described for
Poland. Students were about 40 mm taller than conscripts, but both groups showed
parallel increase in stature by about 10 mmldecade.
There are other examples of parallel secular trends detectable in data from the US,
Holland and former Czechoslovakia, all of which were discussed by ourselves earlier
(Henneberg and van den Berg 1990).
There are some other instances in which secular trends of stature were found not
to follow socio-economic change. For example, drastic worsening of socio-economic
conditions of Sioux Indians following the American invasion of their country had
not been reflected in a negative secular trend (Prince 1995). Documented socio-
economic improvement in the Peruvian District of Nunoa has not been reflected by a
positive secular trend in adult stature (Leatherman et at. 1995).
3. DISCUSSION
The positive secular trend of stature was still going strong in the 1970's and 1980's
in some economically advanced countries, while it has apparently halted in others
(Roche 1979, Malina 1990). Were it true that secular trend is a result of consecutive
generations reaching increasingly greater proportions of their ideal "genetic
potential", privileged groups in various societies would have to halt their trends
earlier than underprivileged ones, who should continue until they "caught up" with
their wealthier compatriots. This certainly was not the case neither in South Africa
nor in Poland. Moreover, the greatest differences in stature between the poorest and
the richest segments of same national populations do not exceed one standard
deviation, i.e. about 60 - 70 mm (Bielicki and Welon 1982, Rosenbaum et al. 1985,
Henneberg and Louw, 1998), while positive secular trends, where they occurred
strongly, shifted stature by as much as 120 - 200 mm (van Wieringen 1986,
Spurgeon et al. 1994). This, coupled with high heritability of the variation in
stature, argues strongly against secular trends being a simple ecosensitive response
to improving general living conditions. In addition, the magnitude of secular trends
in populations experiencing similar changes in socio-economic status varies
considerably from 2 - 6 mm per decade among Australians and South African Whites
to 10 - 15 mm among Europeans.
Secular trends must be caused by specific factors varying from population to
population rather than by the general improvement in living conditions. General
socio-economic improvement may play some role, but it can hardly cause changes
exceeding 70 mm. We cannot yet name specific factors causing greater increases in
height, but certain criteria that they must meet can be specified. This may help to
direct our search for specific factors. Since statural increases in some instances exceed
the range of ecosensitivity, the causative factor must either affect genes determining
stature, or act as a substitute for one of the relevant gene products. If the second is
true then this factor must act early in ontogeny, sometime during the foetal period,
infancy or early childhood, as these are the periods in which most of the body size
determination seems to occur (Bogin 1988, Proos 1993, Delemarrevandewaal 1993,
Henneberg and Louw 1990, 1993). These criteria are met by changing exposure to
pathogens, natural or contained in vaccines, which produce immune responses that
may be biochemically linked to determinants of stature, and by foods which contain
chemical substances influencing regulators of growth. For instance, beef and poultry
may contain additional animal growth hormones used to obtain higher yield on
commercial farms (Moishezon-Blank 1991). Ingestion of dairy products containing
similar natural or artificial compounds is another possibility. Bigger is not always
better, and continued secular increases of body height may be manifestations of the
introduction of undesirable substances into human bodies.
Acknowledgements. I thank Kosette M Lambert and Carmen de Miguel for their

help with anthropometric survey of Aborigines and with some technical aspects of
this study.
4. REFERENCES
Australian Bureau of Statistics, 1995, How Australians Measure Up, Catalogue No. 4359.0,
Commonwealth of Australia, Canberra.
Bielicki, T., and Wei on, Z., 1982, Growth data as indicators of social inequalities: the case of Poland.
Yearbook of Physical Anthropology 25, 153-167.
Bogin, B., 1988, Patterns of Human Growth (Cambridge: Cambridge University Press).
Brown, P., 1992, Recent human evolution in East Asia and Australasia. Phil. Trans. R. Soc. Lond. B 337,
235-242.
Cavalli-Sforza, L.L., and Bodmer, W.F., 1971, The Genetics of Human Populations (San Francisco:
Freeman).
Delemarrevandewaal, H.A., 1993, Environmental factors influencing growth and pubertal
development, Environmental Health Perspectives IOI(Suppl. 2), 39-44. De Miguel, C., and
166 CHAPTER 14
Henneberg, M., 1999, Variation in hominid body size: Do we know how big our ancestors were?
Perspectives in Human Biology, 4, 65-81.
Dettwyler, K.A., 1992, Nutritional status of adults in rural Mali, American Journal of Physical
Eiben, O.G., 1996, Selected data on growth and physical fitness of healthy Hungarian boys and girls in
our changing world, Collegium Anthropologicum, 20, 73-81.
Falconer, D.S., 1980, Introduction to Quantitative Genetics (Edinburgh: Oliver and Boyd).
Frayer, D., 1984, Biological and cultural change in the European Late Pleistocene and Early Holocene.
In The Origins of Modem Humans: A World Survey of the Fossil Evidence edited by F.H. Smith
and F Spencer (New York: Alan R Liss), pp 211-250.
Gerver, W.J.M., Debruin, R, and Drayer, N.M., 1994, A persisting secular trend for body
measurements in Dutch children - The Oosterwolde II study, Acta Paediatrica, 83, 812-814.
Gutierrez-Muniz, J.A., Berdasco-Gomez, A., Esquivel-Lauzurique, M., Jimenez-Hernandez, J.M.,
Mesa-Payan, D., Posada-Lima, E., Romero-Del-Sol, J.M., Quesdada, M.R., and Silva-Ayzaguer,
L.C., 1994, The 1982 Cuban national growth and development study: Secular changes in weigbt and
height, Acta Medica Auxologica, 25, 115-127.
Hauspie, R.C., Vercauteren, M., and Susanne. C .. , 1996, Secular changes in growth, Hormone
Research, 45: (Suppl. 2), 8-17.
Hauspie, R.C., Vercauteren, M., and Susanne, C., 1997, Secular changes in growth and maturation: an
update, Acta Paediatrica, Supplement, 423, 20-27
Henneberg, M., 1976, Reproductive possibilities and estimations of the biological dynamics of earlier
human populations, Journal of Human Evolution, 5,41-48
Henneberg, M., 1988, Decrease of human skull size in the Holocene, Human Biology 60, 395-405.
Henneberg, M., 1992, Continuing human evolution: brains, bodies, and the role of variability,
Transactions of the Royal Society of South Africa 48, 159-182.
Henneberg, M., Budnik, A., Pezacka, M., and Puch, A.E., 1985, Head size, body size, and intelligence:
intraspecific correlations in Homo sapiens sapiens. Homo, 36, 207-218.
Henneberg, M., and Louw, GJ., 1990, Height and weight differences among South African Urban
schoolchildren born in various months of the year, American Journal of Human Biology, 2, 227-233
Henneberg, M., and Louw, G.J., 1993, Further studies on the month-of-birth effect on body size: Rural
schoolchildren and an animal model, American Journal of Physical Anthropology, 91, 235-244.
Henneberg, M., and Louw, GJ., 1998, Cross-sectional survey of growth of urban and rural "Cape
Coloured" schoolchildren: Anthropometry and functional tests. American Journal of Human
Biology, 10, 73-85.
Henneberg, M., and Steyn, M., 1993, Trends in cranial capacity and cranial index in Subsaharan Africa
during the Holocene, American Journal of Human Biology, 5,473-47.
Henneberg, M, van den Berg, E.R, 1990, Test of socioeconomic causation of secular trend: Stature
change among favored and oppressed South Africans are parallel, American Journal of Physical
Huang, Y.C., and Malina, R.M., 1995, Secular changes in the stature and weight of Taiwanese children,
American Journal of Human Biology, 7,485-496.
Jacobs, K.H., 1985, Climate and the hominid postcranial skeleton in Wilrm and early Holocene Europe.
Current Anthropology, 26, 512-514.
Jamison, P.L., 1990, Secular trends and the pattern of growth in arctic populations, Soc. Sci. Med., 30,
751-759.
Kimura, K., 1967, A consideration of the secular trend in Japanese for height and weight by a graphic
method, American Journal of Physical Anthropology, 27, 89-94.
Lancaster, H.O., 1957, Some anthropometrical values of women in Australia, Medical Journal of
Australia (December 1957), 897-900
Leatherman, T.L., Carey, J.W., Thomas, RB., 1995, Socioeconomic change and patterns of growth in
the Andes, American Journal of Physical Anthropology, 97, 307-321.
Loesch, D.Z., Stokes, K., and Huggins, RM., 2000, Secular trend in body height and weight of
Australian children and adolescents, American Journal of Physical Anthropology, 111,545-556.
Louw, GJ., and Henneberg, M., 1997, Lack of a secular trend in adult stature in South African males
born between 1954 and 1974, Homo, 47,54-61.
Malina, R.M., 1990, Research on secular trends in auxology, Anthropologisher Anzeiger, 48, 209-227.
Mathers, K., and Henneberg, M., 1995, Were we ever that big? Gradual increase in hominid body size
over time, Homo, 46, 141-173.
Moishezon-Blank, N., 1991, Commentary of the possible effect of hormones in food on human growth.
Medical Hypotheses, 38, 273-277.
Mueller, W.H., 1976, Parent-child correlations for stature and weight among school aged children: A
review of24 studies. Human Biology, 48, 379-397.
Oyedeji, G.A., Olamilulo, S.K., Osinaike, A.I., Esimai, V.c., Odunusi, E.O., and Aladekomo, T.A.,
1996, Secular trends in the growth of children aged 0-6 years in a rural Nigerian community,
Annals of Tropical Paediatrics, 16, 11-17.
Pretty, G.L., Henneberg, M., Lambert, K.M., and Pro kopec, M., 1998, Trends in stature in the South
Australian Aboriginal Murraylands, American Journal of Physical Anthropology, 106,505-514.
Piontek, J., 1971, Ocena stanu rozwoju fizycznego poborowych powiat6w Inowroclaw i Tuchola
urodzonych w drugiej poTIowie XIX wieku w por6wnaniu z wsp61czesnll mlodzieza Pomorza i
Kujaw, In Rozw6j i Dorobek Antropologii Polskiej w 25-leciu PRL. Poznaj: UAM, pp 71-75.
Prince, J.M., 1995, Intersection of economics, history and human biology - Secular trends in stature in
nineteenth-century Sioux Indians, Human Biology, 67, 387-406
Prokopec, M., 1984, Secular trends in body size and proportions, and their biological meaning. Studies
in Human Ecology, 6, 37-61
Proos, L.A., 1993, Anthropometry in adolescence - Secular trends, adoption, ethnic and environmental
differences, Hormone Research, 39, 18-24
Rao, K.V., Balakrishna, N., and Shatrugna, V., 1993, Secular trends in height of well-to-do adults and
the associated factors, Man in India, 73, 267-273
Roberts, D.F., 1994, Secular trends in growth and maturation in British girls, American Journal of
Human Biology, 6,13-18
Roche, A.F., 1979, Secular trends in stature, weight and maturation, Monographs of the Society for
Research in Child Development, 44, 2-27
Roberts, D.F., Billewicz, W.Z., and McGregor, LA., 1978, Heritability of stature in a West African
population, Annals of Human Genetics, 42, 15-24
Rosenbaum, S., Skiner, R.K., Knight, LB., and Garrow J.S., 1985, A survey of heights and weights of
adults in Great Britain, 1980. Annals of Human Biology, 12, 115-127
Soar, N., 1999, An examination of the secular trend in Australian male stature as seen in the military
data, Perspectives in Human Biology, 4, 157-166.
Sobral, F., 1990, Secular changes in stature in southern Portugal between 1930 and 1980 according to
conscript data, Human Biology, 62, 491-504
Spurgeon, lH., French, K.E., Giese, W.K., Steele, M.F., Utenko, V.N., Bundzen, P.V., and Rogozkin,
V.A., 1994, Somatic comparisons of urban and rural Russian boys ages 6, 9, and 15 years, living in
St Petersburg and surrounding areas, American Journal of Human Biology, 6,141-151
Susanne, C .. , 1971, Recherche sur la transmission des caracteres mesurables de ('homme, Societe
Royale des Sciences Naturelles de la Belgique: Brussels
Styne, D.M., and McHenry, H., 1993, The evolution of stature in humans, Hormone Research,
39(Suppl. 3), 3-6
Tobias, P.V., 1975, Stature and secular trend among Southern African Negroes and San (Bushmen),
South African Journal of Medical Science 40, 145-164
Tobias, P.V., 1985, The negative secular trend, Journal of Human Evolution, 14,347-356
Tobias, P.V., 1990, Adult stature in southern African Negroes - further evidence on the absence of a
positive secular trend, South African Medical Journal, 78, 97-101
Ulijaszek, S.J., 1993, Evidence for a secular trend in heights and weights of adults in Papua-New
Guinea, Annals of Human Biology, 20, 349-355
van Wieringen, lV., 1986, Secular growth changes. In Human Growth. Vol 3 Methodology Ecological
Genetic and Nutritional Effects on Growth edited by F. Falkner and lM. Tanner (New York:
Plenum Press), pp. 307-331
Weber, G., Seidler, H., Wilfing, H., and Hauser, G., 1995, Secular change in height in Austria - an
effect of population stratification, Annals of Human Biology, 22, 277-288
Wolanski, N., 1978, Secular trend in man: Evidence and factors. Collegium Anthropologicum, 2, 69-86
Wrzosek, A., 1931, 0 stosun~u niekt6rych pomiar6w antropologicznych i typ6w rasowych
morfologicznych do sprawnoEci umyslowej. Przeglad Antropologiczny, 5, 1-16.
CHAPTER 15
PATIERNS OF GROWTH AND SCHOLASTIC

PERFORMANCE - A CASE STUDY OF THE
ASSOCIATED FACTORS
K. VISWESWARA RAO, N. BALAKRISHNA, S. VEENA
National Institute of Nutrition, Indian Council of Medical Research, Jamai Osmania,

Hyderabad, India
1. INTRODUCTION
Though there are many studies comparing anthropometric measurements of children
belonging to low socio- economic group with those of other studies and developed
countries, studies are very few on children belonging to well-to-do communities
(Banik 1982, Vijayaraghavan et al. 1971., Nutritional Foundation of India, Scientific
Report No. 11, 1991, Rao and Sastry 1977, Reed and Stuart 1959, Hamill et al.
1977, Tanner et al. 1966). Earlier studies indicated that students of the well
maintained public schools have better anthropometric measurements than those of
other schools where most of the students belong to lower socio-economic status.
Most of the children are found to have substantial retardation in measurements
reflective of skeleton, muscle and fat. Studies on children of well-to-do families with
good parental education are of interest. Also it is of interest to assess the association,
if any, between grades of body built and scholastic performance. A small-scale study
was carried out on school-age children and adolescents of well-to-do and lower middle
class families in Hyderabad. An attempt was therefore made 1) to study the pattern of
growth in children of different socio-economic status; 2) to assess whether a
combination of anthropometric measurements or indices is better for evaluation of
various grades of under nutrition based on body weight, and 3) to study the relative
importance of nutritional status, socio-economic status and paternal nutritional
status associated with scholastic performance of school-going and adolescent
children.

The study was conducted on 837 well-to-do children consisting of 418 boys and 419
girls of 5-18 years age from well maintained residential areas of urban Hyderabad and
data available on 706 children from a nutrition survey carried out on low socio
economic group of urban Hyderabad (Balakrishna 1992) has been utilised.
169
170 CHAPTER 15
Physical growth indicating the current or long-term nutritional status of children

was assessed by taking anthropometric measurements - height, weight, arm
circumference and fat fold at triceps for both income groups. In addition to these
measurements, head and calf circumference and arm length were measured in well-to-
do children. Anthropometric measurements - height, weight, arm circumference, fat
fold at triceps and weight-for-height were expressed as percentage of the available
standards (WHO 1995, lelliffe 1966 lelliffe and lelliffe 1989, Visweswara Rao
1996). Number of subjects surveyed by age, sex and socio-economic status are
provided in Tables 1-5. The sample size studied for each of the anthropometric
measurements was the same. The mean and standard deviation were calculated for all
the measurements and indices like weight-for-height (%) and weightlheight 2 by age
and socio-economic status for both boys and girls. Differences between mean values
were tested using analysis of variance, least significant difference and t-Student test
or normal curve test, whichever was feasible for application (Visweswara Rao 1996,
Snedecor and Cochran 1967, Rao 1952).
Details of scholastic performance from children were collected based on their final
latest year of examination passed in their academic institution.
The best set and relative importance of factors such as socio-economic status,
anthropometric measurements and indices, reflecting growth status of children and
parental nutritional status, associated with scholastic performance, were assessed by
stepwise multiple regression analysis and discriminant function analysis
(Visweswara Rao 1996).
Multiple regression model of body weight with other anthropometric
measurements reflective of skeleton, muscle and fat mass was fitted using the method
of least squares and the best combination of related variables was obtained by
stepwise and step-down regression approach (Visweswara Rao 1976, 1977, 1996,
Drapper and Smith 1981, Mayachoudhry and Visweswara Rao, 1983, 1984,
Mayachoudhry et al. 1990). Stepwise and step-down regression procedures indicated
a good agreement in the selection of predictor variables (Krishna et al. 1996).
The parental nutritional status was assessed by taking weight, height and body
mass index (weightlheight2) of parents.
The details of educational status and occupational status of parents, per capita
family income, family size and type of family were collected with a carefully prepared
family schedule.
Scholastic performance was assessed on the basis of percentage of marks obtained
in the latest school or college examination. The children were classified into the
following grades of marks:
Grade %marks obtained
Good 60 or above
Poor below 60
Mean and standard deviation (SD) of indicators of socio-economic status, parental
nutritional status and growth and development were calculated for children of the
good and poor grades of marks obtained. Differences in mean values of indicators
between children by grades of marks obtained were tested using the appropriate tests
of significance (Visweswara Rao 1996).
The variables which were significantly different between these two groups of
children were chosen and stepwise discriminant function analysis was used to trace
the best combination of variables having the optimal discrimination power. The
GROWTH AND SCHOLASTIC PERFORMANCE 171
relative importance of all the discriminant functions were tested with the use of
Mahalanobis D2 and Fisher's F-test (Visweswara Rao 1977, 1996, Visweswara Rao
et al., 1979). Percentage of subjects properly classified and misclassified into grades
of their scholastic performance was also found with the best sets of variables.
3. RESULTS AND DISCUSSION

The results are shown in Tables 1-5.
3.1 Pattern of Growth

Mean and SD of height, weight, arm circumference, fat fold at triceps and weight-for-
height (%) by age, sex and socio-economic status are given in Tables 1-5. All the
mean values of anthropometric measurements - height, weight and arm
circumference and fat fold at triceps (FFT) varied with age, sex and socio-economic
status.
The mean values of anthropometric measurements and index, weight-for-height
(%) were higher in children of well-to-do families than those of lower socio-
economic group (p < 0.05). Magnitude of differences in measurements between low
and high socio-economic groups varied by age (p < 0.05). This was found to be true
in both boys and girls.
3.2 Differentials of growth with Harvard and NeHS

In well-to-do children, height was closer to that of the Harvard and NCHS children
up to 12 years of age and later it was lower. Weights were lower than the standards
from the age of 5 to 18 years. Studies of Datta Banik, Udani and Vijayaraghavan
showed that children belonging to well-to-do parents were found to have heights and
weights closer to those of Harvard children. This was up to the age of 14 years.
There have been no studies either with a large sample or small sample on well-to-do
children in recent years. A study was therefore, conducted on children belonging to
well-to-do parents who were well educated.
All the measurements except height of study group were lower than those of the
NCHS and Harvard standards. Height was mostly the same up to 12 years and later
lower than the International standards. The magnitude of retardation was found more
in measurements reflective of muscle and fat. These differences may be due to higher
prevalence of adiposity in the children of Harvard and NCHS. The measurements of
the children studied were higher than those of the well-to-do children surveyed by the
Indian Council of Medical Research (1971).
3.3 Relationship between weight, height, weightlheight2 and weightjor-height (%)

To verify and trace whether the combination of height (%) with weight-for-height
(%) or weightlheight 2 was the best to be of use in place of body weight, multiple
regression models were constructed serially by age from 5 to 18 years. Models were
also developed to verify whether there were any age, socio-economic and sex
172 CHAPTER 15
differences. The best set of variables related to body weight (%) was traced with the
calculation of the coefficient of detennination, R2 (%) and adjusted R2 (%).
Table 1. Mean and SD of height and weight by age (years) and socia-economic status
(SES): boys
Height, cm Weight, kg
high SES low SES high SES low SES
Age n mean SD n mean SD mean SD mean SD
5 14 107.2 4.92 53 104.4 5.49 16.3 1.47 15.3 l.89
6 29 116.2 5.03 27 108.6 5.04 19.5 3.11 15.9 l.53
7 25 123.1 4.73 33 114.5 5.42 22.1 3.98 17.4 l.78
8 31 126.7 6.27 30 118.8 6.13 23.8 5.38 18.9 2.18
9 28 133.4 5.15 21 125.7 6.27 26.9 4.67 22.0 3.85
10 33 139.1 5.55 33 128.9 6.30 30.0 4.42 23.2 3.18
11 22 144.0 5.87 37 135.0 6.22 33.3 6.43 25.9 3.73
12 33 148.5 6.86 28 137.9 7.81 35.0 6.02 27.5 4.35
13 31 158.4 7.09 22 145.7 8.29 42.2 6.43 33.1 5.94
14 37 159.8 7.48 20 149.5 7.63 44.0 7.83 34.1 4.62
15 38 167.2 6.88 21 158.5 8.01 49.3 8.03 4l.7 6.75
16 23 167.9 6.86 16 162.5 6.41 52.6 8.22 45.2 4.64
17 37 169.8 4.91 18 163.2 6.55 55.8 9.57 48.3 7.25
18 37 170.0 4.83 14 162.4 5.46 55.7 8.28 47.0 7.05
Table 2. Mean and SD of height and weight by age (years) and socio-economic status
(SES): girls
Height, cm Weight, kg
5 19 110.0 8.35 40 105.0 5.59 16.2 3.30 15.2 l.89
6 20 115.1 6.95 24 110.7 5.39 18.9 4.46 16.7 2.17
7 25 121.5 5.26 30 115.3 4.54 20.6 3.23 18.2 1.80
8 36 129.6 5.67 27 120.2 5.85 24.9 5.96 19.4 2.80
9 29 132.7 6.44 30 122.9 4.83 27.0 5.45 20.9 2.40
10 31 138.9 7.39 27 130.0 6.91 29.2 5.52 23.9 3.14
11 28 145.6 6.94 23 136.3 6.44 35.3 7.75 26.8 4.59
12 30 152.6 5.03 28 14l.0 7.36 41.1 7.64 30.6 5.00
13 38 152.8 6.11 20 145.2 7.84 41.2 6.97 33.2 7.05
14 36 155.9 5.17 25 150.2 6.32 43.3 5.86 37.6 5.09
15 33 156.8 5.55 17 154.5 6.62 47.9 7.68 41.4 6.12
16 29 155.6 5.25 19 151.2 4.86 48.4 8.24 4l.7 5.75
17 33 156.5 5.60 14 154.6 1.31 49.2 6.87 42.9 5.08
18 31 157.5 5.39 9 15l.1 8.41 48.9 5.73 43.4 7.10
Table 3. Mean and SD of arm circumference and fat fold at triceps by age (years)
and socio-economic status (SES): boys
Arm circumference, cm Fat fold at triceps, mm

5 14 15.6 0.77 53 14.5 l.06 6.2 1.06 6.3 l.88
6 29 16.5 l.67 27 14.4 l.09 6.5 1.74 6.0 1.91
7 25 16.9 2.15 33 14.8 0.96 7.0 2.47 5.4 l.25
8 31 17.5 2.15 30 15.1 0.79 7.7 2.69 5.6 l.44
9 28 18.0 2.37 21 15.8 l.62 7.5 3.05 6.0 2.47
10 33 18.8 l.94 33 16.3 0.98 7.9 2.94 6.4 l.94
11 22 20.4 2.63 37 16.8 l.72 8.8 3.77 7.0 3.19
12 33 19.5 2.22 28 17.1 1.30 8.0 2.14 5.8 l.34
13 31 2l.0 2.29 22 18.7 l.79 8.8 3.42 6.4 2.78
14 37 2l.4 2.45 20 18.3 2.07 8.3 3.59 6.6 1.72
15 38 22.4 2.55 21 20.7 2.67 7.9 3.27 6.6 3.18
16 23 23.1 2.40 16 22.0 2.13 10.1 4.64 6.3 l.67
17 37 24.3 2.76 18 23.3 3.31 9.4 3.79 7.1 2.42
18 37 24.5 2.68 14 22.5 2.28 9.9 5.17 6.2 1.75
Table 4. Mean and SD of arm circumference and fat fold at triceps by age (years)
and socio-economic status (SES): girls
Arm circumference, cm Fat fold at triceps, mm

5 19 15.7 l.47 40 14.7 l.03 7.5 l.99 6.7 l.53
6 20 16.3 l.85 24 14.9 l.06 8.2 2.68 6.5 2.14
7 25 16.6 l.92 30 15.3 0.98 7.9 l.94 6.6 l.52
8 36 18.0 l.82 27 15.5 l. 11 9.4 3.05 7.1 2.24
9 29 18.6 2.22 30 15.9 1.19 10.2 2.72 6.8 l.82
10 31 18.8 2.07 27 17.2 2.33 9.3 2.57 7.9 2.08
11 28 20.2 2.46 23 17.3 l.91 11.3 4.48 8.0 2.48
12 30 20.5 2.47 28 18.7 l.91 1l.7 4.55 9.1 3.44
13 38 2l.4 2.31 20 19.5 2.04 12.1 4.18 9.2 3.33
14 36 21.2 l.88 25 20.2 l.70 1l.1 2.85 9.6 2.59
15 33 22.8 2.58 17 20.9 l.77 13.9 4.79 1l.9 3.79
16 29 22.0 2.57 19 2l.2 l.62 1l.5 4.08 1l.4 3.94
17 33 23.2 2.41 14 2l.4 l. 76 14.5 5.09 1l.6 3.50
18 31 23.3 l.63 9 22.3 2.03 12.7 4.75 1l.9 3.05
174 CHAPTER 15
Table 5. Mean and SD of weight-for-height (%) by age (years) , sex and socio-
economic status (SES)
boys girls
Age n mean SD n mean SD n mean SD n mean SD
5 14 94.3 3.8 53 91.2 7.9 19 88.5 12.2 40 90.5 7.8
6 29 92.0 9.0 27 88.6 6.3 20 93.3 14.8 24 89.8 7.4
7 25 93.0 10.8 33 85.5 6.6 25 90.5 8.5 30 90.1 5.1
8 31 93.0 11.5 30 86.5 4.2 36 93.0 11.0 27 86.4 8.0
9 28 94.1 11.8 21 88.5 8.7 29 94.5 14.3 30 87.7 7.2
10 33 94.3 9.8 33 87.3 6.9 31 89.6 10.7 27 87.2 6.6
11 22 94.8 10.1 37 87.2 7.7 28 95.2 16.3 23 87.3 10.1
12 33 91.8 9.2 28 87.7 5.8 30 96.3 14.8 28 88.7 9.4
13 31 93.1 12.3 22 90.3 9.0 38 95.9 15.3 20 86.3 9.9
14 37 92.5 13.3 20 87.0 7.2 36 91.2 11.6 25 87.2 7.4
15 38 91.2 11.1 21 89.3 11.4 33 97.1 13.3 17 86.2 9.5
16 23 94.9 11.8 16 90.0 11.0 29 88.3 13.5 19 87.5 12.5
17 37 94.4 14.5 18 89.6 12.1 33 96.3 12.4 14 85.9 10.7
18 37 92.1 13.7 14 84.4 12.3 31 93.5 9.8 9 85.0 12.7
Height with weight-for-height (%) was found to be the best related to weight than
height alone or height with arm or calf circumferences. The coefficient of
determination was closer to 95%. This was found in both sexes and in all age
groups. This was also found in each of the age groups studied by socio-economic
status. Combination of height and weightlheight 2 was also found best related to
weight. The coefficient of determination was around 99%. This was true in boys and
girls for all ages and socio-economic groups. No other combination yielded such a
high coefficient of determination. Similar findings were observed in studies on
infants, preschool children and adults (Visweswara Rao 1976, 1977, Visweswara Rao
et at. 1986, 1991, Raman et at. 1989). In well-to-do children, weight-for-height was
observed to have higher contribution to variations in body weight than height. In
low socio-economic group, height was observed to have higher contributions to
variations in body weight than that of weight-for-height (%). This was found true in
most of the age groups of low socio-economic status except those of beyond 15 or
16 years.
The contribution of height (%) and weightlheight2 individually to variations in
body weight (%) by age and sex were found different. In well-to-do children,
irrespective of age and sex, weight/height 2 was observed to have higher contribution
to variations in body weight (%) than height. In low socio-economic group, height
(%) was observed to have higher contribution to variations in body weight than that
of weightlheight 2 . That was found true in most of the age groups of boys of low
socio-economic group except those of beyond 15 years. Whereas for girls,
contribution of weightlheight 2 was more than height (%) in most of the ages except
5 to 8 years. This variation may be due to variations in grades of development of
skeletal and muscle reflective measurements and prevalence of various forms of
malnutrition of higher or shorter duration.
Height with weight-for-height or weightJheight 2 was better for quantification and

understanding the forms of malnutrition than those forms of malnutrition using
weight alone. Classifications of malnutrition suggested with the use of height with
weight-for-height or weightJheight2 were of immense use for children of age, 1-18
years (Visweswara Rao 1996).
3.4 Scholastic performance and the associatedfactors.
Of all children, 76.9% of the boys and 79.7 of the girls were with scores of
scholastic performance of 76.0% or above. Sex differences in these scores were not
significant (p > 0.05).
Height, weight, head circumference, child's education, educational status of
parents, occupation of mother, family size and per capita income were different (p <
0.05) between children with poor and good scholastic performance. All these
indicators were better in children of good scholastic performance.
Of all the measurements - weight, fat fold at triceps and calf circumference in
order constituted the best set of measurements related to scholastic performance for
boys. The coefficient of determination of the multiple linear regression models with
these measurements was 7.8 percent. In girls, height and calf circumference in order
were related to scholastic performance. The value of the coefficient of determination
(R2) was 7.5%.
Order of indicators related to scholastic performance were education of mother,
family size and per capita income for boys, whereas in girls, the indicators in order
were family size, type of family, per capita income and education of mother.
Coefficients of determination of the models for boys and girls respectively were 7.1
and 9.0 %.
Of the indicators, maternal weight and weightJheight2 and father's height in that
order were better related to scholastic performance in boys. Coefficient of
determination, R2 (%) was 4.8. In girls, the indicators in order were weightJheight2 ,
weight and height of mothers. Coefficient of determination was 9.3%.
The best set of variables related to scholastic performance was found to be the
levels of education of child and mother and head circumference for boys. The
coefficient of determination was 17.4 percent. In girls, height, education of child,
family size, head circumference and father's weight/height2 were related to scholastic
performance. The coefficient of determination was 17.1 %. Mean values of
anthropometric measurements of height, weight, head and levels of education of
mother, father and child, occupation of mother, family size and per capita income
were different (p < 0.05) between children with poor and good scholastic
performance.
Educational status of mother and child and head circumference of child were found
best for differentiation of good and poor grades of scholastic performance of boys.
Height, arm circumference and weight-for-height of children enhanced discrimination
power. The percentage miscIassification was 33.7(%).
In girls, height and father's weightJheight 2 were found best for differentiation of
variations in scholastic performance. Percentage miscIassification was around 30.0.
Head circumference, education of father and maternal height and weightJheight2
enhanced the improvements in scholastic performance.
176 CHAPTER 15
Parents' nutritional status, child's scholastic perfonnance at school or college and

parents' education and economic status were associated with better nutritional status
of children. Hence, interaction between nutritional and non-nutritional variables was
found significantly associated with scholastic perfonnance of children under study.
Educated parents should playa pivotal role in planning health and child development
programs.
4. REFERENCES:
Balakrishna, N., 1992, Efficiency of some parametric tests of inference in nutritional anthropometry.
PhD Thesis (Hyderabad: Osmania University). pp 191-209.
Banik, N.D.D., 1982, Semilongitudinal growth evaluation of children from birth to 14 years in different
socio economic groups. Indian Pediatrics, 19,353-369.
Draper, N.R, and Smith, H., 1981, Applied Regression Analysis. Second Edition (New York: John
Wiley & Sons Inc.).
Hamill, P.V.V., Drizd, T.A., Johnson, C.L., Reed, RB. and Roche, A.F., 1977, NCHS growth curves for
children - birth to 18 years. Vital and Health Statistics Series 11. No. 65, DHEW Pub. No. (PHS),
78-1650, USHES-PHS, Hyattsville, M.D.
Indian Council for Medical Research (lCMR), 1971, Growth and physical development of Indian
infants and children - All India. (New Delhi: Indian Council of Medical Research).
Jelliffe, D.B., 1966, Assessment of the nutritional status of the community. WHO monograph series No
53 (Geneva: World Health Organisation).
Jelliffe, D.B., and Jelliffe, E.F.P., 1989, Community nutritional assessment with special reference to less
technically developed countries. (Oxford: Oxford University Press), pp. 13-140, 355-443, 524-550.
Krishna, D., Balakrishna, N., Visweswara Rao, K., and Reddy, PJ., 1996, Selection of predictor
variables in multiple regression: A case study of agreement between procedures. In Biostatistics -
A manual of statistical methods for use in Health, Nutrition and Anthropology, edited by
Visweswara Rao K. New Delhi: Jaypee Brothers Medical Publishers (P) Ltd. pp. 612-621.
Mayachoudhury, and Visweswara Rao, K., 1983, Nutritional status of preschool children and the
associated factors. Indian Journal of Nutrition and Dietetics, 20, 18-29.
Mayachoudhury, and Visweswara Rao, K., 1984, Association of growth status and mental function in
preschool children. Indian Journal of Nutrition and Dietetics, 21,1-18.
Mayachoudhury, Balakrishna, N. and Visweswara Rao, K., 1990, Mental ability of preschool children:
Relative importance of some associated factors. In Statistics in Health and Nutrition, Proc. National
Seminar Oct. 27-29, 1988, edited by Visweswara Rao, K., Radhiah, G., and Narayana, V.
Hyderabad: National Institution of Nutrition. pp. 201-218.
Nutrition Foundation of India, 1991, Growth performance of affluent Indian Children - Growth
standards for Indian children (New Delhi: NFL) Scientific report series 11.
Raman, L., Vasanthi, G., Parvathi, e., Vasumathi, H., Rawal, A., Visweswara Rao, K. and Balakrishna,
N., 1989, Growth and development of infants in urban slums of Hyderabad. Indian Journal of
Nutrition and Dietetics, 26, 196-205.
Rao, D.H., and Sastry, J.G., 1977, Growth Pattern of well-to-do Indian adolescents and young adults.
Indian Journal of Medical Research, 66, 696-706.
Rao, e.R, 1952, Advanced statistical Methods in Biometric Research. (London: John Wiley & Sons
Inc.). pp. 236-272, 435-490.
Reed, RB., and Stuart, H.e., 1959, Pattern of growth in height and weight from birth to 18 years of age.
Pediatrics, 24, 904-921.
Snedecor, G.W., and Cochran, W.G., 1967, Statistical Methods. (Ames: IOWA State University Press),
pp.90-106.
Tanner, J.M., White house, RH., and Takahashi, M., 1966, Standards from birth to maturity for height,
weight, height velocity and weight velocity - British Children. 1965, part II. Archives of Diseases of
Childhood, 41, 613-635.
Vijayaraghavan, K., Singh, D., and Swaminadhan, M.e., 1971, Heights and weights of well-nourished
Indian School Children. Indian Journal of Medical Research, 59, 648-654.
Visweswara Rao, K., 1976, Efficiency of anthropometric indices for the diagnosis of malnutrition. PhD
thesis (Hyderabad: Osmanian University) pp. 60-201.
Visweswara Rao, K., 1977, Analysis of relative importance of factors affecting body weight changes in
preschool children by nutritional status. Bulletin of the International Statistical Institute, 41st session,
New Delhi. 47, 717-721.
Visweswara Rao, K., Reddy, PJ., Narayanan, T.P., and Subhadradevi, V., 1979, Discriminant function
analysis: A case study for evaluation of various forms of Protein energy malnutrition. Indian
Journal of Medical Research, 69, 99-108.
Visweswara Rao, K., Balakrishna, N., and Adinarayana, K., 1986, Critical limits of some
anthropometric measurements and indices for the assessment of nutritional status. Indian journal of
Nutrition and Dietetics, 23, 88-99.
Visweswara Rao, K., Balakrishna, N., Veena, S. and Thimmayamma, B.V.S., 1991, Body mass index in
school age children and adolescents. Indian Journal of Physical Anthropology and Human
Genetics, 17, 113-124.
Visweswara Rao, K., 1996, Biostatistics - A manual of statistical methods for use in Health, Nutrition
and Anthropology, edited by Visweswara Rao, K .. (New Delhi: Jaypee Brothers Medical
Publishers Ltd) pp. 47-69,137-182,237-284,461-506.
WHO, 1995, Physical status: The use and interpretation of anthropometry. Report of WHO expert
committee. (Geneva: WHO) pp. 37-115, 424-452.
CHAPTER 16
MENTAL ABILITY AND COGNITIVE THINKING IN

RELATION TO SEX, PUBERTAL STAGE AND SOCIO-
ECONOMIC BACKGROUND
G. TANNER-LINDGREN
Department of Educational Research, Stockholm Institute of Education, Sweden
1. INTRODUCTION
It has long been thought that physical and mental growth and development in
children proceed to some degree in parallel, early developers in the one sense being
on average early developers also in the other, as seen e.g. in parallel secular trends
(Tanner 1961, Muuss 1970, Husen 1974, Emanuelsson and Svensson 1986), but
there is surprisingly little published evidence that this is the case.
The connection has also been asserted because of the sex differences found in
favour of girls concerning school achievement, mental ability and other overt
behaviour during puberty, since girls on average become physically mature about
two years earlier than boys (Tanner 1962, 1989, Lindgren 1978). However, the sex
differences do not always point in the same direction and the results found for boys
compared to the girls in different subjects and behaviours are contradictory (Anastasi
1958, Maccoby and Jacklin 1974, Kelly 1978).
Also evidence for mental growth spurts analogous to physical growth spurts has
been suggested and actually found (Ljung 1965, Epstein 1974, Lindgren 1979).
More directly, the relationship between physical and mental growth and
development has been studied by comparing mental test scores of children and youth
in relation to their physical growth and maturation (Shuttleworth 1939, Boas 1941,
Nisbet and Illesley 1963, Douglas and Ross 1964, Johannesson 1974, Kohen-Raz
1977, Lindgren 1979, Westin-Lindgren 1982, 1984). The main assumption behind
these studies has been that early maturers would achieve better at school than late
maturers. The results, however, have been inconsistent, and very few studies have
been conducted in the last 15 - 20 years (Tanner 1999). In 1980, however, data for
Swedish schoolchildren were collected in order to investigate further the relation
between the physical/physiological and the mental/cognitive growth and
development during puberty. The purpose of the study was to analyse the pupils'
mental and cognitive ability in relation to their sex, pubertal stage and socio-
economic background. Although the study by now is a bit old, its results have never
been published; the aim of the present paper is to present them.
179
P. Dasgupta and R. Hauspie (eds.), Perspectives in Human Growth, Development and Maturation, 179-20l.
180 CHAPTER 16
2. SUBJECTS, MATERIALS AND METHODS
2.11nvestigation group
Two co-educational comprehensive schools with somewhat different socio-economic

status nearby the capital, Stockholm, were chosen for the study. From each school
two classes from Grades 4 (age 10 - 11 years), 6 (age 12 - 13 years) and 8 (age 14 -
15 years) respectively were randomly selected so as to end up with a total of 12
classes. The distributions of boys and girls participating in each Grade from the two
schools are shown in Table 1.
Table 1. Distribution of boys and girls respectively participating in each Grade (Attrition
is given in parenthesis)
Grade Age, years Boys Girls Total

4 1O± 48 (4) 43 (-) 91 (4)
6 12± 39 (6) 60 (6) 89 (12)
8 14± 47 (5) 55 (3) 102 (8)
Total 134 (15) 148 (9)
The loss of individuals, who were either ill during the investigation period or did not
like to participate in the study, was 24 pupils out of 306 (8%). Thus the study
group consisted of 282 pupils.
2.2 Pubertal stages

The children's pubertal stages were rated according to the Tanner-Whitehouse method
described by Tanner (1962). The ratings were done by two experts from the Institute
of Child Health, Department of Growth and Development at the University of
London. The girls' breast development (in five stages) and their pubic hair (in five
stages) were rated by Dr. Claire Burns and the boys' genital development (in five
stages) and their pubic hair (in five stages) were rated by Dr. Michael Preece. Girls'
and boys' axillary hair development (in three stages) was also rated by the respective
female and male paediatrician. Following the recommendation of Mitchell (1980) a
female doctor rated the girls and a male doctor rated the boys. The whole procedure
took place in the school nurse's clinic (there is one in every school in Sweden) and
with her help the children's heights and weights were measured at the same time.
The children were encouraged to ask the doctors anything they wanted to know
concerning their health, body or physical growth and so they did, indeed. During the
whole examination, recording of the observations as well as translating the English
into Swedish and vice versa was performed by the present author.
MENTAL ABILITY AND COONITIVE THINKING 181
Socio-economic status (SES) was defined by father's education as described by

Lindgren (1976) and the children grouped accordingly into three groups:
SES I employers (mostly university graduates)

SES II salaried employees (e.g. small owners, administrators)
SES III manual workers
The distributions on SES-groups in each school and the total distribution on

those groups are shown in Table 2.
Table 2. Distribution of SES-groups in each school as well as for the total group
(percentages are given in brackets)
Social group School A School B Total

SES I 43 (29) 12 (8) 55 (18)
SES II 67 (46) 63 (39) l30 (42)
SES III 34 (23) 81 (51) 115 (38)
Missing data 3 (2) 3 (2) 6 (2)
Total 147 (100) 159 (100) 306 (100)
As can be seen from Table 2, SES-group I is more represented in School A

compared to School B, and SES-group III less represented in School A compared to
School B. However, the total distribution on SES-groups is quite like what is
usually found in densely populated areas in Sweden.
2.3 Mental ability and cognitive thinking

The children's mental ability was measured by means of subtests taken from the so
called DBA test battery (Different Abilities Tests) which consists of twelve subtests.
The contents and construction of these tests were developed by Haemqvist (1960). In
the present study two of the subtests were used which measured verbal ability and
logical-inductive ability respectively. For a detailed description of the contents of
these tests, see Lindgren (1979).
The test "Opposites" is a verbal test, where opposites of given words are required
to be marked. The duration of the test is 10 minutes. There are 40 items to be
marked. The reliability coefficient is r :::: 0.89 for girls and r :::: 0.87 for boys
according to the split-half method.
The test "Letter Groups" is a logical-inductive test, where out of four alternative
letter groups the subject has to choose which one, out of four, differs in logical
construction. The duration of the test is 15 minutes and there are 30 items to be
solved. The reliability coefficient was by the split-half method r :::: 0.87 and r :::: 0.85
for girls and boys respectively.
In addition to these two tests measuring different aspects of mental ability, a
third test "Similarities" measuring cognitive thinking was used. This test was
constructed by Sandgren (1974) for a previous Swedish study. An English
182 CHAPTER 16
translation of the test by the present author is given in the Appendix. The test
consists of 20 pairs of nouns (e.g. a car - a bus) and out of five alternatives for each
pair of nouns the subject has to mark the best and the worst similarity respectively
between the nouns of the pair. The alternatives for each pair of nouns were
constructed according to the theory of Piaget; that is the alternative answers were
constructed to give different stages of cognitive thinking from an egocentric to an
abstract stage of thinking (Inhelder and Piaget 1985), as in the example given; you
can crash into them, have wheels, give out exhaust fumes, you can ride in them and
are vehicles (see Figure 1).
Cognitive level Similarities between a car - a bus
Abstract
logic
C = Conceptual < ~ -
.(~ 5. are vehicles
(abstract overriding concepts)
F = Functional 4. you can ride in them
(their function is alike)
PE = Perceptual < I - ""' 3. give out exhaust fumes
(form, size, place)
PA = Particular 2. have wheels

(part of the whole)
E = Egocentric 1. you can crash into them
(emotional)
Egocentric
logic
Figure 1. Similarities between the pair of nouns "a car - a bus" mirroring different levels
of cognitive thinking from an emotional to an abstract level (Translated from Sandgren
1974)
After instruction and some training (see first page of the test in the Appendix)
the children could work with the test without any time limit - thus minimising
stress. This test was easy to administer and the children liked it since no answer was
ever wrong. However, marking the "worst" similarity did not work out well, since
the children in general thought that once the "best" similarity had been chosen the
other four alternatives were all equally worse. Thus choices of the "worst"
similarities are not analysed here.
2.4 Data analyses

When testing differences between SES-groups in the distributions of pubertal stages
a usual X 2 -test was used. When testing the impact of sex, pubertal stage and SES-
group on mental ability and cognitive level in Grade 4, 6 and 8, three-way ANOV As
were performed using a GLM-procedure (General Linear Model) as described in SAS
User's Guide: Statistics (1985).
2.5 The procedure a/investigation
The tests measuring mental ability and cognitive thinking were administered on the
same occasion and within two weeks of the occasion when the children's pubertal
stages were rated.
Before conducting the study, much work was done to inform the two schools -
including staff, school nurses, parents and children - about the purpose and procedure
of the study. All school personnel, parents and children were informed that they did
not have to participate if they did not want to, and the collected data were to be
treated as strictly confidential. Written consent was asked for and also given by the
parents. On the whole the study was positively received by the school personnel as
well as by the parents and their children. The study was conducted in March 1980.
3. RESULTS
The results will be presented in three parts. First, the results from the pubertal
ratings will be given; secondly, the analysis of the results on the test "Similarities"
measuring cognitive thinking related to sex and pubertal stage; and finally the
results from three-way ANOV A's of the tests measuring mental abilities and
cognitive thinking controlling for sex, pubertal stage and SES-group.
3.1 Pubertal stages and SES-group.
The distributions of pubertal stages in relation to SES-group for boys and girls
respectively are shown in Table 3. (Grades 4, 6 and 8 have been combined in the
Tables, when testing the distributions regarding differences between SES-groups).
No significant differences between the distributions of pubertal stages were found for
SES-groups I, II and III respectively - neither for girls nor for boys (p > 0.05).
Table 3. Distribution of pubertal stages for boys and girls by SES-group
Boys Girls
Pubertal SES-groups SES-groups
stage I II III I II III
I 5 15 17 3 11 8
II 6 13 15 3 8 8
III 1 3 6 9 11 11
IV 2 7 8 9 18 8
V 5 7 4 4 17 10
X 2 = 4.11, p > 0.05 X 2 = 6.79, p > 0.05

184 CHAPTER 16
3.2 The test of cognitive thinking - "Similarities ".

The level of cognitive thinking on a scale of 1 to 5 was rated for each pupil and for
each of the 20 items. Thus a given pupil might score, for example 10 fives, 5 fours
and 5 threes. For a given subsample - boys in Grade 4 say - the average number of 5
ratings, 4 ratings etc. was calculated, and these are illustrated in Figure 2. Thus 10 -
11 year-olds only get ratings of 5 or around 6 out of 20 items, whereas 14 - 15-year-
olds get ratings of 5 on around 13 of the 20 items.
16
... 14 . ::.:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::Gra:de:: 4::(io::; : n : y:e.3.:r.S[::::::::::::
~ 12
g 10
Q.) 8
i?
Q)
6
> 4
« 2
0
0 2 3 4 5
Cognitive level
": ~ f •
~ 12
I'::
Q)
10
8
.....
..... ~
i?
Q)
6
> 4
« 2
0
0 2 3 4 5
Cognitive level
16
b 14 ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::G:rad~}f(i4::;I5:ye3.rs[::::::::::::
.0 12
~ 10
II.)
t:IO
8
E!
II.)
6
> 4
« 2
0
o 2 3 4 5
Cognitive level
Figure 2. Average numbers of choices of different response alternatives (cognitive

levels) on the test "Similarities" in Grade 4, 6 and 8. Boys and girls
MENTAL ABILITY AND COGNITIVE THINKING 185
As can be seen in Figure 2 girls more often than boys consistently tend to
choose answers on the cognitive levels 4 and 5 (the functional and abstract levels)
from Grade 4 up to and including Grade 8. This tendency seems especially marked in
Grade 6. At the same time, of course, boys more often than girls tend to choose
cognitive levels 1, 2 and 3 (the egocentric, particular and perceptual levels) in all
three Grades.
The distributions on levels of cognitive thinking by sex and pubertal stage are
also shown for each Grade in the Appendix: Figures 1 - 3.
Numbers of chosen "abstract" answers were then analysed in relation to sex,
pubertal stage and SES-group for each Grade.
3.3 Mental ability and cognitive level in relation to sex, pubertal stage and SES-
group.
The ANOV A's of the test scores on the two mental ability tests and the test
measuring cognitive level were done after having transformed the raw scores into
stanine-scores. Regarding the test "Similarities" only numbers of answers
concerning the levelS ("abstract"-level) of the cognitive thinking were transformed.
The results from these three-way ANOV A's of test scores related to sex, pubertal
stage and SES-group can be found in the Appendix (Tables 1 - 9). A summary of
the results is given in Table 4 below.
Table 4. Summary of results (significance of F) concerning 3-way ANOVAs (GLM-

procedure) on the tests measuring mental ability and cognitive level in relation to sex,
pubertal stage and SES-group (P: pubertal stage, S: sex)
Source of variation
Type III tests
Test Grade Model S P SES SESxP PxSES S x P xSES
Opposites 4
verbal 6
8 ** **
Letter groups 4 ** * **
logical-induc- 6 * **
tive 8 * *
Similarities 4
abstract cog- 6 * *
niti ve level * *
8 **
** p < 0.01 * p < 0.05 - not significant
Regarding the results on the verbal ability test "Opposites" in Grades 4 and 6, there
were no significant differences between either sexes, pubertal stages or SES-groups.
Neither were there any significant interactions between these factors. In Grade 8,
however, the sex difference is significant (p < 0.01) the girls doing better (see also
Figure 3).
186 CHAPTER 16
8,- Verbal ability grade 4 8 iLogic inductive grade 4 8 iCognitive levels grade 4
",6 6 6
~
o
y
~
u
~
'"
gt 4 4
.~
~ 2 2
o --':-----:':c--=----=~---::'-::--'
L' o L' _-':---:'::---::'::,--=-:-L--' o L' _-':---:'L--'-_-'-_'---'
II III IV V
8 Verbal ability 8 Logic inductive 8 Cognitive levels

grade 6 grade 6 grade 6
",6 6 6
~
o
u
'"gt 4 4
.~
~ 2 2
o 'L--'-_~-==-=-----='---' 0' II III IV V

o L' --':-----:':c---::L------=~---::'-::--'
r
8 8 ognitive levels grade 8
Verbal ability grade 8 Logk ;"d~d" :
",6
~
u
'"gt
(;)
(OJ
l'l
6
4 4
A
'§
~ (')
2 2
0' I II III IV V 0' I II III IV V 0' I II III IV V

Pubertal stages Pubertal stages Pubertal stages
Figure 3. Average scores for boys and girls on the tests measuring verbal ability, logical
inductive ability and cognitive level in relation to pubertal stage in Grade 4, 6, and 8
(0: Boys, .: Girls)
MENTAL ABILITY AND COGNITNE THINKING 187
On the logical inductive test "Letter groups" there were in Grade 4 significant
differences (p > 0.05) between children of different pubertal stages in favour of the
early maturing ones (in this case mainly the girls). At the same time there were
significant differences between SES-groups (p < 0.01). Children from SES-groups I
and II performed better than those in SES-group III.
In Grade 6 there was a significant interaction between sex and pubertal stage (p <
0.01). This meant that there were in this grade (age 12 - l3 years) no significant
differences between girls of different pubertal stages, but there were for boys - the
early maturing boys doing better than the later ones (see Figure 3).
In Grade 8 there was only one significant interaction; the one between sex,
pubertal stage and SES-group (p < 0.05). Regarding pubertal stages, late maturing
boys (except for the prepubertal boys) were performing better than the early
maturing ones, while there were no differences between early and late maturing girls.
Among the girls there were no SES-group differences, while among the boys SES-
group II was doing better than SES-group I.
On the test measuring level of cognitive thinking - "Similarities"- there were no
significant differences or interactions between sex, pubertal stage and SES-group in
Grade 4.
In Grade 6, however, there were significant differences between children of
different pubertal stages - the early maturing ones performed better. At the same
time, there was a significant interaction between sex and pubertal stage (p < 0.05),
which is illustrated in Figure 3.
In Grade 8, there were no significant differences on the test "Similarities"; either
between sexes, the pubertal stages or SES-groups. Two significant interactions
were, however, found. One between sex and pubertal stage (p < 0.05), which is
illustrated in Figure 3; the other interaction between pubertal stage and SES-group
(p < 0.01) indicating that later maturing children in SES-groups I and II performed
better than the earlier maturing ones, while there were no differences between early
and late maturers in SES-group III.
4. DISCUSSION
The results of this cross-sectional study of Swedish schoolchildren lend credence to
the hypothesis that children who are early maturers in the physical sense are prone
to be early maturers in at least some aspects of their mental development also.
A previous study, at that time longitudinal, of Swedish urban schoolchildren
born in 1955 also relating mental ability and school achievement to early and late
maturation (Westin-Lindgren, 1982) gave results quite consonant with those of the
present investigation. However, there were also - apart from the difference in the
design of the two studies - some other disparities between them.
For one thing, the criterion for defining late and early physical maturation was in
the longitudinal 1955-study age at maximum height velocity during puberty (the so-
called Peak Height Velocity age or PHV-age), while in the cross-sectional 1980-
study the criterion was ratings of pubertal stages in five stages (from prepubertal to
adult stage). This means that in the former study, the criterion was the same for
boys and girls and the reliability of the measurement regarding maturational rate was
higher. The relationship between these different criteria of physical maturity is,
188 CHAPTER 16
however, reasonably high (cf. e.g. Lindgren, 1978) indicating that, if a child is early
according to one criterion, he or she is probably also early according to the other
criteria. For boys the highest correlation coefficient is r = 0.79 between age at
pubertal stage IV and PHV-age. For girls the highest correlation is r = 0.63 between
pubertal stage III and PHV-age.
Also the mental ability tests used in both studies were administered in different
Grades. In the longitudinal study they were administered only in Grade 5 (11 - 12
years), while in the cross-sectional study these tests were administered in Grade 4
(10 - 11 years), Grade 6 (12 - 13 years) and Grade 8 (14 - 15 years). In addition a
new type of test measuring cognitive levels of thinking according to Piaget's theory
was administered in all three Grades in the cross-sectional study.
In spite of the differences in design the main results from both studies, however,
give a similar pattern. Verbal ability as measured by the test "Opposites" did not
differ between early and late maturers in Grade 4 (age 10 - 11 years) according to the
present study. In Grade 5 (11 - 12 years), however, according to the former
longitudinal study early maturing girls were performing better than late maturing
girls on this test. In Grade 6 (12 - 13 years) there were no differences between early
and late maturers; neither in Grade 8 (14 - 15 years) according to the present study.
Girls in Grade 8, however, performed better than boys.
Regarding logical-inductive ability as measured by the test "Letter groups" early
maturers were performing better than late maturers (especially the girls) in Grade 4
according to the present study. In addition SES-group I was performing better than
SES-groups II and III. However, according to the former longitudinal study, there
were in Grade 5 no differences between early and late maturers or between SES-
groups. In Grade 6 again, according to the present study, early maturers (and this
time especially the boys) were performing better than late maturers. In Grade 8 there
were no longer any differences between either pubertal stages or between SES-
groups per se.
The cognitive test "Similarities" did not reveal any differences between either
sex, pubertal stages or SES-groups in Grade 4. There was, however, in Grade 6 a
rather clear relationship between pubertal stage and test results for both boys and
girls showing continuously better results from prepubertal stage I to adult stage V.
There was though an interaction between sex and pubertal stage indicating a "dip" in
the trends at pubertal stage III for boys and pubertal stage IV for girls followed by a
catch-up at pubertal stage IV and V respectively. This interaction makes one wonder
whether the timing of PHV-age is in some way related to it. Boys were performing
lower than girls at pubertal stages I, II and III, but after pubertal stage III surpass the
girls at pubertal stages IV and V, that is to say at a time when boys have their
PHV -age. Results from the earlier longitudinal study showed that boys on a mental
arithmetic test scored lower than girls in Grades 5 and 6, but then caught up and
surpassed the girls in Grade 7 (13 - 14 years) when in general boys have their PHV-
age (Lindgren, 1979). In Grade 8 there were in the present study no significant
differences between pubertal stages regarding test scores on the test "Similarities".
There were, however, some interactions between; on one hand sex and pubertal stage
- on the other pubertal stage and SES-group, which were difficult to give
meaningful interpretations.
The results from the present study added to the results from the former
longitudinal study strengthen the impression that tempo of physical maturation
during puberty plays an important role when explaining the success of children's
performance at school, most evident from lO to 14 years of age. After the age of 14
years the direct effect of physical maturation seems to have a less important role - at
least as concerns mental ability and cognitive thinking - and the effects of sex and
socio-economic background take over. However, long-term effects might be
expected, if the late maturing children's self-esteem might have been negatively
affected during the pubertal period. A more detailed attempt to interpret the various
results within their social context as well as the educational implications have been
given by Lindgren (1988, 1995).
Acknowledgements. For very helpful comments on this paper, I thank my

husband Jim Tanner
5. REFERENCES
Anastasi, A., 1958, Differential Psychology. Individual and Group Differences. 3rd ed. (New York:
Mac Millan).
Boas, F., 1941, The relation between physical and mental development. Science, 93, 339-342.
Douglas, J.V.B., and Ross, J.M., 1964, Age at puberty related to educational ability, attainment and
school leaving age. Journal of Child Psychology and Psychiatry, 5, 185-196.
Emanuelsson, I., and Svensson, A., 1986, Does the level of Intelligence Decrease? A Comparison
between Thirteen-Year Olds Tested in 1961, 1966 and 1980. Scandinavian Journal of Educational
Research, 30, 25-37.
Epstein, H.T., 1974, Phrenoblysis. Special brain and mind growth periods. Developmental
Psychobiology, 7, 207-224.
Haernqvist, K., 1960, Individuella differenser och skoldifferenser. [Individual Differences and School
Differences] (Stockholm: SOU 1960: 13).
Husen, T., 1974, Talent, Equality and Meritocracy. Availability and Utilization of Talent. Plan Europe
2000. Project I: Educating Man for the 21st Century. Vol. 9. (The Hague: Martinus Nijhoff).
Inhelder, B., and Piaget, J., 1958, The Growth of Logical Thinking from Childhood to Adolescence
(London: Kegan Paul).
Johannesson, I., 1?74, How does puberty}nfluence school achievement. In Compte-rendu de la XIIe
Reunion des Equipes Chargees des Etudes sur la Croissance et Ie Deve10ppement de I'Enfant
Normal (Paris: Centre International de l'enfance) p. 243.
Kelly, A., 1978, Girls and Science. lEA Monograph Studies No.9. (Stockholm: Almqvist & Wiksell
International).
Kohen-Raz, R., 1977, Psychobiological Aspects of Cognitive Growth (New York: Academic Press).
Lindgren, G, 1976, Height, weight and menarche in Swedish urban schoolchildren in relation to socio-
economic and regional factors. Annals of Human Biology, 3, 501-528.
Lindgren, G., 1978, Growth of schoolchildren with early, average and late ages of peak height
velocity. Annals of Human Biology, 5, 253-267.
Lindgren, G., 1979, Peak velocities in height and mental performance. A longitudinal study of
schoolchildren aged 10-14 years. Annals of Human Biology, 6, 559-584.
Lindgren, G., 1988, Psycho-Social Aspects of Growth with Special regard to the relation between
PhysicaVPhysiological and MentaVCognitive Growth. Collegium Antropologicum, 1,47-66.
Lindgren, G., 1995, Socio-economic background, growth, educational outcome and health. In Essays
on Auxology, edited by R. Hauspie, G. Lindgren and F. Falkner. (Welwyn Garden City:
Castlemead Publications) p. 408.
Ljung, B.-O., 1965, The Adolescent Spurt in Mental Growth (Stockholm: Almqvist & Wiksell).
Maccoby, E.E., and Jacklin, C.N., 1974, The Psychology of Sex Differences (Stanford, California:
Stanford University Press).
Mitchell, J.R., 1980, Male Adolescents' Concern about a Physical Examination conducted by a Female.
Nursing Research, 29, 165-169.
Muuss, R.E.H., 1970, Adolescent development and the secular trend. Adolescence, 5, 267-284.
Nisbet, J.D., and Illsley, R., 1963, The influence of early puberty on test performance at the age of
eleven. British Journal of Educational Psychology, 33, 169-176.
Sandgren, B., 1974, Kreativ utveckling. En empirisk studie av kognitiv utveckling samt en kritisk analys
av intelligensbegreppet [Creative development. An empirical study of cognitive development and a
critical analysis of the concept intelligence] (Stockholm: Almqvist & Wiksell).
190 CHAPTER 16
SAS User's Guide: Statistics., 1985, Version S Edition, Chapter 20: The GLM Procedure. (Cary, N.C.:
SAS Institute Inc) p. 433.
Shuttleworth K.F., 1939, The physical and mental growth of girls and boys age six to nineteen in
relation to age at maximum growth. Monographs of the Society for Research in Child
Development, 4, No.3.
Tanner, 1.M., 1961, Education and Physical Growth. (London: University of London Press).
Tanner, 1.M., 1962, Growth at Adolescence. 2nd ed. (Oxford: Blackwell Scientific Publications).
Tanner, 1.M., 1989, Foetus into Man. 2nd ed. (Ware:: Castlemead Publications).
Tanner, 1.M., 1999, Retrospective: the growth and development of the Annals of Human Biology: a 2S-
year retrospective. Annals of Human Biology, 26, 3-18.
Westin-Lindgren, G., 1982, Achievement and mental ability of physically late and early maturing
schoolchildren related to their social background. 10urnal of Child Psychology and Psychiatry, 23,
407-420.
Westin-Lindgren, G., 1984, Physical and mental growth controlling for social background. In Human
Growth and Development edited by 1. Borms, R. Hauspie, A. Sand, C. Susanne, and M. Hebbelinck
(New York: Plenum Publishing Corporation) p. 70S.
APPENDIX: THE TEST "SIMILARITIES"
o
If you divide a circle like this ...
CD C£ CJD
... you can get halves
that are exactly alike.
o o--~©
If you divide an apple like this ...
... you will see that the halves
aren't exactly alike. But you can
still say that they are alike
O 0,
Compare an apple and a pear!
You can see that they are rather unlike L3 ttp'I-~\
each other. A
But they are also alike in many ways.
Many things are like that. For example, take a mushroom
and an umbrella. How do you think they are alike?
~ / , / A sled and a pair of skies can also be alike in many ways.
6tl/ / / What similarities can you find between them? Write some
down here: .. ........ .. ....... ........ ....... ..... ... ... ....... ........ ...... . .
You understand now that things can be alike in many ways. You can't say that ONE
such similarity is the only right one! There are many similarities, and ALL are
correct, each one in its own way.
Now, if you have several similarities and are going to choose between them, you
usually think that one is better than the others. When people choose between
similarities, they usually choose very different ones. But none can be wrong, since
all of the similarities are right.
Now, please tell me what YOU think is the best similarity between a bun Q and a
loaf of bread Ch
In the squares below, we have written down some similarities, which you can
choose from:
Best Worst
1 2 3 4 5 similarity similarity
can be are baked are light contain taste good
eaten in an oven brown yeast with butter
-- -
As you can see, we have numbered the similarities from 1 to 5. Now, choose first
the similarity you think is the BEST one! Write the figure above that similarity in
the square "Best similarity".
Then, choose the similarity you think is the WORST one! Write this figure above
that similarity in the square "Worst similarity".
On the next page, you will find a lot of items that are like this one. Choose the best
and the worst similarities for each item. Don't worry about what your classmate
chooses. Write only what YOU YOURSELF think is the best and the worst similarity.
192 CHAPTER 16
APPENDIX
Write your name, class and school here before you start to fill out this answering
sheet!
I Name--- ·I~ate --I Class I School
Best Worst
A car have give out you can are you can

A bus wheels exhaust ride in vehicles crash into
fumes them them
A banana taste good have are fruit you can are

An on picnics peels get them nutritious
orange in fruit-
shops
A saw are can hurt are have can cut

A knife tools you sharp handles things into
pieces
Juice are are are mostly are soon are

Soda pop refreshing drinks water gone bottled
A ferry are found they they make are have

A steamer on lakes float you sea- ships keels
sick
Salt consist of are are used are are easy

Sugar small white in food spices to mix up
grains
Boots are protect are can give have heels

Shoes available your feet footwear you
in many blisters
sizes
A violin you can are you must have you keep

A guitar play on musical be careful strings them in
them instru- with them cases
ments
A rifle are can kill have make a you can

A pistol weapons people triggers loud shoot with
sound them
A wild you can have are found need are

rose bush scratch leaves in the water plants
A culti- yourself gardens
vated rose on them
continue on the next page!

APPENDIX
Best Worst
An arm- have four have the you can are furni- are com-
chair legs same sit in them ture fortable
A sofa shape
A tele- can be can are square have an are

gram sent contain address messages
A letter bad news on them
A fann are found are have are good you can

A summer in the buildings windows to have in live in
house country summer them
vacation
A hawk frighten have are birds can fly are dark

An eagle other feathers brown
animals
A one are money you can are reaso- you have have pic-
dollar bill pay for nable them in tures of
A five things weekly your famous
dollar bill with them allowance wallet men on
them
A school- have a go to learn new are pupils can come

boy schedule school things too late
A school-
girl
A cap are round you can protect are head- are easy
A hat have them you gear to forget
on your
head
Chess you can are games have have a have

Checkers win at certain square "men" or
them rules playing pieces
board
A book are found give you have are fun to are

A news- in shops infor- printed read printed
paper mation letters matter
A radio send the can be have nice are look like

A tele- news turned off programs communi- boxes
vision for cations
children media
THE END
194 CHAPTER 16
APPENDIX
10 I
Grade 4: Girls
8 ...........
•
ti D Pub. stage I
,..0 Pub. stage II
~
6 ··········· Pub. stage III
~
d)
on
ro 4
ti
:>
<t:
2
2 3 4 5
Cognitive level
Grade 4: Boys
... 81·. ·. ·····1 D Pub. stage I
•
d)
.0
E
a 61-···· ····1 Pub. stage II
~
~ 4
d)
:>
<t:
2
0
0 2 3 4 5
Cognitive level
Figure 1. Average number of choices of different response alternatives (cognitive levels)

on the test "Similarities" in Grade 4 in relation to pubertal stages. Girls and Boys.
APPENDIX
Grade 6: Girls
16 ...........
•
~ 0 Pub. stage I
..0 Pub. stage II
~ 12 ...........
~ Pub. stage III
0
~
8 Pub. stage IV
r:!
C1.)
8 ··········· D Pub. stage V
;:-
<
4
0
0 2 3 4 5
Cognitive level
2
Grade 6: Boys
I-<
C1.)
D Pub. stage I
..0 • Pub. stage II
§ 12 ~ Pub. stage III
o
B Pub. stage IV
&C1.)
o Pub. stage V
;:-
<
0' 0 2 3 4 5
Cognitive level

196 CHAPTER 16
APPENDIX
16 l .- ......... __ ....... ...................................................... .................................. .
Grade 8: Girls
~
12 ~···········
D Pub. stage I
1
d 8 • ..........
•
~
a
Pub.
Pub.
Pub.
stage II
stage III
stage IV .......................................................................... ~
& D Pub. stage V
<~ 4 +·············· . . ... . .. . ........ . ... . ... _..... _....... .
0
0 2 3 4 5
Cognitive level
16
I
Grade 8: Boys
•
12 ...........
1-0
Q)
D Pub. stage I
.0 Pub. stage II
aE 8 ··········· ~
g
Pub. stage III
Pub. stage IV
&
Q)
;>
0 Pub. stage V
< 4
! !
0' 0 2 3 5
Cognitive level

APPENDIX
Table 1. Results concerning three-way ANOVA (GLM-procedure) regarding the verbal test
"Opposites" in relation to sex, pubertal stage and SES-groups in
Grade 4 (10-11 years):
N= 76
Source of Sum of squares df F Significance

variation level
Model 39.983 15 1.08 NS
Type III tests
Sex 1.158 1 0.47 NS
Pubertal stage 13.524 3 1.83 NS
SES-group 1.354 2 0.27 NS
Sex x pub. stage 0.106 0.04 NS
Pub. stage x SES-group 6.234 4 0.63 NS
Sex x pub. stage x SES- 0.690 4 0.07 NS
group
NS: not significant
N=81

variation level
Model 58.895 23 1.03 NS
Type III tests
Sex 6.717 1 2.71 NS
SES-group 11.416 2 2.30 NS
Sex x pub. stage 13.851 4 1.39 NS
Sex x pub. stage x SES- lO.738 4 1.08 NS
group
NS: not significant
198 CHAPTER 16
APPENDIX
N=84

variation level
Model 113.583 20 2.27 **
Type III tests
Sex 23.325 1 9.32 **
group
**: p < 0.01 NS: not significant
Table 4. Results concerning three-way ANOVA (GLM-procedure) regarding the logical

inductive test "Letter groups" in relation to sex, pubertal stage and SES-groups in
N= 76

variation level
Model 68.481 15 2.90 **
Type III tests
Sex 2.278 1 1.45 NS
Pubertal stage 13.520 3 2.86 *
SES-group 15.505 2 4.92 **
Sex x pub. stage 0.115 0.07 NS
group
**: p < 0.01 *: p < 0.05 NS: not significant
MENTAL ABILITY AND COGNITIVE TIllNKING 199
APPENDIX

N=81

. variation level
Model 174.222 23 2.04 *
Type III tests
Sex 4.937 1 1.33 NS
SES-group 11.259 2 1.51 NS
Sex x pub. stage 55.595 4 3.74 **
group

N=82

variation level
Model 112.107 20 1.87 *
Type III tests
Sex 10.175 1 3.40 NS
Sex x pub. stage x SES- 40.995 5 2.74 *
group
*: p < 0.05 NS: not significant
200 CHAPTER 16
APPENDIX
Table 7. Results concerning three-way ANOVA (GLM-procedure) regarding the cognitive

test "Similarities" in relation to sex, pubertal stage and SES-groups in
N=83

variation level
Model 40.258 15 0.90 NS
Type III tests
Sex 2.414 0.83 NS
group
NS: not significant

N=83

variation level
Model 94.476 23 1.37 NS
Type III tests
Sex 0.085 0.03 NS
Pubertal stage 39.183 4 3.26 *
Sex x pub. stage 36.482 4 3.03 *
group
*: p < 0.05 NS: not significant
APPENDIX

N=86

variation level
Model 91.484 26 2.01 *
Type III tests
Sex 0.000 0.00 NS
Sex x pub. stage 18.510 2 4.06 *
Pub. stage x SES-group 46.798 6 3.42 **
group
POPULATION DIFFERENCES IN
GROWTH
CHAPTER 17
HOW GENETIC ARE HUMAN BODY PROPORTIONS?
B. BOGINt, M. KAPELLt, M.1. VARELA SILVAt, A.B. ORDEN:j:, P.K.

SMITJIlj[, J. LOUCKY§,
tDepartment of Behavioral Sciences, University of Michigan-Dearborn, Dearborn, MI,

USA, :f:Centro de Investigaciones en Genetica Basica y Aplicada (CIGEBA), Universidad
Nacional de La Plata, Argentina, 'f[Department of Social Sciences, University of
Michigan, Dearborn, MI, USA, § Department of Anthropology, Western Washington
University, Bellingham, WA., USA
1. INTRODUCTION
Children tend to resemble their parents in stature, body proportions, body
composition, and rate of development. It may be assumed that barring the action of
obvious environmental influences on growth (such as chronic illness or long-term
malnutrition) these resemblances reflect the influence of genes that parents contribute
to their biological offspring. A study published by Prokopec and Lhotska (1989),
based on a sample of 81 boys and 78 girls, is an example of this view. The subjects,
all from Prague, were measured annually from birth to age 20 years. The Preece-
Baines growth curve was fit to the longitudinal data of each subject. From these
fitted curves for all the boys and girls, the three tallest, the three shortest, the three
slowest maturing, and the three fastest maturing of each sex were selected. None of
these extreme cases was known to have any major chronic or acute diseases. Neither
the subject's history of common childhood diseases, nor the occupation of the fathers
had an effect, positive or negative, on growth and development. In contrast, the mid-
parent height did predict the adult stature of offspring. Mid-parent height is the
average of the stature of the mother and the father. Inspection of the Preece-Baines
curves showed that tall or short stature at age 20 could be predicted from stature at
age four years. The positive impact of mid-parent stature on offspring growth and the
predictability of adult height from stature at age four are prima/acia evidence for the
role of heredity. Moreover, these findings attest to the early establishment of
individual patterns of growth and their stability over time.
Other well-known research strategies are used to demonstrate the genetic
determination of human growth. These strategies include the study of monozygotic
and dizygotic twins, correlations in growth between biological relatives (non-twins),
and the effects of genetic abnormalities on growth (reviewed in Bogin 1999). Studies
such as these lend support to the concept of a "genetic potential" for body size, body
composition and body proportions. The term "genetic potential" usually means that
every human being has a genetically determined upper limit to adult stature, the ratio
205
206 CHAPTER 17
of leg-length to sitting-height, and other anthropometric dimensions. It is further

assumed that an individual may achieve this "genetic potential" if the environment is
free of insults that delay or retard growth.
The origin of the idea of "genetic potential" in human development may be found
in the theory of epigenesis. In 1651 the English physician and anatomist William
Harvey hypothesised that the specialised tissues, organs, and structures of the
individual develop from unspecialised and undifferentiated cells in the ovum.
Experimental support for this hypothesis came only in 1759 with the research of the
German anatomist Kaspar Friedrich Wolff. He was able to show that the
development of the organs of the chick in the egg did indeed develop from
undifferentiated cells and tissues. Further research into embryology during the 19th
century came to view prenatal development as largely fixed, or predetermined, in
nature. With the rediscovery of Mendel's principles of heredity in 1900, the
mainstream view within developmental biology was that, "the basic potential nature
and organisation of the structures of the organism are determined by the genetic
constitution of the fertilised egg" (Roche and Boell 1999). The quoted material is
from a popular electronic encyclopaedia. Note that the authors of this quote link the
words "potential nature," "determined," and "genetic" to describe how developmental
biology operates. This is one more example of the way in which the concept of
"genetic potential" is ingrained in scientific and popular literature.
Some scholars extend the concept of "genetic potential" from the individual to
populations. Body-proportion differences between populations and ethnic groups are
well known, and are often used to exemplify the concept of "genetic potential."
Norgan (1998), for example states that, "The largest differences between ethnic
groups, when all are growing up in good environments, are those of shape rather
than size." Eveleth and Tanner (1990, p. 186) state, "These differences are certainly
genetic in origin .... " Both Norgan and Eveleth and Tanner reinforce these statements
with a good deal of compelling data from growth research conducted on diverse
human populations. The data are, however, circumstantial as no genetic basis for
body shape is provided in any of the studies cited. Indeed, there is no known basis for
the determination of human body proportions.
In addition to the use of circumstantial data to support the belief in a genetic
determination of body proportions, several authors use a series of somatotype-like
photographs of men of European or African origin. One version of these
photographs, from the Cambridge Encyclopedia of Human Growth and Development
(1998, p. 360), is reproduced here as Figure 1. The caption for this version of the
figure reads, "Comparison of European and African physiques, showing the relatively
longer legs of Africans." The implication of this statement, and the accompanying
article on body proportions, is that population differences in body shape are
genetically determined. Note that the photographs of each man are reduced or enlarged
in size so that all of the men have the same sitting height, as this makes the
differences in leg length more clear.
GENETICS OF BODY PROPORTIONS 207
Figure 1. Comparison of European and African Olympic athlete physiques showing the
relatively longer legs of Africans. 1n each of the four comparisons the European athlete is
on the left. This figure was re-imagined from a photograph in The Cambridge
Encyclopedia of Human Growth and Development (Ulijaszek 1998, p. 360). Professor
Tanner supplied the original photograph).
208 CHAPTER 17
The photographs were originally published in Tanner (1964), and are of Olympic
athletes. Olympic athletes are a highly selected group of people. The body shape of
Olympic athletes, including the ratio of leg length to sitting height, is strongly
correlated with the type of sport performed. Moreover, Olympic athletes constitute
only a very small segment of humanity, far smaller in number than would be
required for reliable and accurate statistical analysis. Olympic athletes, then, are in no
way representative of the general population in terms of biology or behaviour. Yet, a
version of the original photographs has been reprinted in several textbooks,
including both editions of Eveleth and Tanner's Worldwide Variation in Human
Growth (1976, 1990), and is used to demonstrate genetic differences in growth
potential between human populations.
The concept of "genetic potentials" in growth and body shape of human
populations is well entrenched in the field of human growth research. Some scholars
take the concept of "genetic potentials" as a given and use it without definition or
justification from research, that is, without reference to the scientific literature.
Interested readers may peruse recent issues of journals dealing with human biology,
physical anthropology, paediatric medicine, and related fields, where they will find
the concept of "genetic potential" used in this way.
The problem with this casual usage is that the assumption upon which it is
based is not true. Existing research shows that adult stature, body proportions, and
body composition are highly plastic. One of the clearest examples of the plasticity
of human body proportions comes from Tanner et at. 's (1982) studies of relative lag
length of Japanese children. In 1957 the relative leg length of adult Japanese was
significantly less than that of Northern Europeans, and it was assumed that this
difference was due mostly to genes. By 1977 the two ethnic groups had become
virtually identical in shape. This invalidated the genetic explanation, which was
replaced by an explanation favouring improvements in the physical and social
environment of Japan. Research conducted in China (Zhang and Huang 1988),
Argentina (Bolzan et at. 1993), Poland (Wolanski 1979), and Mexico (Gurri and
Dickinson 1990, Wolanski et at. 1993, Wolanski 1995, Siniarska 1995) shows that
the development of body proportions during the growing years is exquisitely
sensitive to the quality of the local environment (reviewed in Bogin 1999 and
discussed below). In these studies boys and girls of identical ethnicity and genetic
background are compared, and children from lower SES families are significantly
shorter in leg length than children of higher SES families.
In this article we re-examine the existing literature to see to what extent ethnic
and population differences in body-proportions which are assumed to be of a genetic
nature are, in fact, better explained by environmental plasticity in growth. We also
present the findings of a new study of body proportion change in the children of
Guatemalan immigrants now living in the United States. Finally, we make use of
experimental findings on the determinants of fluctuating asymmetry in insects to
develop an ecological model of body-proportion development.
2. WORLD-WIDE VARIATION IN HUMAN BODY PROPORTIONS - A NEW

ANALYSIS
The first and second editions of the book Worldwide Variation in Human Growth
(Eveleth and Tanner 1976, 1990) provide the most comprehensive compendia of
human growth studies. The authors present data for several measures of body
proportions. Here we re-analyse the data for relative sitting height, which is
calculated as [(sitting height/stature) x 100]. Relative sitting height is a ratio that
expresses the percent contribution of sitting height -- the length of the head, neck
and trunk -- to total stature (Lohman, Roche, and Martorell 1988). In practice, this
ratio is most often used as an indication of leg length differences between individuals
or populations. The reason for this is due to the cephalo-caudal gradient of growth,
which means that during the years of growth the head and trunk are always closer to
their final adult size relative to the legs. The sitting height ratio changes with age as
the legs begin to grow relatively faster than the head and trunk of the body.
From the data found in Eveleth and Tanner (1976, 1990) we analysed a total of
874 samples of relative sitting height of boys and girls, measured at different ages
and representing populations from many countries. The samples include people
living on most of the major continents and islands of the world. When discussing
the relative sitting height data, Eveleth and Tanner focus on only a few of these
samples, which they consider to be representative of four major geographic regions
of the world. The 1976 edition of their book presents samples from London
(Europe), from Ibadan, Nigeria (Africa), Hong Kong (Asia), and of Australian
Aborigine origin (Australia). In proportion to sitting height, the Australians had the
longest legs followed, in order by Africans, Europeans, and Asians. Expressed
quantitatively, "at a sitting height of 60 cm, for example, London boys have leg
lengths averaging 43 cm, Ibadan boys 53 cm and Australian Aborigine boys 61 cm"
(Eveleth and Tanner 1976 p. 229). In the 1990 edition of their book they analyse
data from Bergen (Europe), the People's Republic of China (Asia), "Afro-Americans"
(Africa) measured for the NHANES II study (a national study of health and nutrition
in the United States), and the Australian Aborigines from 1976 (Australia). The
1990 analysis finds almost the same differences in leg length as found in 1976.
By organising the samples in this way, Eveleth and Tanner are employing a
geographic "racial" typology (Garn and Coon 1955, Gam 1971). Garn (1971, p. 17-
18) states that, "To a large extent the geographical races of mankind coincide with
the major continents and ... may also be spread over major island chains, as is evident
in the Pacific today." Over the past century, many researchers have assumed that the
body proportion differences between geographic populations are explainable only in
terms of a genetic or "racial" model. The population difference in body proportions is
usually considered to be an adaptation to regional climates, for example, long arms
and legs in hot and humid regions and short appendages in cold regions. There is
considerable evidence to support the correlation between climate and body
proportions (Roberts 1953 is the classic study). However, no genes for the
determination of body proportions are known and a plausible genetic mechanism that
might account for population differences in body shape has never been formulated.
We use quote marks when writing the word "race" because that term implies a
biologically definable group of people. Definable biological distinctions between so-
called "races" do not exist at the genotypic or phenotypic level (Bogin 1993, Lasker
1999). "Race" does have some value as a shorthand term to categorise people into
groups that differ in economic opportunities, social organisation and resources, and
political power. In the United States those people who are classified as "white"
enjoy, on average, greater socio-economic opportunities and political power than
those classified as "black." These differences have an impact on social resources that
210 CHAPTER 17
often translate into biological effects, such as patterns of growth and development
(see Bogin 1999 for a more general and comprehensive discussion of "race" and
human growth).
The analysis presented by Eveleth and Tanner compares four samples from
different geographic areas. These areas conform to the four major world regions
defined by many of those researchers using a "racial" typology to divide humanity --
Africa, Asia, Australian, Europe. Eveleth and Tanner (1990, p. 188) attribute the
differences in relative sitting height and other body proportions to "racial
differences." Indeed, all of the growth data presented in both editions of Worldwide
Variation in Human Growth are organised into categories based upon geographical
"races." These "races" are Europeans, Africans, Asian, Australian Aborigines and
Pacific Islanders (including New Guineans and Maoris of New Zealand), Indo-
Mediterraneans (includes Algeria, Egypt, Ethiopia, India, Iran, Israeli Kurds and
Yemenites, Pakistan, Saudi Arabia, Turkey, and Yugoslavian gypsies), and a final
group of "inter-racial crosses", such as Cuban mulattos, South African Coloureds,
and Mexican Mestizos. The geographic "racial" classification is reinforced in their
work in that Eveleth and Tanner are more concerned with the putative geographic
origin of the samples than with the location at which they were measured. For
example, Black children measured in Tanzania and Black children measured in
Washington, DC are considered to be "African race." Native Canadians (Indians of
Canada), the Maya of Guatemala, and Chilean Indians are considered to be "Asian
race." Any samples of children in the United States with light skin colour and
claiming European origin are considered to be of the "European race."
2.1 Materials and methods of the re-analysis

We retain the geographical "race" scheme employed by Eveleth and Tanner in our re-
analysis of their data on relative sitting height and we add several new variables. In
all, we assigned an AGE, SEX, RACE, REGION, WORLD, SES, and SES x
WORLD variable to each sample. The numeric values assigned to sub-samples
within each variable are given in Table 1. These new variables are defined as follows.
AGE is the chronological age of a sample. Whole years are used to define AGE,
meaning that AGE 7 included all individuals between 7.00 and 7.99 years old and so
on for all other ages. Any individuals age 20 or older were considered adults and were
assigned the AGE score of 20. The SEX variable indicates the biological sex of an
individual, that is, either male or female.
RACE is the geographical "race" categorisation as found in Eveleth and Tanner
(1976, 1990). As stated above, this means that Tanzanians and fifth generation
Blacks living in the United States are "Africans." REGION is the geographic
location where each sample was measured regardless of "race" categorisation. This
means that Blacks and Whites measured in the United States and Canada are coded as
North American REGION, and any sample measured in Africa are African REGION.
The RACE variable assesses variability in body proportions by sorting people into
purported historical populations. The REGION variable provides an assessment of
the effect of current geographic location on body proportions of a sample.
Table 1. Variables used to re-analyse Eveleth and Tanner's data, the scores assigned to
each variable and the meaning of these scores.
Variable Score
AGE 1 to 18 - indicates the chronological age of a sample; e.g. a
score of 7 is for individuals between 7.00 and 7.99 years old. A
score of 20 indicates age 20 years old or older.
SEX 1 = male and 2 =female
RACE 1 = Origin in Australia, New Zealand, or Papua New Guinea, 2
= Origin in Africa, 3 = Origin in Europe, 4 = Origin in Asia,
includes Native Americans (North, Central, and South
American Natives).
REGION I = Europe, 2 = North America (Canada and United States), 3 =
Central America (Mexico to Panama), 4 = South America, 5 =
Africa, 6 = Asia, 7 = Indo-Mediterranean, 8 = Australia, New
Zealand, Papua New Guinea, and Pacific Islands
WORLD 1 = Industrialised/developed nations, 2 = former Soviet Union
and its satellites, 3 = non-industrialisedllesser developed
nations, 1.5 = Hong Kong, Singapore, Southern Europe
SES 1 = very high SES, 2 = high SES, 3 = middle SES in the
industrialised nations, 4 = urban poor/working poor, 5 = rural
poor
SES x WORLD Multiplication product of SES and WORLD
The WORLD variable indicates that a sample was measured in a first world,
second world, or third world country. First world countries in the Eveleth and Tanner
database are those of North-western Europe, the United States, Canada, Japan,
Australia, and New Zealand. Second world nations in the database are those that
belonged to the former Soviet Union and its satellite republics. Third world nations
in the database are those of Africa, Asia (except Japan), Latin America, the Pacific
Islands, New Guinea, and India. Some nations in Eveleth and Tanner's database were
most difficult to categorise. We assigned the value of 1.5 to Hong Kong, Singapore,
and southern Europe (Spain, Italy, and Greece) indicating that these nations were
halfway between first and second world in development indicators at the time the
growth data were collected. The WORLD variable is intended to assess the effect of
the general level of industrial development and standard of living on body
proportions.
The SES variable is an estimate of socio-economic status for each sample
varying from a high of 1 to a low of 5. Scores of 1 or 2 were assigned to samples
living in any country if the sample was described as "very high" or "high" SES. A
score of 3 is for middle SES samples in the industrialised nations of the first and
second world. A score of 4 is for urban poor and a score of 5 is for rural poor.
Parental occupation and parental education, especially of the father of a child, are
usually used to estimate SES. But, such information is not always known, and this
212 CHAPTER 17
makes the SES variable the most difficult to assign with accuracy. We tried to use
the SES information provided in each of the original publications cited by Eveleth
and Tanner. Most often, the authors of these papers describe their samples as "low
SES," "middle SES," or "high SES." However, not all of the original papers
included measures of SES for their sample. Also, we could not locate all of the
original papers. Our assignment of a SES score to each sample is, therefore, a
mixture of quantifiable information and qualitative assessment based on our best
estimate of the general socio-economic status of the sample.
The SES x WORLD variable is the multiplication product of the variables SES
and WORLD. We included this variable based on existing empirical research
showing an interaction between SES and WORLD on human growth (e.g.
Henneberg and Van Den Berg 1990, Bogin 1999). For example, low SES Maya
children growing up in the United States, a first world nation, are significantly taller
and heavier than low SES Maya children of the same ages growing up in their third
world homeland of Guatemala (Bogin and Loucky 1997).
For analysis, each of these variables was entered into a ridge regression model.
Ridge regression is used when multicollinearity exists between the independent
variables. A test for multicollinearity found that the variables WORLD and SES x
WORLD were highly correlated with the other variables. Ridge regression reduces
the effects of such high correlations by adjusting the regression model for more
reliable beta coefficients.
2.2 Results
A summary of the regression analysis is presented in Table 2. We find that AGE

explains 42% of the variance in relative sitting height of the 874 samples in the
Eveleth and Tanner database. As is well known, relative leg length increases from
infancy to adulthood. The geographical RACE organisation of the samples explains
3.6% of the variance and SES variable accounts for an additional 1.0% of the total
variance. The SEX variable explains only 0.002% of the variance, which is not
statistically significant. The other variables did not enter the regression model.
Table 2. Summary of the stepwise ridge regression model for relative sitting height
(lambda set at 0.10). R = 0.684, R2 = 0.465, F(4, 869) = 191.03, p < 0.000, Standard
error of estimate = 2.083
Step Multiple R Multiple R2 R2 change p-level

AGE 1 .648 .420 .420 .000
RACE 2 .675 .456 .036 .000
SES 3 .682 .466 .010 .000
SEX 4 .684 .468 .002 .120
2.3 Discussion
Eveleth and Tanner state that, "The ultimate size and shape that a child attains as an
adult is the result of a continuous interaction between genetical and environmental
influences during the whole period of growth" (1990, p.176). We agree with this
statement. Our re-analysis of Eveleth and Tanner's data helps to refine the role of
"genetical and environmental influences." Our regression analysis shows that
geographic RACE does contribute to variation in body proportions, but at a
relatively low level. Explaining only 3.6% of the total variation, RACE contributes
much less to the variability in body proportion than is assumed by many researchers
and the general public. The socio-economic variable SES, also contributes a small
and significant amount to the variance in body proportion. Indeed the relative
contribution of RACE and SES to the variance in body proportion are statistically
equal in our analysis, as a test for equality of slopes of their regression coefficients
(the unstandardised betas) shows no differences (t = 0.94). It is important to note
that, due to the imprecise definition of both RACE and SES in our re-analysis, our
results must be taken with caution.
To better understand variation in body proportions it is necessary to have more
accurate control of genetic and environmental variables. Some studies of body
proportion have such control. For example, Ramos-Rodriguez (1981) shows that
Mexico City children of middle SES have relatively longer legs than low SES
Mexican children from Oaxaca. She also shows that genetics cannot account for this
difference. Genetics played no role in the secular trend of Japanese body proportions
studied by Tanner et al. (1982). Prior to 1960, the Japanese were considered to be
both a short stature and short-legged "race." Between 1960 and 1977 the Japanese, on
average, gained 10 cm in stature and almost all of this increase was in leg length.
After 1977, the average height of Japanese men and women continued to increase,
but at a slower rate (Takaishi 1995), and both leg length and sitting height seem to
have increased at about the same rate, at least for young women (Rojo et al. 1981).
Since 1990 there is little evidence for further increase in stature. Today, Japanese
have, on average, virtually the same body proportions as many European
populations.
Several researchers working in Argentina, Poland and Mexico report similar
findings on the plasticity of body proportions. A team of researchers working in
small towns located in the Province of Buenos Aires measured 569 boys and girls,
seven to 13 years old, attending several schools (Bolzan et al. 1993). The sample
was divided into groups according to age, sex, and occupational status of the father.
Both boys and girls with fathers of lower occupational status (lower SES) were
shorter, and especially shorter in leg length, than subjects of higher family SES. In
Poland (Wolanski 1979), improvements in living conditions in towns and villages,
such as nutrition and health care, are associated with increases in leg length relative
to stature. Similar results are reported from a series of Mexican studies that find that
the body proportions of boys and girls vary according to relatively small differences
in family SES. The authors of these studies (Wolanski et al. 1993, Wolanski 1995,
Siniarska 1995) measured children living in the Yucatan region, including both
ethnically Maya and non-Maya populations. All of the families were of, generally,
low SES, but children from families of slightly better economic means were longer-
legged than children from lower SES families. Other studies from Mexico find even
more subtle influences of life style on body proportions. The leg-Iength-to-stature
proportions of women living in Chiapas Mexico, all of low SES, differ according to
socio-economic status, ecological region, and demography (Gurri and Dickinson
1990). All of the 421 women studied were 20 years old or older and all were of
214 CHAPTER 17
generally low SES. The sample was divided into four SES regions: 1) a region of
intensive export agriculture, 2) a region of cattle for meat production, 3) a region of
mixed agriculture and dairy herding, both for national consumption, and 4) a region
of subsistence agriculture. The authors also divided the sample into four ecological
regions: 1) Pacific coastal plain, 2) Sierra Madre mountains, 3) Central Valleys, and
4) Central Plateau. Finally a demographic division was made between those women
living in rural or urban areas. The authors found that 80% of the variance in stature
in the sample was due to SES region and that 20% of the variance was due to
ecological region. Almost all of the differences in stature within the sample were due
to variation in growth of the leg. Women from the SES regions of export agriculture
and the cattle raising area were the tallest. Women from the highland ecological
regions were, generally, the shortest. Similar findings are reported for children and
adults living in the Yucatan (Murguia et at. 1990, Dickinson et at. 1990). These
Mexican studies indicate that even within a generally low SES population, life style
differences exist and exert influence on body proportions.
3. RAPID CHANGE IN THE BODY PROPORTIONS OF MAYA CHILDREN
3.1 Materials and methods

Since 1992, we have studied the growth of samples of Maya children living in the
United States (Bogin and Loucky 1997). These children are the offspring of Maya
adults who immigrated from Guatemala to the United States, mostly from the late
1970s to the early 1990s. Some of the Maya children measured in our 1992 survey
were born in Guatemala or Mexico and some were born in the United States. Our
1992 survey shows that the children of Maya immigrant parents are significantly
taller than Maya children living in Guatemala (Bogin and Loucky 1997). We did not
measure body proportions in that study, but we hypothesised that an increase in
relative leg length accompanies the increase in stature.
A new study of the growth of Maya children in the United States began in 1999.
We measured the height, weight, and sitting height of 360 Maya children ages 6 to
12 years old, living in south central Los Angeles in February 1999 and in
Indiantown, Florida in March, 2000. With these data we computed the sitting height
ratio. We compared these newer data with the data for height ofthe 1992 samples and
with the data for height and sitting height ratios of a sample of Maya schoolchildren
living in rural Guatemala measured in 1998 (n = 1297). The Guatemalan data were
kindly supplied by Luis Rios of the Universidad Aut6noma de Madrid.
3.2 Results
Mean values for height and sitting height ratio for each sample are shown in Figures
2 and 3. Anthropometric reference data from the NHANES I and II surveys of the
United States (Frisancho 1990) are used as a baseline for comparison in each graph.
Analysis of variance (ANOV A) was used to evaluate differences between samples.
160
- Maya-Guat
150 • Maya-USA1992
- Maya-USA2000
E 140 - - - - - NHANES
E
.1::"
01)
130
.Qj
:r::
120
110
100
5 6 7 8 9 10 11 12 13
Age, years
Figure 2. Mean height of Maya children living in the United States measured in 1992
(Maya-USA 1992) and in 1999-2000 (Maya-USA 2000) compared with Maya children
living in Guatemala measured in 1998 (Maya-Guat) and the United States reference data
from NHANES 1 and 1/.
58
- Maya-Guat
57 - Maya-USA2000
- - - - - NHANES
0
.~
56
...
.1::01) 55
.Qj
..c:
01)
:::
54
·il
iZi 53
52
---
51
5 6 7 8 9 10 11 12 13
Age, years
Figure 3. Mean sitting height ratio of Maya children living in the United States measured
in 1999-2000 (Maya-USA 2000) compared with Maya children living in Guatemala
measured in 1998 (Maya-Guat) and the United States reference data NHANES 1 and 1/.
216 CHAPTER 17
After adjusting for the effect of age and sex, our results show (Figure 2) that the
Maya children of the 1999 and 2000 samples (abbreviated as "Maya-USA 2000") are
significantly taller (Figure 1) than the Maya samples measured in 1992 (abbreviated
as "Maya-USA 1992") and Maya children living in Guatemala (abbreviated as
"Maya-Guat"). The Maya living in the USA also have smaller sitting height ratios
than Maya living in Guatemala (Figure 3). All differences are significant at p < .01.
A smaller sitting height ratio generally indicates a child with relatively longer legs.
Compared with the NHANES references, all of Maya are shorter and have higher
sitting height ratios, i.e, relatively shorter legs.
3.3 Discussion
Our findings add further support to the literature on developmental plasticity in body
proportions. The results indicate that between 1992 and the present, there is a clear,
and positive, trend in growth of Maya children living in United States. The reasons
for this trend are likely due to improvements in the environment for growth. All
Maya in the USA have access to clean drinking water, health care, and education.
The Maya children living in Indiantown, Florida participate in school breakfast and
lunch programs. Nearly all of the Maya children have at least one parent with a
wage-earning job. All of this provides the Maya in the United States with a higher
standard of living than that found in Guatemala, as clean water, health care, education
and wage paying employment often do not exist for Maya living in rural Guatemala.
These health, economic, social, and nutritional changes are known to result in
greater stature. Our findings support the hypothesis that the increase in stature is due
mostly to relatively longer legs. We predict that the positive trend in height and
sitting height ratio seen between 1992 and 2000 will continue for the Maya in the
United States. Eventually heights and body proportions should achieve values that
approximate those for long-term residents of the United States.
4. CONCLUSION
In this article we have taken a critical look at the concept of genetic potential or
genetic determination of human body proportions. We find that population
differences in body proportions are influenced very little by genetic background, at
least as imputed from geographic, ethnic, or "racial" categorisations. Our findings
stand in sharp contrast to the opinion found in several of the most widely consulted
books on human growth. In a more general sense, the whole concept of "genetic
determination" in human growth is seriously flawed. That notion implies that the
flow of information about how any human trait is developed, be it height, body
fatness, personality, or intelligence, originates in the DNA and then unfolds into the
phenotype. Within this scenario, one may allow for a greater or lesser amount of
environmental influence on the phenotype, but the flow of information basically
begins with the DNA and moves one way.
The roles of DNA in human development are much more complex, and often
much less direct, than this. Genes do not directly cause growth and development.
Rather, the many proteins that genes produce, which are mediated by the endocrine
and neurological systems, regulate the expression of a genetically inherited pattern of
growth. The physical and social environment also mediates growth. In the case of
the Prague study discussed in the introduction of this article, it is important to note
that the parents and children were living in the same households and, therefore,
shared a very similar environment. It is too simplistic, therefore, to ascribe the
similarities in growth between parents and offspring to genes alone.
The interactions between genes, hormones, and the environment may flow in all
directions. A marvellous example of the interaction of genes, proteins, and the
endocrine system may be seen in the action of homeobox genes and Hox genes. The
description and elucidation of homeobox and Hox genes is one of the most important
advances of molecular biology of the past two decades. The homeobox is sequence of
180 DNA base pairs that codes for a 60 amino acid segment of a protein. First
discovered in the genome of the fruit fly, Drosophila, homeobox sequences are found
in all eukaryote organisms so far examined. These highly conservative DNA
sequences -- the same homeobox sequences are found in organisms as diverse as
hydra, nematodes, all arthropods (the group that includes insects) and all chordates
(the group that includes human beings) -- produce proteins that regulate the
expression of other genes, " ... and control various aspects of morphogenesis and cell
differentiation" (Mark et al. 1997, p. 421). Hox genes are a category of homeobox
genes that encode transcription factors (Holland and Garcia-Fernandez 1996), which
are proteins that initiate and regulate the conversion of the DNA code to the RNA
sequence that is used to make amino acid polypeptide chains.
In multicellular animals, homeobox genes act to delimit the relative positions of
body regions, for example the head, thorax and abdomen of insects, or the general
body plan and limb morphology of vertebrates. Homeobox genes seem to have their
greatest impact during the earliest stages of development. The proteins that
homeobox genes produce are needed to regulate the expression of other DNA to,
" ... sculpt the morphology of animal body plans and body parts" (Carroll 1995, p.
479). The DNA affected by homeobox proteins will, in tum, produce other proteins
that mediate cellular differentiation, growth, and development. These "down stream"
proteins do not act alone. Some of them must combine via a process called
molecular zipping before they have any effect on a given segment of DNA
(McKnight 1991). These and other proteins need an appropriate environment to have
any effect. In placental mammals, the biochemical environment of the egg cell and, a
bit later in time, of the mother's womb and the placenta, provide a host of factors
needed for growth, including nutrients and hormones.
Throughout life, the endocrine system often provides the necessary biochemical
environment for gene action. The human adolescent growth spurt, for example,
requires adequate amounts of two hormones, growth hormone and testosterone (boys)
or oestradiol (girls), to be secreted into the blood stream. Without these two
hormones the genes that regulate growth of skeletal, muscle, and adipose tissue will
not increase enough in activity to produce the growth spurt. The endocrine system
also responds to the influence of many environmental factors that affect human
development. Under-supply or over-supply of many nutrients, such as vitamin A or
D and folate, can have major effects on the growth and development of tissues,
organs, and the body as whole. These nutrients influence growth via their effect on
the regulation of DNA expression, protein synthesis, and hormonal regulation (see
Bogin 1999 for several additional examples). Because it is situated between the
action of genes and the external environment, the endocrine system serves as a
218 CHAPTER 17
mechanism that unifies the genes we inherit and the environments in which we live
to shape the pattern of growth of every human being.
Fluctuating asymmetry and human body proportions

Animal models are beginning to clarify the underlying biological processes of the
development of body proportions. Here we describe one experiment that manipulated
the size of forewings and hindwings in the buckeye butterfly (Precis coenia). The
authors of this study (Klingenberg and Nijhout 1998) were interested in the control
of fluctuating asymmetry (FA), that is, departures from the anticipated or normal
symmetry between the parts of growing organisms. In bilateral organisms one
measure of FA is to measure and compare the left and right sides of an individual. In
organisms with fore and hind limbs, wings, or other appendages it is possible to
measure FA by comparing the size of the distal and caudal segments. "Fluctuating
asymmetry... is increasingly being discussed as a potential indicator of
environmental stress" (Wilsey and Saloniemi 1999, p. 341). The working
hypothesis is higher level of FA are indicators of greater developmental instability
and greater environmental stress (Klingenberg and Nijhout 1998). The search for the
factors underlying FA, in both plants and animals, is an active area of developmental
biology research .
............. •
.." ....
4
• .. .'
'
~ '
gf
.~
3
.E
~
&.
~
.~
2
~
o
1+1--~--r-~--~--r-~--~--~~--4
25 35 45 55 65 75
dry weight of body (mg)
Figure 4. Regression of dry weight per forewing (the average of left and right sides) on dry
weight of the body for butterflies from which one, both, or no hindwing imaginal discs
had been removed (from Klingenberg and Nijhout 1998, plJ 36, with permission of the
author)
In their experiments with the butterflies, Klingenberg and Nijhout (1998, p.

1135) " ... removed the hindwing imaginal discs from one or both body sides of
caterpillars. Emerging butterflies were thus missing one or both hindwings, but had
heavier forewings, midlegs, and hindlegs than untreated controls. When only one
hind wing was removed, the forewing and hindleg on the treated side were heavier
than the untreated side." The results of these experiments for the growth of the
forewing are shown graphically in Figure 4. The authors note that the degree of
asymmetry and increase in weight of the growing forewings or legs diminished with
greater physical distance between the removed imaginal disc and the remaining
tissue.
Imaginal discs are embryonic structures of insects that provide the basic
information and the tissue structures to form body segments in adult organisms. The
developmental biology of imaginal discs are best known from the fruit fly
Drosophila, where the limbs and eyes arise from discrete imaginal discs. The
imaginal discs of Drosophila are epithelial sacs composed of a " ... columnar cell
monolayer covered by a squamous epithelium known as the peripodial membrane"
(Gibson and Schubiger 2000). During the fly's development the peripodial cells are
able to send biochemical information to the imaginal disc via microtubule
connections. Experiments by Gibson and Schubiger (2000) demonstrate that
biochemical communication between the peripodial cells and the imaginal discs
influences the control of growth and pattern formation of cells that will become the
eye and wing in the adult fly.
Based on the results of their experiments with butterflies, Klingenberg and
Nijhout (1998, p. 1135) concluded that, "Our findings are consistent with the
hypothesis that the growing imaginal discs compete for a haemolymph-borne
resource, such as a nutrient or growth factor. Such competition is a possible
mechanism for feedback interactions and may thus participate in the developmental
control of asymmetry." This hypothesis is strengthened by the work of Gibson and
Schubiger which finds that a physical connection, via microtubules, exists between
the imaginal disc and the peripodial cells, which can transmit information and
resources to the disc cells.
These findings, while based on work with insects, strongly indicate that the
development of human body proportions is more complex than can be accounted for
by any simple "racial" or genetic model. The known and hypothesised function of
homeobox and Hox genes in human development is compatible with the insect
models. This means that there is a role for genetic, environmental, and hormonal
interactions in the regulation of the growth of human body segments. Human body
proportions are important determinants of the biomechanical efficiency of movement
and performance in many daily activities and in sport. Body proportions are also
increasingly used as primary indicators of healthy growth and development (Bogin
1999). Accordingly, more sophisticated research on the regulation of development of
body proportions needs to be done. The new research must discontinue the use of
alleged geographic origin, skin colour, and other markers of "race." The new research
must also discontinue the uncritical use of the concept of genetic potential. The new
research must instead make use of better-defined, bioculturally valid, samples of
people and the independent variables that may influence their body proportions.
220 CHAPTER 17
5. REFERENCES
Bogin, B., 1993, Biocultural studies of ethnic groups. In Research Strategies in Human Biology, edited
by G. W. Lasker and CGN Mascie-Taylor (Cambridge: Cambridge University Press) pp. 33-61.
Bogin, B., 1999, Patterns of Human Growth, 2nd edition (Cambridge: Cambridge University Press).
Bogin, B., and Loucky, J., 1997, Plasticity, political economy, and physical growth status of Guatemala
Maya children living in the United States. American Journal of Physical Anthropology, 102, 17-32.
Bolzan, A. G., Guimarey, L. M., and Pucciarelli, H. M., 1993, Crecimiento y dismorfismo sexual de
escolares segun la ocupacion laboral paterna. Archivos Latinoamericanos de Nutricion, 43, 132-
38.
Carroll, S. B., 1995, Homeotic genes and the evolution of arthropods and chordates. Nature, 376, 479-
485.
Dickinson, F., Cervera, M., Murguia, R., and Uc, L., 1990, Growth, nutritioual status and environmental
change in Yucatan, Mexico. Studies in Human Ecology, 9, 135-149.
Eveleth, P. B., and Tanner, J. M., 1976, Worldwide Variation in Human Growth. (Cambridge:
Cambridge University Press).
Eveleth, P. B., and Tanner, J. M., 1990, Worldwide Variation in Human Growth, 2nd edition,
(Cambridge: Cambridge University Press).
Frisancho, A. R., 1990, Anthropometric Standards for the Assessment of Growth and Nutritional Status
(Ann Arbor: University of Michigan Press).
Gam, S. M., 1971,Human Races, 3rd edition (Springfield, Illinois: Charles C Thomas Publisher).
Gam, S. M., and Coon, C. S., 1955, On the number of races of Mankind. American Anthropologist, 57,
996-1001.
Gibson, M. C., and Schubiger, G., 2000, Peripodial cells regulate proliferation and patterning
Drosophila imaginal discs. Cell, 103, 343-350.
Gurri, F. D. and Dickinson F., 1990, Effects of socioeconomic, ecological, and demographic conditions
on the development of the extremities and the trunk: A case study with adult females from Chiapas.
Journal of Human Ecology, 1, 125-138.
Henneberg, M. and Van Den Berg, E. R., 1990, Test of socioeconomic causation of secular trend:
stature changes among favored and oppressed South Africans are parallel. American Journal of
Physical Anthropology, 83,459-465.
Hojo, T., Takemoto, R., and Shinoda, K., 1981, The secular unchangeability in relative sitting height of
female Kyushuites. Journal of the University of Occupational and Environmental Health, Japan, 3,
203-5
Holland, P. W. H. and Garcia-Fernandez, J., 1996, Hox genes and chordate evolution. Developmental
Biology, 173,382-95.
Klingenberg, C. P. and Nijhout, H. F., 1998, Competition among growing organs and developmental
control of morphological asymmetry. Proceedings of the Royal Society London, 265, 1135-1139.
Lasker, G. W., 1999, Race. Microsoft Encarta Encyclopedia 2000 (Redmond, Washington: Microsoft
Corporation).
Lohman, T. G., Roche, A. F., and Martorell, R., 1988, Anthropometric Standardization Reference
Manual (Champaign, Illinois: Human Kinetics Publishers).
Mark, M., Rijli, F. M., and Chambon, P., 1997, Homeobox genes in embryogenesis and pathogenesis.
Pediatric Research, 42, 421-29.
McKnight, S. L., 1991, Molecular zippers in gene regulation. Scientific American, 264, 54-64.
Murguia, R., Dickinson, F., Cervera, M., and Uc, L., 1990, Socio-economic activities, ecology and
somatic differences in Yucatan, Mexico. Studies in Human Ecology, 9, 111-134.
Norgan, N. G., 1998, Body-proportion differences. In: Cambridge Encyclopedia of Human Growth and
Development, edited by S. J. Ulijaszek, F.E. Johnston, and M. A. Preece (Cambridge University
Press), pp. 378-379.
Prokopec, M. and Lhotska, L., 1989, Growth analysis of marginal cases of normal variation. Anthrop.
Kozl., 32, 65-79.
Roberts, D. F., 1953, Body weight, race, and climate. American Journal of Physical Anthropology, 11,
533-558.
Ramos Rodriguez, R. M., 1981, EI significado del miembro superior: una hip6tesis a considerar. Bolotin
de Medicina de Hospital de Infant Mexicana, 38, 373-377.
Roche, J, and Boe!l, E. J., 1999, Development (biology), Microsoft Encarta Encyclopedia 2000
(Redmond, Washington: Microsoft Corporation).
Siniarska, A., 1995, Family environment and body build in adults of Yucatan (Mexico). American
Journal of Physical Anthropology, Supplement 20, 196.
Takaishi, M., 1995, Growth standards for Japanese children-an overview with special reference to
secular change in growth. In Essays on Auxology, edited by R. Hauspie, G. Lindgren, and F.
Falkner (Welwyn Garden City, England: Castiemead), pp. 302-311.
Tanner, J. M., Hayashi, T., Preece, M. A., and Cameron, N., 1982, Increase in length of leg relative to
trunk in Japanese children and adults from 1957 to 1977: comparison with British and with Japanese
Americans. Annals of Human Biology, 9, 411-23.
Tanner, J. M., 1964, Physique of the Olympic Athlete (London: George Allen & Unwin).
Ulijaszek, S.J., 1998, Between-population differences in human growth. In The Cambridge
Encyclopedia of Human Growth and Development edited by SJ. Ulijaszek, F.E. Johnston, and
M.A. Preece (Cambridge: Cambridge University Press) p. 361.
Wilsey, B. J. and I. Saloniemi, 1999, Leaf fluctuating asymmetry in tree-line birches (Betula pubescens
ssp. tortuosa): genetic or environmentally influenced? Oikos, 87, 341-346.
Wolanski, N., 1979, Parent-offspring similarity in body size and proportions. Studies in Human Ecology,
3,7-26.
Wolanski, N., 1995, Household and family as environment for child growth (Cross cultural studies in
Poland, Japan, South Korea and Mexico). In Human ecology: Progress Through Integrative
Perspectives, edited by S. D. Wright, D. E. Meeker, and R Griffore (Bar Harbor, Maine: The
Society for Human Ecology), pp. 140-152.
Wolanski, N., Dickinson, F., and Siniarska, A., 1993, Biological traits and living conditions of Maya
Indian and non-Maya girls from Merida, Mexico. International Journal of Anthropology, 8, 233-
246.
Zhang, X. and Huang, Z., 1988 The second national growth and development survey of children in
China, 1985: children 0 to 7 years. Annals of Human Biology, 15,289-306.
CHAPTER 18
THE SARSUNA-BARISHA LONGITUDINAL GROWTH

STUDY: DYNAMICS OF GROWTH
R.C. HAUSPIEt, P. DASGUPTA:\:

t Laboratory of Anthropogenetics, Free University of Brussels, Brussels, Belgium,
:t Anthropology and Human Genetics Unit, Indian Statistical Institute, Calcutta, India
1. INTRODUCTION
The Sarsuna-Barisha Longitudinal Growth Study was conducted between 1952
and 1966 by the late Professor S.R. Das with the assistance of Mr. D.P. Mukherjee,
Mrs. Lalita Bose and Mrs. Asha Das, and with the financial support of the
Anthropological Survey of India. The study was named after the two Bengali
villages, near Calcutta, where all children were examined. More than 500 boys and
girls, belonging to some two hundred families, were examined between birth and
adulthood over varying lengths of time (sometimes up to 14 years). The study is
most remarkable by the fact that it was one of the first longitudinal growth surveys
in India, including 22 anthropometric measurements all taken by one single
measurer, Professor S.R. Das himself, using standardised anthropometric techniques.
Professor Das got thorough training and practice in anthropometry under the able
guidance of Dr. E.C. Biichi - then a visiting scientist in the Anthropological Survey
ofIndia.
In the early 70's, the hand-written data sheets were handed over to Professor I.M.
Tanner for further processing under his direct care and guidance at the Department of
Growth and Development of the Institute of Child Health in London. It was in late
1978 that these data was operationalised by one of the authors of the present chapter
(RCH) who had received a Postgraduate NATO Research Fellowship to work for one
year on this material under the guidance of Professor I.M. Tanner, Dr. N. Cameron
and Dr. M.A. Preece in the Institute of Child Health. Since then several results have
been published in co-authorship with Professor. S.R. Das. In 1985, Professor Das
set the raw data of the Sarsuna-Barisha study available to the scientific world through
a special volume published by the Anthropological Survey of India, Calcutta (Das
1985).
The aim of this contribution is to highlight the major outcomes of the analyses
of this most valuable longitudinal growth data.
223
224 CHAPTER 18
2. THE SARSUNA-BARISHA SERIES

The Sarsuna-Barisha study was conducted over a period of 14 years, from 1952 to
1966, in two Bengali villages, Sarsuna and Barisha. These two contiguous villages
are situated in a semi-urban area, within the South Suburban Municipality of
Calcutta, at a distance of about 16 km from the heart of the city. The sample was
composed of three subcastes, namely the Rarhi Brahman (64%), Dakshin Rarhi
Kayastha (27%) and Vaidya (9%), which occupy the highest rungs of the Bengali
Hindu caste hierarchy in West Bengal. In spite of differences between families due to
income variations, they possessed a remarkable degree of homogeneity in respect of
the distribution of various blood group genes, or serum proteins, red cell enzymes
and haemoglobins (S.R. Das, unpublished data). For these reasons, all three classes
have been pooled together in the present analysis.
All subjects belonged to educated, middle-class families, who had lived in the area
for a number of generations. The mother's education status varied from primary to
graduated level and their occupation was primarily their own housekeeping. The
father's education status varied from school final to the graduate level with
postgraduate level in a few cases. Their occupations were in diverse areas:
administrative, secretarial, technical, scientific services in industrial, commercial,
governmental and public bodies. Some were engaged in teaching in schools and
undergraduate colleges.
Table 1. Frequency of families with different numbers of children
Number of sibs in 2 3 4 5 6 7 8 9
the family
Number 73 48 36 26 19 4 3 3 2
There were 303 boys and 260 girls, covering the age range 0.5 to 20 years. These
563 subjects belonged to 214 families; the frequency distribution of number of
sibships are shown in Table 1. Each child underwent a brief medical examination on
each measurement occasion and any with obvious illness or defect was omitted from
the record. Twenty-two anthropometric measurements were taken at six-monthly
intervals up to the age of five years, yearly from five to ten, six-monthly from lO to
14 and yearly thereafter. Only height will be presented in this paper. All children
were measured between 0700 and 0900 hours and all measurements, without
exception, were taken by S.R. Das throughout the entire period of the study.
Standard anthropometric instruments were used with the techniques described by
Martin (1928). For children below 2 - 3 years the measurement of stature was
substituted by the supine length from crown to heel, made on a special child-
measuring board. The true ages of the children were known in all cases as only
families having authentic birth records were included in the study; these records
included family horoscopes, notebooks and birth certificates. Target dates for
measurements were the birthdays or half-birthdays, and children were usually
measured within 15 days of the target. Over the whole material, attendance averaged
SARSUNA-BARISHA LONGITUDINAL GROWTH STUDY 225
six days late in boys and 7 days late in girls, and the standard deviation of the
difference between observational age and target age was about 14 days in both sexes.
Table 2 shows the number of children present in the survey for different lengths of
time. Table 3 shows the frequency distribution of the number of measurement
occasions per child.
Table 2. Number of children present in the study for different lengths of time (years)
Period <2 2.5-4 4.5-6 6.5-8 8.5-10 10.5-12 12.5-14

Boys 56 36 31 37 47 34 62
Girls 52 23 23 51 33 71 7
Table 3. Frequency distribution of the number of measurement occasions for individual

children.
Occasions 1-5 6-10 11-15 16-20 21

Boys 100 53 86 61 3
Girls 83 55 83 39
3. METHODS
Mean height for boys between 2 and 20 years of age was estimated by fitting Preece
Baines model I directly to the raw height-for-age data (Preece and Baines 1978). The
model has the following mathematical formulation:
2(hi - he)
y = hI eso(t-e) + e Sl (t-e)
with five function parameters hI, he, so' SI, (), while y is height in em and t is
age in years. Although the model is originally designed to describe individual
growth, it is also very apt to fit population average height growth (Jolicoeur and
Pontier 1993, Zemel and Johnston 1994). The model was fitted by an efficient non-
linear least-squares technique developed by (Powell 1969) for the Harwell Subroutine
Library. This method is a compromise of three different algorithms for minimising a
sum of squares, namely Newton-Raphson, steepest descent and Marquardt (Hauspie et
al. 1980b).
Centile lines for attained height were estimated by applying the LMS-method
(Cole and Green 1992) to the residuals obtained after fitting the Preece Baines model
to the raw data, i.e. to the deviations of the height measurements from the fitted
mean curve. We used a FORTRAN computer program, kindly set available by T.
Cole and slightly adapted for the present analysis. The LMS-method is a powerful
and compact technique designed to construct growth standards, allowing for
departures from normality while at the same time allowing the centiles to be
calculated from the mean and standard deviation (Cole 1989). The program provides
values of L (the Box-Cox power transformation), M (the mean value), and S (the
226 CHAPTER 18
coefficient of variation) for distinct values of age. These values of L, M and Scan
then be used to calculate the desired centiles ( Pi) according to the formula:
Pi = M (1 + LSz)1L
where zi is the normal equivalent deviate for the required centile.

Yearly increments in height were calculated for all height measurements 1 year
(± 0.15 year) apart. The centile distribution of yearly increments was estimated by
means of the LMS-method directly applied to the incremental data.
4. RESULTS
4.1 Growth charts

The centile distribution of height is shown in Figure 1 for boys and Figure 2 for
girls. This part of the analysis is limited to the age range 2 - 20 years where
measurements of standing height were available. The average pattern of growth is
very similar to the results obtained earlier by Hauspie et al. (1980) on the same data
applying methods for estimating mean height in mixed longitudinal data as described
by Patterson (1950) and Tanner (1951). In the present analysis we have treated the
data as if they were cross-sectionally but applying the Preece Baines model to
estimate the central tendency and the LMS-method to estimate the other centiles.
Mean height was 166.4 cm for boys at 20 years and 151.4 cm for girls at 18
years. As it can be seen from the distance charts there was no clear plateau reached by
the age of 20 years in boys and 18 years in girls. It seems that children continue to
grow slowly but gradually beyond that age. Hauspie et al. (1980) showed that the
average growth pattern of Sarsuna-Barisha children fluctuated between the 5th and
10th centile of the British references and that Indian children had a constantly lower
velocity than the Brits in pre-adolescent years, but similar velocity during the
adolescent growth spurt.
The centiles for yearly increments are shown in Figure 3 and reveal a classical
pattern of mean velocity for both boys and girls. From this analysis the mean age at
reaching maximum velocity can be estimated as 14.13 years in boys and 11.88 years
in girls. These values are, quite naturally very close to the ones obtained in the 1980
analysis of the data, using efficient methods for estimating mean values in mixed
longitudinal data: respectively 14 and 12 years and do not substantially differ from
corresponding values in Western populations. Although the 1980 analysis ofthe data
provided efficient estimates of the means, the age at maximum increment was, by
the nature of data structure, linked at one of the target ages while the present
methodological approach allows to estimate smoothly the age at maximum
increment in-between target ages. Figure 3 also shows that average growth velocity
did not really tend to zero at 20 years of age in boys and beyond 18 years in girls.
The centiles estimated by the LMS-method correctly show the strongly skewed
distribution of growth velocity in the second half of the adolescent growth spurt.
Because of this strong skewness, means and standard deviations of yearly increments
are no good estimates of central tendency and variation in growth velocity in this
period.
100
P97
P90
170 P75
P50
P25
100 P10
P3
100
140 .... -,--

, ---- .., ,
,-------~.-.--
· .
......,, ... -.. ,,, ..........,,--- .... -,.---
: ' : :
.: .... -,---
.....·, ....... , ..... ---.---
:
,
E 13) ,
u l ' 1 1
E .:-------~-- .. ----~--
Ol
'(j)
I 12) -/,./.. ;/...;('......,.......;....... ;...... ) , ....... .j...... ;....... ) , ....... .j...... ,.
, .. ..
:, :, :: ::
... ~ ... ---j-...... +... --.. ~ ....... ~ ....... i-' -.... -t··· ····t·-· .. ··~
" "
, , " "
::,I;,1;lllil
110 ····'f----
100
.. ,,<..../-, ....... j ....... , ....... , ....... , ...... ; ....... , ....... , ....... j ........ ~ ....... j .....•. ; .......• , ....... j....... ;
9) v-!//I/l······~···············;···· ...~ .......: ......... ··r·······~·······,········r·······:················r· .............. j
; i : ; i l i j i : i i : i :
·····j·······,········;·······:·······j········t·······)······T······,·······)·······,
._... 1. .. /. .. :/. ..... ~........ ~ ....... )....... , ........ , ..
00 hI.;;./····;·······,········,··· ... ;....... , .......~ .......... ···1······ .... ·· ···· .... ········· ..,.... ···: ...... ··...... ··r .. ·.... ;.... ···1
_... -r"'---:-.. ---T----··-~ .. -----: -...... y·····--r--···· "j"' --- -~-.--- .. -. --

-~
7O'L__L_ _L-~__~~L_-L_ _~-L__~~__~__L - _ L_ _~~_ _~_ _L_~
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 2)
Age, years
Figure 1. Centile distribution of height in boys from the Sarsuna-Barisha Longitudinal

Growth Study.
228 CHAPTER 18
100 ··········~····-· .. r·······l .-:-. ----°r-.-----J :
170
100
150
140
oE 133
1:
Ol
·iii
111::11;;1;111]
1;+ ,····!ttl;:I.
I 12)
110
100
7OL1__ ~~ __ ~~ __- L__ ~~ __ ~~ __- L__ ~-L __ ~~ __- L__ L--L~
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 2)
Age, years
Figure 2. Centile distribution of height in girls from the Sarsuna-Barisha Longitudinal

Growth Study.
12
11
n; 10
Q)
~
E 9
0
E 8
·wCl
.r; 7
c
(/) 6
"E
Q)
E 5
~
0 4
.~
>.
3
~
Q)
>- 2
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 2)
Age, yea rs
n;
Q)
~o
E
Cl
w
:~ ::~ir'flii I
7 ,P50.
.r;
.~
(/) 6 1P1O :
"E ~.... j'"'<;.: .. ," ..... ~ ...~ "',' ...... ~ ... j / " ....... ~. '\.! .. \:" .. \ .. ~.\ .. !. --:- ... -.--~.- ..
Q)
5
E
~
o 4
.~
>.
3
~
Q)
>- 2
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 2)
Age, years
Figure 3. Centile distribution of yearly increments for height in boys (upper part) and
girls (lower part) from the Sarsuna-Barisha Longitudinal Growth Study.
230 CHAPTER 18
4.2 Dynamics of individual growth patterns

The sample of 563 subjects comprised 63 boys and 42 girls with sufficient
longitudinal data for determining their individual growth pattern by means of curve
fitting techniques. Preece-Baines model I was also used to obtain a smooth growth
curve for those subjects and to estimate a set of biological parameters which
characterise the shape of the growth curves. Preece Baines model I has proven to be a
very robust model to describe the adolescent growth of longitudinal data in the age
range of 2 years to adulthood (Hauspie et at. 1980b, Hauspie and Chrzastek-Spruch
1999) Table 4 shows the means and SD's for the 5 function parameters and six
biological parameters.
Table 4. Means and standard deviations (SD) for the five function parameters of Preece-
Baines modell and for a series of biological variables in 63 boys and 42 girls.
Boys Girls
mean SD mean SD
hI 165.72 5.18 151.77 4.68
he 153.40 5.18 141.90 4.79
so 0.099 0.023 0.130 0.023
sl 1.297 0.235 1.260 0.262
() 14.61 0.975 12.90 0.927
Ti: Age at take-off, years 10.47 1.47 9.32 1.11
HI: Height at take-off, em 129.70 6.08 121.31 6.40
Vi: Velocity at take-off, em/year 3.90 0.82 4.57 0.64
12: Age at peak velocity, years 14.26 1.00 12.35 0.99
H2: Height at peak velocity, em 150.59 5.04 138.19 5.04
V2: Peak height velocity, em/year 8.73 1.27 7.16 1.15
On the basis of this subsample of 63 boys and 42 girls, we could conclude that
these Bengali boys reach age at peak velocity 1.9 years later than girls but at ages
fairly similar to those seen in Western populations. A comparison of the Sarsuna-
Barisha data with longitudinal data from the Harpenden Longitudinal Study and the
London group of the International Children's Centre Longitudinal Study (Preece,
Tanner and Whitehouse, unpublished data), analysed in the same way it seemed that
the adult British boys were taller than Indian ones by 10 cm and that this difference
was almost entirely due to a difference in prepubertal growth, mean height at take-off
being 138 cm in the British boys and 129 cm in the Indian ones. The mean ages at
take-off in the two samples were practically identical and the total gain during
adolescence in the boys differed by less than a centimetre. The adult height of British
girls exceeded that of Indian girls by approximately 11 cm, of which 7 cm was due
to a difference in prepubertal growth and 4 cm to a greater adolescent gain in British
girls.
Figure 4 shows the sex difference in height growth of Bengali adolescents on the
basis of mean-constant curves, i.e. the curves obtained by feeding the boys' and girls'
mean values of the 5 function parameters into the model (Hauspie 1989). Mean-
constant curves show the typical average growth pattern in the sample, and do not
suffer from the so-called phase-difference effect which smooths out the adolescent
growth spurt in cross-sectional growth data or in longitudinal growth data that is
treated cross-sectionally (unconditioned for tempo of growth). The velocity curve in
this graph is obtained by taking the mathematical first derivative of the distance
mean-constant curve.
170 Boys
100
150
E
u 140
1:
Ol
.(j)
1:Il
I
1~
110 10
9
100
8
00 7
6 <
CD
5"
5 Q.
4 ~
0
3 3
2
-<CD
~
,-- .•. r:=:---' '0

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 2l
Age, years
Figure 4. Sex dimorphism in growth of Bengali children. Preece-Baines model I mean-

constant curves.
According to Tanner et al. (1976) the sex dimorphism in adult stature of the
Indian boys can be decomposed into three additive parts:
• the difference in adolescent gain between boys and girls,
• the difference in pre-pubertal growth, measured as the size difference between boys
and girls at the girls' take-off,
232 CHAPTER 18
• the amount of growth achieved by the boys between the girls' and the boys' take-
off.
On a slightly different subsample, Hauspie et al. (1985) found that the sex
difference in adult height was greater in Indians than in British children (14.2 cm
compared to 12.0 cm). When expressed in percentage of the boys' adult height the
difference is even more pronounced in Indians than in their British counterparts. The
contribution of the difference in adolescent growth to the sex difference in adult
height was smaller in the Brits (2.0 cm) than in the Indians (5.2 cm). On the other
hand the contribution of the later onset of the pubertal growth spurt in boys was
bigger in Great Britain (7.9 cm) than in Indians (6.3). The sex difference in pre-
pubertal growth were fairly similar in British and in Indian children (respectively 2.1
and 2.7 cm). So it seemed that the greater sex difference in adult height among
Indians, compared to British children, was mainly due to a relatively greater
adolescent gain in Indian boys than in Indian girls.
4.3 Family data analysis

Of the 63 boys and 42 girls, for whom growth curves could be obtained by means of
curve fitting techniques, 45 were singletons and the other 60 were sibs, distributed
among 25 families with numbers of children varying from 2 to 5 (see Table 5).
Table 5. Distribution of children (Boys + Girls) among families
Singletons Sibships
------------------------------
Number of children per family 1 2 3 4 5
Number offamilies 45 19 3 2 1
A component of variance analysis was used to estimate the degree of resemblance

of growth patterns within families. The sample variation of an attribute is
decomposed into the within-family variation and the variation due to random effects
differing from family to family. The calculations were made according to Smith's
(1980) method, which allows the information derived from singleton families to be
incorporated with due weight.
The variables involved were the biological parameters derived from the fitted
curves: adult height, age, height and velocity at take-off and at peak height velocity.
The boys' and girls' values for these biological parameters were pooled after
standardisation for sex. The population against which the standardisation was done
consisted of (1) the 25 boy and 20 girl singletons of the sample plus (2) one
randomly chosen subject of each sex (where available) per family from the sibling
sample (22 boys and 15 girls).
The results of the component of variance analysis on the 105 subjects distributed
among 70 families are shown in Table 6. The last column gives the percentages of
total variation of the characteristic which is due to differences between families,
additional to the variation seen amongst groups of sibs.
The results indicate that the timing and intensity of the adolescent growth spurt
tended to be similar among sibs of the same family. Thirty three percent of the
variation in peak height velocity and 26% of the variation in peak height velocity
could be attributed to differences between families. Twenty two percent of the
variation in age at take-off could also be so attributed. Unfortunately the standard
errors of these percentages were quite high, so that only peak height velocity reached
twice its SE.
Table 6. Results of the component of variance analysis of growth curve parameters of 105
subjects (boys and girls pooled together after standardisation for sex) distributed among
70 families.
Mean square Mean square % variation due

to differences
within family between family
between families
Adult height 0.79 ±0.17 0.17 ±0.16 18 ±16
Age at take-off 0.77 ±0.16 0.22 ±0.16 22 ±lS
Height at take-off 0.8S ±0.18 O.lS ±0.16 IS ±16
Velocity at take-off 0.80 ±0.16 O.OS ±0.13 6 ±lS
Age at peak velocity 0.76 ±0.17 0.27 ±0.17 26 ±lS
Height at peak velocity 0.88±0.19 0.16 ±0.16 IS ±16
Peak height velocity 0.61 ±0.13 0.31 ±O.lS 33 ±14
The percentage of the variance of adult height due to between-family variation

was surprisingly small (18 ±16). Hauspie et at. (1982) were able to analyse this
parameter in a larger subsample of the Sarsuna-Barisha data since adult height was
reached in many more subjects than those who had sufficient data to yield successful
curve fits. An analysis of variance performed on the standardised values of adult
height of 22S boys and girls distributed among lOS families (maximum sibship 7)
gave a value of 49 ±7 percent of variation due to between family variance. In this
larger group it was also possible to study the sexes separately. For 121 boys
distributed among 81 families (maximum sibship 4) the between-family contribution
was 42 ±13% and for 104 girls distributed among 71 families (maximum sibship 4)
the between-family variance accounted for 38 ±14% of the total variation. These
three estimates of between-family variance were just within the upper 9S%
confidence limits of the estimate obtained from the smaller sample described earlier.
This analysis of longitudinal data on height growth in 70 Bengali families, of
which 2S contained siblings, showed that siblings resemble each other to a
considerable degree both in the time at which the adolescent growth spurt occurs and
in its intensity. The proportion of the total population variance due to variance
between families, which measures sibling similarity, was 33% for peak height
velocity (derived from the fitted curves) compared with 38% and 42% for adult height
in girls and boys of a larger sample. The percentage values for ages at take-off and
peak velocity were scarcely smaller. The standard errors of these percentages are
admittedly large, due to the sample being small. The between-family variance is a
measure of what Boas called the family-line reduction in variance (Boas 1916, Tanner
234 CHAPTER 18
1959). How much of this reduction in variance is due to genes and how much to
environment cannot be estimated from such family data. It reflects not only the
various and complicated interactions of genes and environment, but also the mating
structure of the population, especially of inbreeding.
4.4 Short-term growth in height, sitting height and biacromial diameter

This part of the analysis of the Sarsuna-Barisha material consists of a study of the
short-term variations in growth velocity of three different segments of the body:
height, sitting height and biacromial diameter. Several studies have shown that most
children show a small rise in growth velocity during childhood, the pre--pubertal
growth spurt or mid-childhood spurt (Tanner and Cameron 1980, Meredith 1981,
Prader 1982) and more recently Butler et at. (1989) suggested the presence of a series
of small spurts during childhood. In a longitudinal analysis of 59 girls measured at 3
and 6 monthly intervals, Hauspie and Chrzastek-Spruch (1999) found that only 7%
of the subjects had no pre-pubertal spurt at all, but that most had between 1 and 3
such spurts. In a longitudinal study of monozygotic and dizygotic twin data from
Wroclaw (Poland), Hauspie et al. (1994) found evidence that there is a genetic
component in the short-term variations of growth velocity for height.
In the present study we adopted the same methodology as Hauspie et at. (1994)
for the analysis of twin data, to study whether the short-term variations in height
velocity were, within individuals, also reflected in the pattern of growth of other
body dimensions such as sitting height and biacromial diameter. The methodological
approach essentially consists in fitting Preece Baines model I to the growth data of
those three traits and to use the tiny deviations (or residuals) of the measurements
from the fitted curves as a proxy for short-term variations in growth velocity.
Comparison of the within-subject resemblance in the patterns of residuals of the
three traits was done by calculating the average Euclidean distance coefficients for the
three traits, two by two. See Hauspie and Das (1995) for the details of the
methodology. Table 7 shows the means, SD's and the results of the t-Student test of
the Euclidean distances between the traits two-by-two after standardisation.
Table 7. Means and standard deviations (SD) of the standardised Euclidean distances of the
residuals (N = 59)
Mean SD Prob. one-tailed t-

Student test
Height versus sitting height -1.521 1.086 <0.001
Height versus biacromial diameter -0.533 1.254 <0.001
Sitting height versus biacrom. diam. -0.439 1.538 0.016
The mean values are well below zero. comparison of those mean values with the
expected value of zero (under random conditions) revealed significance for all three
combinations of traits. Since the Euclidean distance is a dissimilarity coefficient,
this means that the pattern of residuals of the three considered traits are significantly
more similar to each other than what would be expected if only random effects were
in play, most likely reflecting similarity in the age-related short-term variations in

growth velocity of the three considered traits, i.e. a reflection of synchronised
changes in growth velocity of these traits. The results also indicate a greater
similarity between the patterns of height - sitting height, than between height -
biacromial diameter and sitting height - biacromial diameter. Possibly, there is a
greater similarity in short-term variations in growth rate between vertical body
dimensions (height and sitting height) than there is between the vertical and the
horizontal body dimensions like biacromial diameter for instance.
5. DISCUSSION
The Sarsuna-Barisha series has proven to be a very useful source of information to
study variability in human growth patterns. The present overview highlights the
major outcomes of the analysis of this data.
• The charts for growth and growth velocity in height have been useful references
for the growth status of middle class Bengali children but their validity should be
checked in the light of a possible secular trend.
• The analysis of individual growth patterns by means of growth curve fitting has
thrown light on the dynamics of growth in an Indian population. It has shown
that the interrelation between various aspects of the human growth curve and the
dynamics of sexual dimorphism in the growth pattern is quite different between
Indian and Western children.
• The familial structure of the Sarsuna-Barisha data has allowed to examine family
resemblance in various aspects of the pattern of growth, an aspect of the study of
growth which is generally very difficult to tackle because of the lack of proper
material.
• Finally, the comparison of the tiny changes in the patterns growth of three body
dimensions has revealed that short-term variations in growth rate seem to be fairly
synchronised in traits such as height, sitting height and shoulder width.
Acknowledgements. The project was conducted and financed by the

Anthropological Survey of India, Government of India. We wish to thank Dr. B.S.
Guha, then Director of the Survey, for his continuous encouragement and Drs. D.P.
Mukherjee, Lalita Bose, Asha Das and E.C. Biichi for their help.
Analysis of these data was made possible thanks to a NATO Research Fellowship
awarded to one of the authors (RCH). We are also grateful to the Belgian National
Science Foundation for having allowed RCH to work abroad. Many thanks go to Dr.
N. Cameron and Dr. B. Carter of the Institute of Child Health, University of
London, for their helpful comments and suggestions.
6. REFERENCES
Boas, F., 1916, On the variety of lines of descent represented in a population. American Anthropologist,
18: 1-9.
Butler, G.E., McKie, M., and Ratcliffe, S.G., 1989, An analysis of the phases of mid-childhood growth
by synchronization of growth spurts. In Auxology 88: Perspectives in the science of growth and
development edited by J.M. Tanner (Niigata-Shi: Nishimur; London: Smith-Gordon), p. 77-84.
236 CHAPTER 18
Cole, T., 1989, Using the LMS method to measure skewness in the NCHS and Dutch national height
standards. Annals of Human Biology, 16: 407-419.
Cole, TJ., and Green, PJ., 1992, Smoothing reference centile curves: the LMS method and penalised
likelihood. Statistics in Medicine, II: 1305-1319.
Das, S.R, 1985, Mixed-longitudinal growth data for 22 measures - The Sarsuna-Barisha series, India.
Volume I boys and Volume 2 girls. Anthropological Survey of India, Calcutta.
Hauspie, R, Das, S.R., Preece, M.A., Tanner, 1.M., and Susanne, e., 1985, Decomposition of sexual
dimorphism in adult size of height, sitting height, shoulder width and hip width in a British and West
Bengal sample. In Human Sexual Dimorphism edited by 1. Ghesquiere, RD. Martin, and F.
Newcombe (Taylor & Francis: London), p. 207-215.
Hauspie, R.C., 1989, Mathematical models for the study of individual growth patterns. Revue
d'Epidemiologie et de Sante Publique, 37: 461-476.
Hauspie, RC., and Chrzastek-Spruch, 1999, Growth models: possibilities and limitations. In Human
Growth in Context edited by F.E. 10hnston and B. Zemel (London: Smith-Gordon) p. 15-24.
Hauspie, Re., and Das, S.R, 1995, Short-term variations in growth rate of height, sitting height and
biacromial diameter in Bengali children. In Essays on Auxology presented to lames Mourilyan
Tanner edited by R Hauspie, G. Lindgren, and F. Falkner (Welwyn Garden City: Castlemead
Publications), p. 260-268.
Hauspie, Re., Bergman, P., Bielicki, T., and Susanne, e., 1994, Genetic variance in the pattern of the
growth curve for height: a longitudinal analysis of male twins. Annals of Human Biology, 21: 347-
362.
Hauspie, Re., Das, S.R, Preece, M.A., and Tanner, 1.M., 1980. A longitudinal study of the growth in
height of boys and girls of West Bengal (India) aged six months to 20 years.
Hauspie, Re., Das, S.R, Preece, M.A., and Tanner, 1.M., 1982, Degree of resemblance of the pattern
of growth among sibs in families of West Bengal (India). Annals of Human Biology, 9: 171-174.
Hauspie, Re., Wachholder, A., Baron, G., Cantraine, F., Susanne, C., Graffar, M., 1980, A
comparative study of the fit of four different functions to longitudinal data of growth in height of
Belgian girls. Annals of Human Biology, 7: 347-358.
10licoeur, P., and Pontier, 1., 1993, Peut-on utiliser des modetes de croissance longitudinale dans
l'analyse de donnees transversales? Biometrie et Praximetrie, 33: 33-44.
Martin, R, 1928, Lehrbuch der Anthropologie. Erster Band-Somatologie (lena: Gustav Verlag).
Meredith, H.V., 1981, An addendum on the presence and absence of a mid-childhood spurt in somatic
dimensions. Annals of Human Biology, 8: 473-476.
Patterson, H.D., 1950, Sampling on successive occasions with partial replacement of units. 10urnal of
the Royal Statistical Society Bulletin, 12: 241-255.
Powell, MJ.D., 1969, Harwell Subroutine Library. Routine VA05A, AERE.
Prader, A., 1982, Biomedical and endocrinological aspects of normal growth and development. In
Human Growth and Development edited by 1. Borms, R Hauspie, A. Sand, e. Susanne, and M.
Hebbelinck (New York: Plenum Press), p. I -22.
Preece, M.A., and Baines, M.l., 1978, A new family of mathematical models describing the human
growth curve. Annals of Human Biology, 5: 1-24.
Smith, C.A.B., 1980, Estimating genetic correlations. Annals of Human Genetics, 43: 265-284.
Tanner, 1.M., 1951, Some notes on the reporting of growth data. Human Biology, 23: 93-159.
Tanner, 1.M., 1959, Boas' contributions to knowledge of human growth and form. In The Anthropology
of Franz Boas: Essays on the Centennial of his Birth edited by W. Goldschmidt. Memoirs of the
American Anthropological Association, nO 89. American Anthropologist, 61: 76-111.
Tanner, 1.M., and Cameron, N., 1980, Investigation of the mid-growth spurt in height, weight and limb
circumferences in single-year velocity data from the London 1966-67 growth survey. Annals of
Human Biology, 7: 565-577.
Tanner, 1.M., Whitehouse, RH., Marubini, E., and Resele, L.F., 1976, The adolescent growth spurt of
boys and girls of the Harpenden Study. Annals of Human Biology, 3: 109-126.
Zemel, B.S., 10hnston, F.E., 1994, Application of the Preece-Baines growth model to cross-sectional
data: problems of validity and interpretation. American 10urnal of Human Biology, 6: 563-570.
CHAPTER 19
LINEAR COMPONENTS OF GROWTH AMONG

RURAL INDIAN CHILDREN
S.RAO
Biometry & Nutrition Group, Agharkar Research Institute, Pune, India
1. INTRODUCTION
I consider it a great honour to be invited to write an article for a Commemorative

Volume of Auxology being published in memory of Prof. S.R. Das, a noted Human
Auxologist and Human Geneticist of India. I would therefore, like to pay my
respectful homage to this great man at the outset. I have been working on human
growth in particular on adolescent growth of rural Indian children for the last decade.
Since the data on linear components of growth in relation to stature growth is scarce
in Indian Auxological literature, it was thought to be befitting for this special
Volume.
Although growth is known to be influenced by genetic as well as environmental
factors, the latter are of greater importance. For, it is only under optimal levels of
environmental conditions that genetic factors become effective. Rural Indian children
living in poor environmental conditions reveal varying degrees of wasting and
stunting in pre-school age itself. Adolescence is considered to be a 'second
opportunity' as it facilitates catch-up in height, and therefore, investigations on
adolescent growth patterns assume significance.
Growth during adolescence contributes significantly to attainment of final
anthropometric measurements of an individual. Adolescence witnesses rapid changes
in most physical measurements but stature growth is more pronounced. Adolescence
is thus, characterised by spurt in height, and therefore changes in other measurements
in relation to spurt in height provide useful information.
Age at maximum growth in body dimension provides important information
about the timing of adolescent growth in general and about the individuality of
physical growth (Satake, Kikuta and Ozaki 1993). Children pass through the
adolescent phase of accelerated growth at widely different chronological ages. As such
marked differences are observed between children of same age-sex, with regard to age
at which the spurt begins and ends, as also in magnitude of the peak height velocity.
Large between-individual variations in adolescent growth performance among
children from the same community are likely to be due to the fact that children enter
adolescence with differing nutritional status determined by the past experiences in
237
238 CHAPTER 19
life. The present paper therefore, examines the effect of undernutrition on the growth
spurt as well as on other linear components of growth.
2. SUBJECTS AND METHODS
2.1 Subjects
A longitudinal adolescent growth study was undertaken in seven villages around
Pune during 1992-96. All children in the age group 9-18 year from these villages
were considered for the study. Most children were contacted in school while
remaining were recruited through house-to-house visit and an attempt was made to
contact a maximum number of study children in each subsequent round in order to
reduce the loss in follow-up. The majority (79%) of parents had cultivation as major
occupation, but most were engaged in subsistence farming due to poor land holdings.
People generally consumed vegetarian diets consisting of cereals, pulses and
vegetables, and the consumption of milk, meat, fruits was negligible. None of the
villages had safe drinking water supplies and lacked adequate hygiene and sanitary
conditions. Body weight, height, sitting height, foot length, shoulder width were
measured by a trained team of researchers once in every six months.
2.2 Measurements
Measurements were considered one year (±1 week) apart and all available
observations were used in the analysis. Weight (WT) was measured up to 20 g using
an electronic weighing balance (ATCO, India) while height (HT) was measured up to
0.1 cm using stadiometer (UNA, India). The same equipment was used to measure
sitting height (SHT), while leg length (LEGLN) was estimated by subtracting
sitting height from stature. Foot length (FLN) was measured with the help of a
specially designed board fixed with a fine metal ruler while shoulder width (SW) was
measured with the help of a measuring tape. All the measurements were recorded by
trained investigators.
3. RESULTS AND DISCUSSION
3.1 Stature growth

Data on 673 boys, collected for the period of three years, is discussed here for this
purpose. Mean attained values of weight and height using all cumulative annual
measurements (n = 1785) are given in Table 1 and compared with NCHS (Figure 1).
It can be observed that rural children are closer to 3rd percentile values only up to the
age 12-13 years, but then fall below it during adolescence, and the deviation increased
as adolescent growth advanced. Height velocity has special significance in the
context of adolescent growth as onset of growth spurt is characterised by the rapid
increase (almost doubles) in the height velocity. Cumulative data on annual
increments in weight and height was therefore, used to develop velocity curves
GROWTH IN RURAL INDIAN CIDLDREN 239
(Figure 2) by age. The velocity curve for height shows a shift to the right as
compared to the NCHS velocity curve for healthy children. The maximum
increments in weight (4.65 kg/year) and height (6.4 cm/year) occurred around the age
of 14+ years, which is higher than that reported for well-to-do Indian children
(Vijayaraghavan, Gowrinath and Indubhushan 1974).
70l • P50
180
.• ' P50
Weight 170J Height
60
160
50
eo 8u 150
~
•....40
..<::
eo
'Q) ~140
0)
~30 ::t:
130
20
120
10 110
01 100
8 10 12 14 16 18 20 8 10 12 14 16 18 20
Age, years Age, years
Figure 1. Mean weight and height of rural boys by chronological age compared to the 3rd
and 50th NCHS centiles
Table 1. Mean and SD afweight and height· rural boys
weight, kg height, cm
Age, years n mean SD mean SD
8+ 86 18.98 ** 1.9 118.2 ** 5.5
9+ 190 20.60 ** 2.3 122.8 ** 5.3
10+ 259 22.59 ** 2.7 127.7 ** 5.6
11+ 252 24.70 ** 3.5 132.0 ** 6.3
12+ 208 27.65 ** 4.6 137.4 ** 7.5
13+ 193 30.37 ** 6.6 142.1 ** 8.3
14+ 195 33.46 ** 6.1 147.7 ** 8.4
15+ 163 37.37 ** 6.5 152.7 ** 8.0
16+ 122 40.68 ** 6.1 157.0 ** 7.9
17+ 76 43.74 ** 6.0 160.0 * 7.3
18+ 33 45.59 5.8 161.2 7.5
19+ 8 50.24 7.0 164.7 5.3
*: p < .05 **: p < .01
240 CHAPTER 19
The difference is not so much in actual peak height velocity, but the age at which
peak height velocity occurs is significantly (p < 0.05) different. Adolescent growth
of rural boys is thus delayed by 1-2 years in comparison to healthy children. The
velocity curves also show higher height gains compared to healthy boys in later
ages. Thus, slow but gradual height gains as late as 18+ and 19+ years not only
prolong the adolescent growth span, but are indicative of catch-up growth in rural
Indian children.
7 8
Weight velocity Height velocity
6 7
a
Q)
a6
<-OIl 5 ~
..:.: ] 5
>. 4 >.
., 'g
.<;::
u
0 4
;:-
3 ~
.:E00
'0)
~3
'0)
:=: 2 ::r::
2
'P50
01 " o 1 "
8 10 12 14 16 18 20 8 10 12 14 16 18 20
Figure 2. Weight and height velocity by chronological age for rural boys compared with
the NCHS PSO
3.2 Undernutrition and stature growth

In view of the large extent of undernutrition, prevailing in the rural community,
adolescent growth performance of normal, underweight (weight-for-age < 75) and
stunted (height-for-age < 90) boys was examined on the basis of weight-for-age and
height-for-age ratios computed for their measurements in the initial round. The
underweight boys had significantly (p < .01) lower attained values for weight and
height throughout adolescence than normal boys (Table 2). The difference in the two
groups at 11 years of age is about 4 kg for weight and 8 cm for height, but increases
to almost 12 kg and 10 cm by adulthood. Underweight boys are thus clearly at a
disadvantage with respect to adolescent growth and final adult size. The trends were
almost similar for stunted boys.
GROWTH IN RURAL INDIAN CHILDREN 241
Table 2. Attained weight and height for normal ("N") and underweight ("U") boys. All
differences (Dif) were statistically significant (p < .01)
Number weight, kg height, em
Age liN" "U" "N" "u" "N" "u"

yrs n n m SD m SD Dif m SD m SD Dif
8+ 23 63 21.3 1.3 18.1 1.2 3.2 123.4 4.1 116.4 4.7 7.0
9+ 45 145 23.6 1.6 19.7 1.7 3.9 128.1 3.8 121.2 4.6 6.9
10+ 59 200 26.0 2.4 21.6 1.8 4.4 133.6 4.4 125.9 4.7 7.7
11+ 52 200 29.1 3.5 23.6 2.4 5.6 139.1 4.6 130.2 5.4 8.9
12+ 38 170 34.0 4.8 26.2 3.1 7.8 146.5 5.8 134.5 6.2 12.0
13+ 23 170 39.7 3.1 28.8 4.3 10.9 154.8 4.1 140.1 7.1 14.4
14+ 21 174 44.0 3.1 32.2 5.1 11.8 159.4 4.4 146.3 7.6 13.1
15+ 19 144 48.2 2.8 35.9 5.4 12.3 162.4 4.4 151.4 7.5 11.0
16+ 12 110 51.2 2.6 39.5 5.2 11.7 165.5 5.0 156.1 7.7 9.4
17+ 9 67 54.4 4.1 42.3 4.6 12.1 168.2 3.5 159.0 7.0 9.2
18+ 4 37 57.9 6.0 45.3 5.0 12.7 170.6 3.7 160.9 6.9 9.1
The velocity curves based on cumulative data on mean annual increments for
these groups also bring out the delay in adolescent growth among underweight as
well as stunted boys (Figure 3) as compared to their normal counterparts. The
differences in maximum increments either for weight or height, were not prominent
but were significant for ages at which maximum increments occurred. Thus, more
than growth rates, it is the timing of peak height (or weight) velocity which is more
sensitive to undernutrition. This perhaps results in significant gains at later ages
among undernourished boys compared to normals.
3.3 Linear components of growth
It would be useful to examine patterns of growth in other linear measurements such

as sitting height, leg height, foot length, shoulder width etc. Such data may help to
understand better, the contribution of different body segments in linear growth during
adolescence. However, such measurements are not so frequently measured unless for
a special purpose or study. For example, sitting height and pelvic width were
recorded for clinical purpose (Behrman et at. 1992) while leg heights have been used
in investigating secular trends (Tanner et at. 1982) or for describing racial differences
in body size (Verghese et at. 1969, Eleveth and Tanner 1976).
Data on such measurements for Indian populations are scanty (Pathmanathan and
Prakash 1994). While weight, height, chest circumference and skinfolds are
commonly recorded in many Indian studies (Tripathi et at. 1976, Banik 1982) the
measurements on other physical dimensions are rarely studied for evaluation of
adolescent growth in Indian children. Only few have examined growth in leg length
and sitting height, and they are pertaining to well-off (Agarwal et at. 1992,
Pathamanathan and Prakash 1994) or middle class Indian children (Dasgupta and Das
242 CHAPTER 19
1997)0 However, no such data, other than weight or height growth, is reported on
children from rural Indian communities wherein undernutrition has been a persistent
problemo
7 7
Weight velocity Weight velocity
6 6
...til Normal ...til
~ 5 ~ 5
---
00
.>d
---
00
.>d
o~
u
4 o~
u
4
0 0
~ ~
;;- ;;-
Q
3 3
.:E00 .:E00
0d) 0 0d)
~ 2 ~ 2
01 0
8 10 12 14 16 18 20 8 10 12 14 16 18 20
91 9
Height velocity Height velocity
8 8
tiltU 7 ...til 7 Normal a. Stunted

;>, ,0' "\Underweight tU
;>,
8u 6
0 8u 6
>. 5
0<;::: o~ 5
0
u .. u
0
~ 0 ~
;;- 4 ;;- 4
.:E00 .:E00
0d) 0d) 3
3
/--
::t: ::t:
2 0 2
O~I---.---.--~--~--~--. O~I---.---.--~--~--~--.
8 10 12 14 16 18 20 8 10 12 14 16 18 20
Figure 30 Weight and height velocities by chronological age for rural boys
3.4 Sitting height and leg length

Longitudinal data for these measurements available on 587 boys and 433 girls for 3
years period is analysed. Mean values for sitting height and leg length for boys and
girls differed marginally in pre-pubertal age but became clearly distinct (p < 0.01)
beyond 14+ years of age (Table 3). Mean sitting height as well as leg length was
smaller for girls than boys after 14 years of age and consequently, adult stature (152
cm) too was significantly (p < 0.01) smaller than that for boys (164 cm).
Table 3. Mean and standard deviation SD for sitting height (SHT) and leg length (LEGLN)
of boys and girls.
Boys Girls
sitting leg length, sitting leg length,
Age height, em em height, em em
yrs n mean SD mean SD n mean SD mean SD
9+ 64 65.5 2.8 59.0 3.9 66 64.8 2.8 59.2 3.7
10+ 120 67.2 2.9 61.9 3.8 125 66.9 2.9 62.0 3.9
11+ 204 68.7 3.2 64.4 4.1 191 69.1 3.5 65.2 4.0
12+ 267 70.3 3.2 67.2 3.8 202 71.8* 3.5 67.8 3.7
13+ 321 72.8 3.7 70.4 4.4 204 74.0* 4.0 70.2 3.8
14+ 315 76.1 4.6 73.6 4.8 209 76.5 3.6 71.8* 3.8
15+ 270 79.3 4.5 76.3 4.5 157 78.0* 3.3 72.6* 3.7
16+ 210 81.5 4.4 77.8 4.6 125 78.4* 3.3 72.5* 3.8
17+ 130 83.1 4.0 78.8 4.5 95 79.0* 3.1 73.0* 4.2
18+ 61 84.3 3.6 79.5 4.2 36 78.1* 2.5 72.3* 4.2
19+ 23 85.2 2.6 79.7 4.8 11 80.3* 2.2 71.5* 3.0
*: p < .05
The cumulative data on annual increments was used to develop velocity curves
(Figure 4). It was observed that mean annual increments in sitting height were
smaller than mean annual increments in leg length up to 13 years in boys and 10
years in girls. However, velocities in sitting height became significantly (p < 0.01)
greater than velocities in leg length beyond 14 years in boys and 11 years in girls. It
was thus observed that leg length grows faster and reaches its peak velocity before
sitting height, i.e. at 13+ years and 14+ years respectively for boys and at 10+ years
and 11 + years respectively for girls. In fact, it showed that growth of leg length in
relation to stature growth was faster before spurt among girls than boys.
3.5 Foot length and shoulder width

Mean values of foot length for girls were closer to that for boys only up to age of
13+ years but were significantly lower beyond 13+ years, and by adulthood it was
lower by 2 cm as compared to boys. However, unlike other linear measurements,
differences in shoulder width between boys and girls were not sizeable (Table 4).
244 CHAPTER 19
8
Boys 81 Girls
7 7
6 6
a ....
Q)
'"
Q)
~ 5 ~ 5
u u
z:; 4 SHT z:; 4
·u0 ·u0
~
"i3 3 ~ 3
>- D' '.
• 0
2 o--o-G ... 2
oI , , , , , , 0
8 10 12 14 16 18 20 8 10 12 14 16 18 20
Figure 4a. Velocity curves for height (RT), sitting height (SRT) and leg length (LEGLN)
in boys and girls
8l Boys 81 Girls
7 7
6 6
....
~
....
~
Q) Q)
~5 ~5
u u
z:; 4 z:; 4
'uo '0
0
"i3 3 "i3 3
>- >-
2 2
0- ,,_,,_0 _,,F~N
~-G_
~ '1 0- "-G_ FLN
°-0_
oI , , , ,0-,-0 , 0 , , ,0- 0 -,-
8 10 12 14 16 18 20 8 10 12 14 16 18 20
Figure 4b. Velocity curves for height (RT), foot length (FLN) and shoulder width (SW) in
boys and girls
Table 4. Mean and standard deviation, SD, for foot length (FLN) and shoulder width (SW)
of boys and girls.
Boy"s Girls
foot length, shoulder width, foot length, shoulder width,
Age em em em em
yrs n mean SD n mean SD n mean SD n mean SD
9+ 52 19.4 1.2 1 29.5 - 52 19.4 1.1 3 29.2 2.3

10+ lO3 20.2 1.2 35 29.1 2.4 112 20.2 1.3 32 29.7 2.1
11+ 189 20.9 1.2 lO6 30.1 2.3 170 21.0 1.4 78 30.6 2.6
12+ 215 21.7 1.2 129 30.9 2.4 163 21.5 1.2 82 31.5 2.7
13+ 248 22.6 1.6 119 31.5 2.5 172 22.1** 1.2 76 32.0 2.7
14+ 235 23.4 1.5 106 33.4 3.0 170 22.4** 1.0 76 33.4 2.7
15+ 223 24.0 1.3 74 34.4 3.3 117 22.5** 1.0 54 34.5 3.0
16+ 185 24.2 1.2 59 36.2 3.4 lO7 22.6** 1.8 40 35.5 2.7
17+ 117 24.4 1.2 41 37.7 2.7 83 22.6** 1.0 30 36.1* 2.6
18+ 54 24.8 1.2 12 38.5 2.4 29 22.6** 1.1 9 36.7 1.9
19+ 23 24.6 0.9 11 22.3** 1.1
*: p < .05; **: p < .01
Velocity curves for these measurements show that mean velocities in foot length
were larger before peak height velocity but taper off gradually both in case of girls
and boys (Figure 4). The age at maximum velocity of foot length was 13+ years in
boys and 10+ years in girls, i.e. one year before their age at peak height velocity.
Growth in foot length was faster before peak height velocity among girls than boys.
The trend in growth of foot length was thus almost similar to that observed in leg
length.
The age at maximum increment for shoulder width was however, 15+ years in
boys and 13+ years in girls. Thus, shoulder width reaches its peak value two years
after peak height velocity and one year after sitting height reaches its peak, both in
case of girls and boys. The trend in growth of shoulder width was thus consistent
with the growth in sitting height.
In general, it was observed that age at maximum increment for each measurement
was earlier for girls than boys. Further, the sequence of attainment of maximum
growth in these measurements was much the same in boys and girls, but the ages at
which maximum growth is achieved differed. Thus, the spurt in foot length and leg
length occurred before the spurt in height velocity while that for sitting height and
shoulder width occurred after peak height velocity in both sexes.
Sitting height and leg length of rural boys is compared with reported Indian
studies (Agarwal, Agarwal, Upadhyay, Mittal, Prakash and Rai 1992, Dasgupta and
Das 1997) and British data (Eleveth and Tanner 1976) in Figure 5. It shows that rural
boys have significantly lower leg lengths. However, this difference is reduced in case
of middle-class Bengali boys while Indian well-off have in fact higher leg lengths
than British children. Sitting height too showed a similar socio-economic gradient,
246 CHAPTER 19
but, unlike leg length, even for well-off Indian boys, it was lower throughout
adolescence when compared to British children. In fact, the difference in final size for
leg length of rural boys, when compared with British children, was only 3 cm (79.3
Vs 82.9 cm) but was 8.4 cm (83.9 Vs 92.3 cm) in the case of sitting height.
110 110
100 Sitting height 100 J Leg length
§ Indian well-off British 81 Indian well-off
.:E 90 90
.-~ 80
u.
~ £
~
5 80
gf ~ ~. British
.-
1 •
'B 70 OJ
~
70
(/)
60 60
50 50
9 10 11 12 13 14 15 16 17 18 9 10 11 12 13 14 15 16 17 18
110 110
100 Sitting height 100 Leg length
§ Middle class British 8 Middle class
..... 90 Bengal u 90 Bengal ..
.-~ 80
~ -5
5
~
BntIsh
~
~
-
~
80
:€
(/)
70 ~ 70
60 60
50 50
9 10 11 12 13 14 15 16 17 18 9 10 11 12 13 14 15 16 17 18
110 110
100 Sitting height 100 I Leg length
§ Indian rural British 8 1 Indian rural
,E. 90 ... -- u. 90 B't'h
.-~
1
~ £ n IS
-
5
~
80 80
~
.-
~ ~
'B 70 OJ
~
70
(/)
60 60
50 50
9 10 11 12 13 14 15 16 17 18 9 10 11 12 13 14 15 16 17 18
Figure 5. Comparison of sitting height and leg length of Indian well-off, Bengali middle
class and Indian rural boys with British boys.
Similar comparisons of observed mean foot length and shoulder width values of rural
children with British children showed that foot length values were significantly (p <
0.01) lower throughout adolescence and the differences increased progressively up to
adulthood. However, differences in shoulder width were not prominent when
compared with British children.
3.6 Undernutrition and linear components of growth
Rural Indian children are known to suffer from malnutrition in early life, as also
from a lack of optimal nutrition during adolescence. It would be therefore, worth to
examine ages at maximum increment for various physical dimensions and compare
with well-to-do Indian children for examining the effect of undernutrition on linear
components of growth.
We have observed in the previous section that differences in sitting height were
more prominent than those in leg length in case of rural children. Thus,
undernutrition prevailing in rural communities has affected all the components of
linear growth but sitting height is relatively more affected than other parts.
The rural group as a whole, however, consists of individuals suffering from both
short-term and long-term undernutrition. Therefore, in order to examine the effect of
current undernutrition, children were classified as normal (weight-for-age ~ 75) or
undernourished (weight-for-age < 75) as before and their body measurements were
compared.
Underweight children had lower attained values for all the measurements viz., leg
length, sitting height, foot length and shoulder width compared to normal children
(Figure 6). Further, it can be observed (Figure 6) that children with normal weight
status had leg lengths similar to that of British children, but had significantly lower
sitting heights. Therefore, most of the difference in adult height was due to
differences in sitting height. However, underweight children had lower values for
sitting height and leg length when compared with their normal counterpart as well as
British children. Thus, differences in adult height were mainly attributed to
differences in sitting height in case of normal children, but additionally to differences
in leg length in the case of underweight children.
Similar comparisons for foot length and shoulder width showed (Figure 6) that
mean foot length values for normal boys were close to British values up to 15 years
of age, but were lower thereafter. In contrast, mean values for shoulder width for
normal boys were higher than British values throughout adolescence. Thus, children
with normal weight status during adolescence had satisfactory leg length, foot length
and shoulder width, but lower sitting height while underweight children had lower
values for all body dimensions.
Undernutrition delayed ages at maximum increments of all the measurements, but
the sequence of their attainment remained the same as in normal children. Despite the
fact that for underweight boys, velocities at later ages were higher for all the
measurements, compared to normal children, their final adult sizes were lower than
the normal children. Thus, although there is some evidence for catch-up growth, it
was not sufficient to achieve comparable final adult sizes.
248 CHAPTER 19
4. SUMMARY
In conclusion, rural children had lower attained values than British or Indian well-
off children for all the linear components of stature growth considered in this study.
The sequence of the adolescent spurt in these linear components is the same in boys
as in girls, i.e. the peak velocity in foot length and leg length is reached before peak
height velocity, while that for sitting height and shoulder width occurs after peak
height velocity. While foot length was the first trait to reach peak velocity, shoulder
width was the last one in this sequence. The ages at maximum increment for each of
the linear components were earlier for girls than boys.
100, 100
Sitting height Leg length
I
au 90 90
au
~80 0;9" 80
'i) 01)
..c: I':
(!.)
.S
t::
70 -; 70
(!.)
....l
i;i)
60 • Normal
---{]-- underweighl
60 ~ • Normal
---{]-- Underweight
.. ------- British - - - - - - - - - British
501 , , , , , , , , , , 50
9 10111213141516171819 910111213141516171819
28, 45
Foot length I Shoulder width
26
a § 40
u 0;9"
0;9" 24 -0
01)
I': .~ 35
~ (!.)
-0
-00 22 "3
0
t5
~
30
____ Normal
---{]-- Underweight
~-
• Normal
---{]-- Underweigh
--------- British --------- British
181 , , , , , , , , , , 25
9 10111213141516171819 910111213141516171819
Figure 6. Comparison of British, normal and underweight boys for sitting height, leg
length, foot length, and shoulder width.
The growth of length in relation to stature growth was faster before the spurt
among girls than among boys. In general, trends in growth of foot length and
shoulder width were similar to those observed for leg length and sitting height
respectively.
Undernutrition during adolescence not only delayed the adolescent growth spurt,
but also affected all the linear components of growth. Comparison of rural data with
Indian well-off and British children revealed that the impact of undernutrition was
relatively more on sitting height as compared to other measurements. Further, it was
observed that leg length improved as social class improved, and Indian well-off
children had actually higher values of leg length than British children, but this was
not true for sitting height. It thus, appeared that growth in leg length was influenced
by post-natal environmental influences while growth in sitting height had probably
influence of intra-uterine nutritional environment apart from genetic influence. In
that case, relative growth in sitting height to leg length can provide an index of
disproportionate growth.
Recent hypothesis of 'foetal origins of adult diseases' describes risks of adult
diseases in relation to disproportionate growth at birth (Barker 1994). In view of
this, the findings presented in this paper indicate the need for more observational
studies to understand implications of disproportionate growth of linear components
in adulthood in relation to the risk for adult diseases. The likelihood for such a
relationship appears possible as it is known that prevalence of hypertension, diabetes
and coronary heart disease are low in rural populations and increase as we go to urban
middle and urban affluent classes. The observations discussed in this paper thus have
significant implications for adult health.
Acknowledgements. I am thankful to Dr. Rao V.S., Director of Agharkar

Research Institute for providing facilities to carry out the longitudinal growth study
on rural Indian children. I am also thankful to my colleagues Dr. Mrs. Kanade, Dr.
Mrs. Joshi and Dr. Mrs Gokhale and Mr. S. Girigosavi for their help in conducting
the field work and data analysis. I am also thankful to the Indian Journal of Nutrition
& Dietetics, Indian Pediatrics, and Annals of Human Biology, for permitting me to
reproduce some tables and figures published earlier in these journals.
5. REFERENCES
Agarwal, D.K, Agarwal., KN., Upadhyay, S.K., Mital, R, Prakash, R, and Rai, S., 1992, Physical and
sexual growth pattern of affluent children from 5 to 18 yrs of age. Indian Pediatrics, 29, 1203-1283.
Banik, N.D.D., 1982, Semi-longitudinal growth evaluation of children from birth to 14 yr in different
socio-economic groups. Indian Journal of Pediatrics, 19,353-359.
Barker, DJ.P., 1994, Mothers, Babies and Diseases in Later Life (BMJ Publishing Group).
Behrman, R.E., Kleigman, R.M., Nelson, W.E., and Vaughan, V.C., 1992, Textbook of Pediatrics
(Philadelphia, P.A.W.B.: Saunders Company), 14th Edition.
Dasgupta, P., and Das S.R., 1997, A cross-sectional growth study of trunk and limb segments of the
Bengali boys of Calcutta. Annals of Human Biology, 24, 363-369.
Eveleth, P.B., and Tanner, J.M., 1976, Worldwide variation in Human Growth (Cambridge University
Press: Cambridge).
Pathmanathan, c., and Prakash, S., 1994, Growth of sitting height, subischial leg length and weight in
well-off North-Western Indian children. Annals of Human Biology, 21, 325-334.
Satake, T., Kikuta, F., and Ozaki, T., 1993, Ages at peak velocities and peak velocities for seven body
dimensions in Japanese children. Annals of Human Biology, 20, 67-70.
Tanner, J.M., Hayashi, T., Preece, M.A., and Cameron, N., 1982, Increase in length of leg relative to
trunk in Japanese children and adults from 1957-1977: Comparison with British and Japanese
Americans. Annals of Human Biology, 9, 411-423.
Tripathi, A.M., Agarwal, D.K., and Agarwal, KN., 1976, Physical growth during adolescence in Delhi
school children. Indian Pediatrics, 13, 191-200.
Verghese, KP., Scott, RB., Teixeira, G., and Ferguson, A.D., 1969, Studies in growth and development,
XII. Physical growth of North American Negro Children. Pediatrics, 44, 243-249.
250 CHAPTER 19
Vijayaraghavan, K., Gowrinath, S.T., and Indubhushan, M.e., 1974, Growth performance of well-to-do
Hyderabad children - a follow up study. Indian Journal of Medical Research, 62, 117-124.
CHAPTER 20
GROWTH STUNTING AMONG CHILDREN, AGED

BIRTH TO FIVE YEARS, IN PERI-URBAN
KATHMANDU, NEPAL
T. MOFFAT
McMaster University, Hamilton, Ontario, Canada
1. INTRODUCTION
Anthropometric studies of children in developing countries elucidate a growth pattern

that has been found in numerous studies of geographically diverse populations
(Anderson 1979, Sastry et al. 1989, Watts et al. 1990, Dettwyler 1991, Neumann
and Harrison 1994, Leonard et al. 1995, Ricci and Becker 1996). Essentially, linear
growth tracks reference standards until approximately six months of age, at which
point growth faltering is detectable and continues until two to three years of age. The
second year of life is often the growth nadir, although this can vary. If conditions are
not too severe, there may be a "catch-up" period, when children begin to track
reference standards again by age three or four (Martorell et al. 1994).
The exact inception and cessation of growth faltering is somewhat debatable
because of discrepancies between cross-sectional and longitudinal studies (Beaton
1992). Indeed, since most growth studies have been cross-sectional in design, it has
been assumed that the onset of growth faltering begins at six months. Some
longitudinal studies show, however, that the process of growth faltering, i.e. a
decrease in growth velocity, can begin earlier than six months, as early as the second
to fourth month after birth (Naborro et ai. 1988, Neumann and Harrison 1994,
Leonard et al. 1995, Waterlow 1988). In terms of lowered growth velocity, or the
process of becoming stunted, the most intense period of growth retardation occurs
generally between three to 12 or 18 months, although it has, in some populations,
been known to extend into the third year or longer (Martorell et al. 1994).
The highest prevalence of linear growth retardation, or stunting, is found in the
southern hemisphere of the world, cutting across a variety of ethnic groups (Keller
1988). Within the southern hemisphere, some of the highest rates of stunting world-
wide are found on the South Asian subcontinent (Victora 1992). Located in South
Asia, the Himalayan Kingdom of Nepal is home to some of the smallest people in
the world (Eveleth and Tanner 1990). Considering this, it is surprising that relatively
few growth studies have been conducted in Nepal. This dearth is probably connected
to the relative political and geographic isolation of the nation until the 1960s, only
after which the country was rendered more accessible through the establishment of
251
252 CHAPTER 20
transportation and communication lines. Studies that have been done in Nepal in
various locations document an extremely high prevalence of stunting for children
less than five years of age with little change in growth status from the 1970s to the
1990s (Brink et al. 1976, Martorell et al. 1987, Costello 1989, UNICEF 1996,
Panter-Brick 1997).
Much less research on child growth has been conducted among urban children in
Nepal. An exception here is Panter-Brick et al. /s (1996) study of the growth of
homeless urban boys compared to urban rural boys. Nepal is a mainly rural, but
rapidly urbani sing country; in the past two decades it has undergone an unprecedented
rate of urbanisation (UNICEF 1992). The studies that do examine urban children
assume a rural/urban contrast without exploring the complex mosaic of differing
socio-economic landscapes within urban environments (Brockeroff 1995). Thus,
intensive research on urban sub-populations is needed to understand the complexity
of child growth within the larger urban communities.
Those living in peri-urban locations are noted as being distinct within the urban
environment. Peri-urban communities are situated on the periphery of the city and
are comprised mainly of rural-to-urban migrants working in industry or living in
squatter settlements. In general, people living in peri-urban communities are more
impoverished than those living within the city proper having less resources and
essential services such as electricity, piped water and sewage systems (Harpham et al.
1988).
This study evaluates the growth of a sample of children less than five years of
age residing in peri-urban Kathmandu. Growth is examined by employing both
cross-sectional and longitudinal anthropometric data in order to get at what is
described above as both the process and the end product of linear growth retardation.
Aspects of growth faltering in low-income countries such as the relationship
between wasting and stunting, gender differences, seasonality, and comparisons to
children in rural locations are examined.
2. METHODS
2.1 Subjects and study design

This study is part of a larger research project that examined the impact of mothers'
wage labour on child health and development. All the mothers of the children in the
sample were working in the carpet-making industry in peri-urban Kathmandu. Data
collection extended for nine months from January to September 1995. A sample of
17 carpet-making factories and 59 households from a squatter settlement were
selected as study sites. All mothers with children less than five years of age were
asked to participate. The author and research assistants followed many of the families
for a total of three rounds of anthropometry.
All anthropometric dimensions were taken by the author either at the carpet
factories or at two co-operating health clinics, located at convenient distances from
the study sites. Measurements were done according to Lohman et al.'s (1988)
standardised techniques. For each measurement children's shoes, socks and caps were
removed, although total nudity was not permissible because of low temperatures in
GROWTH STUNTING IN NEPALESE CHILDREN 253
the cold season. Heavy sweaters and belts, however, were removed. Children's weight
was measured with a portable suspended infant/child weighing scale (Perspective
Enterprises, PE-HS-25, Kalamazoo, MI) to the nearest 0.1 kg. The scale was set to
zero at the beginning of each session and checked periodically throughout the
sessions. The scales were calibrated bi-monthly with a 2 kg weight and there were no
changes detected throughout the study period. Recumbent length was taken for
children up to and including 24 months, and height was measured for those over 24
months with a portable wooden adult/infant measuring board (Perspective
Enterprises, PE-AIM-lOl, Kalamazoo, MI) to the nearest millimetre. Arm
circumferences were measured to the nearest millimetre with insertion slot measuring
tapes (Perspective Enterprises, Kalamazoo, MI) and triceps skinfolds in millimetres
with a Lange skinfold calliper (Beta Technology Incorporated, Cambridge, MD). For
triceps skinfolds, three measurements were taken in quick succession and an average
of the three was used as the measurement (Frisancho 1990).
Nepalese children have no official birth certificates, so the age of the child had to
be ascertained from the parents. In most cases parents did not know the year of birth,
but they did know the age of the child and the month and day on which the child was
born. Dates were converted from the Nepalese calendar to the Western calendar and
then checked to confirm that they corresponded with the age stated by the parent. The
caregiver was asked at each subsequent measuring session to state the date of birth
and the age of the child in order to check the accuracy of the information.
2.2 Data Analysis

There were a total of 283 children in the sample, although portions of this sample (n
= 83) and (n = 12) joined the study in rounds II and III respectively. Fifty of the
mothers of the children in the sample had two or more children less than five years
of age. In order to test whether this biased the anthropometric data, the eldest
siblings were removed from the sample, leaving a remaining 218 index children, i.e.
one child from each household. Since there were no significant differences between
this sample (n = 218) and the overall sample (n = 283), the data for the whole
sample are presented here.
For cross-sectional data each anthropometric measurement was converted in
Epinfo version 6 software to Z-scores of the National Center for Health Statistics
(NCHS) reference population for boys and girls for height-for-age (HAZ), weight-for-
age (W AZ) and weight-for-height (WHZ). For longitudinal growth data, mean
growth velocities were calculated by comparing differences in height and weight
between the first and third measuring sessions. The differences were divided by the
exact time period in months between the measuring sessions to obtain an average
monthly growth rate. These rates were then multiplied by six to make them
comparable to the reference population's velocity over a six month period (See
Leonard et al. 1995 for this method). Student t-tests and Analysis of Yariance
(ANOY A) for single and repeated measures were used to test for statistical differences
by age, gender and season.
254 CHAPTER 20
3. RESULTS
3.1 A cross-sectional view of growth

This section presents an overall view of attained growth for all of the children
who were measured at least one time throughout the study (n = 283). Mean weight,
height and their standard deviations by sex are presented in Table 1, arm
circumferences and triceps skinfolds in Table 2. Mean Z-scores for each of the indices
-- HAZ, W AZ, and WHZ -- are presented in Table 3.
Table 1. Mean and standard deviation (SD) of heightf and body weight of children aged 0-
60 months, measured at first survey
Males Height, cm Weight, kg

Age, months n mean SD n mean SD
0.0 - 5.9 18 61.8 3.3 18 6.2 1.0
6.0 - 11.9 15 67.4 3.9 16 7.6 1.4
12.0 - 23.9 40 74.8 3.2 41 9.1 1.1
24.0 - 35.9 21 81.4 2.8 24 10.9 1.2
36.0 - 47.9 24 88.5 5.2 25 12.4 1.4
48.0 - 59.9 10 92.5 6.6 10 13.6 1.8
Females Height, cm Weight, kg
Age, months n mean SD n mean SD
0.0 - 5.9 22 57.5 4.6 22 4.9 1.4
6.0 - 11.9 25 65.9 3.0 27 7.1 0.8
12.0 - 23.9 37 75.0 3.6 38 8.6 0.9
24.0 - 35.9 23 80.0 4.1 23 10.6 1.3
36.0 - 47.9 25 87.4 4.6 26 11.9 1.7
48.0 - 59.9 12 90.7 3.5 12 12.9 1.1
t Recumbent length for children under 24 months.
A comparison of mean HAZ, WAZ and WHZ scores by age for both sexes
presented in Figure 1 shows that, apart from the infants, these children, on average,
are moderately-to-severely retarded in terms of linear growth and mild-to-moderately
underweight for age. Their WHZ scores, a measure of acute weight loss, however,
are very close to the NCHS reference median, indicating that they are not wasted
relative to their body frame.
Table 2. Mean arm circumference (e), triceps skinfoids (Ts) and Upper Arm Areas(UMA)
of children aged 12 - 60 months, measured at first survey.
Males C, em Ts, mm UMA, em 2

Age, months n mean SD mean SD mean SD
12.0 - 23.9 41 13.9 1.2 10.2 2.4 9.1 1.7
24.0 - 35.9 22 14.3 0.9 12.2 2.5 8.7 1.6
36.0 - 47.9 25t 14.4 1.3 11.1 3.4 9.7 2.1
48.0 - 59.9 10 14.5 1.1 11.0 3.4 9.8 0.9
Females C, em Ts, mm UMA, em2
12.0 - 23.9 41 13.9 1.2 10.2 2.4 9.1 1.7
24.0 - 35.9 22 14.3 0.9 12.2 2.5 8.7 1.6
36.0 - 47.9 25t 14.4 1.3 11.1 3.4 9.7 2.1
48.0 - 59.9 10 14.5 1.1 11.0 3.4 9.8 0.9
t There were 26 boys and 38 girls measured for upper arm circumference
Table 3. Mean and standard deviation (SD) ofZ-scoresfor HAZ, WAZ and WHZ in males
and females, aged 0 - 60 months, at first survey.
Males HAZ WAZ WHZ

Age, months n mean SD n mean SD n mean SD
0.0 - 5.9 18 -0.36 0.75* 18 -0.23 0.80* 18 -0.00 0.77
6.0 - 11.9 15 -1.50 1.36 16 -1.35 1.47 15 -0.44 0.92
12.0 - 23.9 40 -2.12 0.59 41 -1.81 0.95 40 -0.72 0.95
24.0 - 35.9 21 -2.65 0.95 24 -1.91 0.79 21 -0.33 0.81
36.0 - 47.9 24 -2.62 1.21 25 -1.89 0.74 23 -0.38 0.78
48.0 - 59.9 10 -3.18 1.27 10 -2.07 0.73 10 -0.31 0.72
Total sample 128 -2.06 1.31 134 -1.59 1.08 127 -0.43 0.87
Males HAZ WAZ WHZ
Age, months n mean SD n mean SD n mean SD
0.0 - 5.9 22 -1.08 0.88* 22 -1.06 0.98* 22 -0.35 1.01
6.0 - 11.9 25 -1.42 0.88 27 -1.29 0.71 25 -0.18 0.76
12.0 - 23.9 37 -1.80 0.96 38 -1.79 0.62 37 -1.03 0.75
24.0 - 35.9 23 -2.59 1.03 23 -1.75 0.92 23 -0.23 0.60
36.0 - 47.9 25 -2.66 0.94 26 -1.98 0.97 25 -0.36 0.70
48.0 - 59.9 12 -3.24 0.94 12 -2.07 0.69 12 -0.22 0.76
Total sample 144 -2.02 1.15 148 -1.64 0.87 144 -0.46 0.83
* mean difference between males and females is significant at p < 0.01
256 CHAPTER 20
o
-0.5
- I f- ........................ ~
~ -1 .5 f- .........................= ...................
~
8
N -2 f- ............................................ ······~··················· f
VJ
-2.5 r-·t ~ HAZ

• WAZ
-3 1-·1 D WHZ
-3.5 -=--=-=----:----:--------------~
o - 5.9
,-I
6 - 11.9 12 - 23.9 24 - 35.9 36 - 47.9 48 - 59.9

Age, months
Figure 1. Mean Z-scores of HAZ, WAZ and WHZ by age category for all children measured
for the first time (n = 283)
Figure 1 shows the progressive decrease in mean Z-scores for all three indices as
the ages of the children increase. One-way ANOV A tests of all indices indicate that
the mean Z-scores for children in each age category are significantly different (p <
0.0001). A Tukey-HSD post-hoc analysis of the means indicates that for HAZ,
children less than six months old are significantly different from all other age
groups; children from six to 12 months are significantly different from two, three
and four-year olds and one-year olds are significantly different from two, three and
four-year olds. For WAZ, only infants « 12 months) are significantly different from
all other age categories. For WHZ, all age groups differ significantly from the one-
year olds. This analysis indicates that under-weight-for-age and stunting peaks during
years one and two respectively, and continues on through the fourth year of life. The
most serious state of low WHZ, or wasting, occurs among the one-year olds.
Upper arm circumferences were converted to muscle areas and then converted to
Z-scores (ZAM) along with triceps skinfolds (ZTR) relative to the NCHS reference
medians (Frisancho 1990) (Table 4). Mid upper arm circumference Z-scores (ZCIRC)
for all children range between -2 and -1 SD below the NCHS reference medians.
Triceps skin fold Z-scores (ZTR), however, are all close to the reference median, i.e.
they are abnormal or slightly above normal in terms of fatness for that area (Figure
2). The arm measurements corroborate what was found for height and weight; that is,
these children are small but not wasted. Low arm muscle area Z-scores (ZAM)
indicate that there is more wasting of the upper arm muscle than there is of upper
arm fat. It has been proposed that arm circumference be used as a quick and simple
measure of malnutrition (Anderson 1979). Arm circumference and arm muscle area,
however, do differ as shown by their respective Z-scores, ZCIRC and ZAM (Table
4). A child can appear to be better off because he/she is plump and has a moderate
size arm circumference, but the child may still have poor muscularity as indicated by
a low ZAM.
Table 4. Mean and standard deviation (SD) ofZ-scoresfor triceps skinfolds (ZTR), mid
upper arm circumference (ZCIRC) and upper arm muscle area (ZAM) of children aged I2 -
60 months.
Males ZTR ZCIRC ZAM

12.0 - 23.9 41 -0.06 0.81 -1.85 0.96 -1.77 0.72
24.0 - 35.9 22 0.74 0.85 -1.54 0.62 -1.68 0.50
36.0 - 47.9 25t 0.45 1.24 -1.77 0.93 -1.79 0.67*
48.0 - 59.9 10 0.67 1.25 -1.91 0.77 -2.42 0.31*
Total sample 98:j: 0.33 1.04 -1.77 0.86 -1.82 0.66
Females ZTR ZCIRC ZAM
12.0 - 23.9 37t -0.25 0.80 -1.86 0.85 -1.69 0.76
24.0 - 35.9 22 0.38 1.00 -1.63 0.62 -2.00 0.55
36.0 - 47.9 25 0.59 0.91 -1.72 0.77 -2.34 0.85*
48.0 - 59.9 12 0.48 0.57 -1.60 0.36 -2.04 0.46*
Total sample 96:j: 0.20 0.92 -1.74 0.73 -1.97 0.75
t There were 26 boys and 38 girls used in the calculation of ZCIRC
** There were 99 boys and 97 girls in the total samples used in the calculation of ZCIRC
Mean difference between males and females is significantly different at < 0.05
p
0 ZAM
0.51··~ ZTR
0
~ -0.5
0
u
N'" -1
I
-1.5
-2
-2.5
12-23.9 24-35.9 36-47.9 48-59.9
Age, months
Figure 2. Mean Z-score for upper arm muscle area (ZAM) and triceps skinfolds (ZTR) for
all children measuredfor the first time (n = 194)
258 CHAPTER 20
One-way ANOV As for mean ZCIRC, ZTR and ZAM by age category indicate
that ZCIRC does not differ significantly in terms of age. This conforms to the
generalisation that upper arm circumference can be used as a test of malnutrition
independent of age, because it remains relatively constant from 12 to 60 months of
age (Anderson 1979, Gorstein et al. 1994). Mean ZTRs do differ significantly by age
category (p = 0.00). A post-hoc Tukey's-HSD test at the significance level of 0.05
indicates that one year olds have a significantly lower mean ZTR than the older age
groups. ZAM, which takes both ZCIRC and ZTR into account, is also significantly
different by age category (p = 0.01); three and four-year-olds have significantly lower
ZAM scores than one and two year olds. These age trends in arm measurements
corroborate height and weight age trends. The lowest skinfold thickness occurs
among one year olds, which corresponds with the period of greatest wasting or low
weight-for-height. The lowest mean ZAMs occur after the first year of Hfe, mirroring
the drop in mean W AZ and HAZ. After two years of age children remain short, with
low muscularity, but not thin.
Bivariate correlations between ZAM and HAZ are positive and significant (r =
0.32, p = 0.00), but only moderately so. ZAM and W AZ are more highly positively
correlated (r = 0.47, p = 0.00). Thus, children who are stunted and underweight are
likely to have poorer muscularity and vice versa. ZTR is also positively correlated
with WHZ (r = 0.47, p = 0.00). In other words, on average, those who are wasted
will have correspondingly low triceps skinfold scores. There is no significant
correlation between ZTR and HAZ (r = 0.-14, p = 0.05) and a weak one between
ZTR and W AZ (r = 0.22, p = 0.002). This makes sense because both ZTR and WHZ
are measures of thinness, whereas W AZ and HAZ are measures of body size.
3.2 Prevalence of stunting, wasting and low weight among infants and children
Since many studies report anthropometric findings in terms of prevalence of
stunting, underweight and wasting (HAZ, W AZ and WHZ < -2 SD), it is important
to report these results here. Moreover, although the mean Z-scores may show where
the group as a whole stands, they do not indicate what proportion of the sample is in
the lower range of the NCHS reference population.
Table 5. Prevalence of low (5 - 2SD) ht/age (stunted), wt/age (underweight) and wtlht
(wasted) for children aged 0 - 60 months
stunted underweight wasted

Age, monthst n % n % n %
0.0 - 5.9 40 5.0 40 5.0 40 5.0
6.0 - 11.9 40 32.5 43 23.3 40 2.5
12.0-17.9 49 42.9 51 35.3 49 8.2
18.0 - 35.9 72 65.3 75 42.7 72 4.2
36.0 - 60.0 71 77.5 73 47.9 70 1.4
6.0 - 36.0 161 50.3 169 35.5 161 5.0
t The age categories are divided according to UNICEF (1996) national surveillance report
on the health and nutrition of Nepalese children for the sake of comparison. For the total
sample, a rate for children aged 6 to 36 months of age is presented, corresponding to the
age range used in the national study.
The percentages of children below the cut-off point of -2 SD are presented in

Table 5. Again, it is apparent that similar to other studies of Nepalese children
discussed above, stunting is the major problem for these peri-urban children: half of
them were classified as stunted, whereas only 5% of them were below the cut-off
point for wasting. Similar to the findings for mean Z-scores, there was a higher
proportion of stunted and underweight children among the older compared to the
younger children. Only 5% of infants under six months were stunted and/or
underweight compared to 78% of children between 36 and 60 months.
Figure 3 compares the prevalence of stunting with 95% confidence intervals in
this peri-urban sample to the national rural and urban prevalence found in a UNICEF
(1996) survey. The rural prevalence is significantly higher than that ofthe peri-urban
sample for the 12 to 17 month old children. The same is true for the six to 36
month group. By comparison, the peri-urban and urban prevalence of stunting are
within the same range for all age groups shown.
70
60
50
1
p...
40
30
________ peri-urban
20 ~ rural national
--i'r- urban national
10
6 - 1l.9 12 - 17.9 6 - 36.9
Age, months
Figure 3. Stunting (HAZ < -2SD) prevalence by age category of Nepalese National Survey
of urban and rural population (UNICEF 1996) and peri-urban sample.
3.3 Longitudinal growth

Table 6 presents longitudinal, age-specific, average growth rates for staturellength
(cml6 months) and body weight (kg/6 months) for males and females relative to US
reference data from the Fels longitudinal growth study (Baumgartner et al. 1986).
Mean growth rates are from Rounds I to III. The growth spans an average of 6.16
±0.32 months between rounds I (January to March) and III (July to September).
260 CHAPTER 20
18
16
----0-- US 3rd
14 ~US50th
12 - Nepalese boys
~'" 10
0
E 8
\0
] 6
4
2
0
0 6 12 18 24 30 36 42 48 54 60
Age at second measurement, months
Figure 4. Linear growth velocity (Round I to Ill) for Nepalese boys relative to US
longitudinal data (3rd and 50th percentiles)
18
16
----0-- US 3rd
14 ~US50th
12 e--- Nepalese girls

'" 10
.p
Q
0
E
\0
8
] 6
4
2
0
0 6 12 18 24 30 36 42 48 54 60
Age at second measurement, months
Figure 5. Linear growth velocity (Round I to Ill) for Nepalese girls relative to US
longitudinal data (3rd and 50th percentiles)
Mean linear growth rates from Table 6 (Rounds I to III) are plotted against
normative data (US 3rd and 50th percentiles) from Fe1s longitudinal study
GROWTH STUNTING IN NEPALESE CHlLDREN 261
(Baumgartner et at. 1986) for linear growth velocities (cml6 months) (Figure 4). In
this sample Nepalese males have a slow linear growth rate compared to US males;
the Nepalese means track the 3rd percentile until the third year of life, after which
they approach the 50th percentile. On the other hand, the mean linear growth rate for
females declines precipitously from 6 to 18 months, passing from the 50th down to
the 3rd US percentile. As with the males, however, the mean linear growth rate
increases after the third year of life, approaching the 50th percentile again.
Table 6. Mean and standard deviation (SD) of growth rates (cml6 mths) of stature and
body mass for children aged 6 to 60 months between Rounds I and Ill.
Males Stature Body Mass

Age, monthst n mean SD n mean SD
6.0 - 11.9 7 7.71 3.13 7 1.85 0.67
12.0 - 23.9 19 4.17 1.38 21 0.27 0.52
24.0 - 35.9 12 3.22 1.47 12 0.55 0.53
36.0 - 47.9 12 3.59 0.74 12 0.39 0.32
48.0 - 59.9 9 2.90 0.61 9 0.32 0.57
Females Stature Body Mass
Age, monthst n mean SD n mean SD
6.0 - 11.9 7 10.89 2.87 7 2.55 0.96
12.0 - 23.9 12 4.93 1.09 14 0.59 0.49
24.0 - 35.9 13 3.21 1.00 13 0.34 0.63
36.0 - 47.9 10 3.59 1.00 10 0.47 0.55
48.0 - 59.9 8 3.02 1.00 8 0.26 0.33
t Age at second measurement
Table 7. Repeated measures ANOVA tests for differences in mean HAZ taken for children
measured in Rounds I, II and III and covaried by age at measurement (n = 107)
Source of Sum of Mean

variation Squares df squares F-ratio F prob.
Within + residual 19.93 211 0.09
Regression 0.16 1 0.16 1.64 0.20
Season 0.17 2 0.09 0.91 0.41
The longitudinal growth data can be used to investigate seasonal variation in

mean growth as measurements were taken in each of the cold, hot and rainy seasons
(Rounds I, II & III respectively) through the study year. This type of analysis
requires the use of repeated measures ANOV A, since the same children's
measurements are compared in each season. The repeated measures ANOVA tests use
child's age at the second and third measurements as a covariate because growth
velocity and mean HAZ-scores decrease with increasing age. Since a child's age was
262 CHAPTER 20
not a significant covariate for WHZ, it was not included in the repeated measures
ANOVA test for differences in mean WHZ by season.
As can be seen from the results of the repeated measures ANOVA tests (Tables 7
to 9), there are no statistically significant differences in HAZ and W AZ in cold, hot
and rainy seasons. Mean WHZ, however, does vary by season (Table 9). Mean WHZ
scores for children measured in the winter, hot and rainy seasons respectively are as
follows: -0.40, -0.72, -0.84.
Table 8. Repeated measures ANOVA tests for differences in mean WAZ taken for children
measured in Rounds I, II and III and covaried by age at measurement (n = 113)

variation Squares df squares F-ratio Fprob.
Regression 0.63 1 0.63 5.87 0.02
Season 0.49 2 0.25 2.31 0.10
Table 9. Repeated measures ANOVA tests for differences in mean WHZ taken for children
measured in Rounds I, II and III (n = /07)

variation Squares df squares F-ratio Fprob.
Season 10.62 2 5.31 33.81 0.00
4. DISCUSSION
Linear growth retardation or stunting is one of the most prevalent forms of growth
faltering affecting children in low-income countries. It is estimated that
approximately 50 to 70% of pre-school children in South Asia are growth stunted
(Victora 1992). This is also the case in Nepal (UNICEF 1996), as well as for the
peri-urban sample reported on here.
Although linear growth retardation, or stunting, is the salient issue for these
children, there may be some connection between growth in stature and body mass.
Waterlow (1992) is adamant that linear growth retardation and wasting are separate
physiological processes, although he has since acknowledged that there is evidence
that episodes of wasting may precede stunting (Waterlow 1994). Naborro et al.
(1988), Costello (1989) and Panter-Brick (1997) all show evidence that the thinnest
children in their samples gained relatively more body weight than other children in
succeeding months, but at the expense of growth in stature. In terms of cross-
sectional growth for this peri-urban sample, this relationship can be seen from the
differential age pattern of growth status. The lowest mean WHZ occurs in the sample
from 12 to 24 months of age. This thinness is corroborated by a significantly lower
mean ZTR for this age group. Notably, this state precedes the peak in low mean
HAZ, or stunting, from 24 to 60 months.
Longitudinal growth patterns lend a more dynamic view of child growth.
Although cross-sectional or attained growth indicates that the Nepalese children in
this sample were not stunted until after 12 months of life, boys from early infancy
have low linear mean growth velocities. Girls' mean velocity, however, only drops
after the first year. As mentioned above, this may be due to the latter being
significantly smaller at birth. Lack of birth weight data is a drawback of this study,
as there is evidence that birth weight may determine the pace and absolute size of
early infant growth (Chen et al. 1980). Early infant growth faltering, moreover, may
be regulated by deficient in utero nutrition and poor micro-nutrient stores at birth
rather than the postnatal environment (Allen 1994).
Linear growth velocities do improve after the third year of life, when both males
and females approach the 50th percentile of the US normative data. This is in
contrast to cross-sectional attained growth where three and four-year-olds have the
lowest mean HAZ scores and the highest prevalence of stunting. The difference in
these two types of measures illustrates the distinction between the process of
stunting and the state of being stunted. The process of stunting began at six months
(3 months for boys) and lasted through to 36 months, but the children were stunted
from 24 months onward. There appears to be an increase in linear growth velocity
for both male and female three and four year olds. This corresponds with what is
termed catch-up growth for three and four year old children in low-income countries
(Martorell et al. 1994), although it is not enough to overcome original deficits.
This state of being stunted is also corroborated by significantly reduced upper arm
muscularity for three and four-year-olds compared to the younger children. Results
from arm measurements indicate that there is more wasting for these children in
upper arm muscle compared to mid-upper arm body fat. This is further evidence of
protein-energy-malnutrition; children may be obtaining enough calories, but not
enough protein to build body muscle (Frisancho 1990).
There appear to be some gender differences in arm measurements. Interestingly,
among the four-year-olds (Table 4), it is the girls who have slightly higher mean
ZAM scores, whereas among the three-year-olds, the boys have higher mean ZAM
scores. This may be due to cohort differences rather than systematic gender
differences. Indeed, there is no difference in mean ZAM scores between boys and girls
for the sample as a whole.
The only difference in mean Z-scores between the sexes is found for mean HAZ (t
= 2.74, df = 38, p = 0.009) and W AZ (t = 2.91, df = 38, p = 0.006) of infants less
than six months of age (Table 3). In this case the mean Z-score is lower for females
than males. It is surprising that female infants would be significantly smaller than
male infants at such a young age, as gender differences in Z-scores usually signal
preferential care of boys over girls. Infants less than six months are being breast-fed
with little in the way of supplementation, so they would not have much opportunity
for preferential treatment through better food. It is possible that mothers are breast-
feeding male infants more frequently than female ones, although it is difficult to
prove this without detailed observational data.
While females begin growth faltering after six months of age, males appear to
have slower growth rates earlier in the infancy period (Figures 4 and 5). The females
less than six months of age, as discussed above, are significantly smaller than their
264 CHAPTER 20
male counterparts. Therefore, the faster female infant growth velocity may be due to
post-natal (in the first six months) catch-up growth and regression to mean. See
Baumgartner et at. (1986) for explanations of these phenomena. That is, they may
have been growing faster during the infancy period to make up for deficits in size at
birth. This may explain why they are comparable in size to the males in later years.
Unfortunately, without birth weights, it is difficult to test this hypothesis. Birth
weights were unobtainable for this sample because more than half of the children
were born at home and no birth weight was recorded. Those born in the hospital
were not given a record of their birth weight.
There is no doubt that discrimination against girl children is an important issue
in South Asia and evidence of preferential care of males over females has been
highlighted in studies of household food allocation in Nepal; however, differences are
found only during adolescent and adult years (Gittelsohn et at. 1997). The fact that
no systematic evidence of gender differences in body size was found in this peri-urban
sample concurs with the results of other growth studies in Nepal (Panter-Brick 1997)
and Himalayan India (Himmelgreen et at. 1991). Himmelgreen et at. (1991), in fact,
argue that the assumption of female nutritional disadvantage in northern India should
be re-evaluated given the lack of gender differences in their findings. There is
evidence from an Andean community in Peru that infants and pre-school children of
both genders are buffered by nutritional stress due to child preferential food allocation
within the household (Leonard 1991). Thus, gender differences may be more
pertinent to adolescent age groups than infants and pre-school children.
An investigation of urban-rural differences in Nepal is made in comparing the
results from this peri-urban sample to those from the 1995 Nepalese National Health
Survey (UNICEF 1996). The children in this peri-urban sample, at least for those
over 12 months of age, have a prevalence of stunting in the same range as urban
Nepalese children; both of these groups have a lower prevalence of stunting than the
rural Nepalese children. This points to the possibility of slightly better linear growth
status for the peri-urban children, a result which is often found for comparisons
between urban and rural children in other low-income countries (WHO 1988,
Williams 1990).
One way that growth retardation may manifest itself differently in rural and urban
environments in Nepal is in terms of seasonal variation. However, like the results of
seasonal growth variation among rural Nepalese children found by Costello (1989)
and Panter-Brick (1997), the most pronounced effect of seasonality was the
significantly lower mean WHZ scores during the summer months. Costello (1989)
attributes this drop in weight to the dearth in food availability before the harvest;
Panter-Brick (1997) examines a combination of factors including harvest cycles and
child illness. Given that the peri-urban children in this sample are completely
dependent on the urban food markets, i.e. there is no pre-harvestlpost-harvest
difference in food availability, it is most likely that seasonal differences are not due
to changes in food consumption. Rather, it is probably associated with the increased
incidence and severity of illness, particularly gastrointestinal infections, during the
hot and monsoon seasons. The peak in diarrhoeal diseases during these seasons is
documented in other low-income countries located in tropical regions of the world
(Rowland and Barrell 1980, Rowland 1983).
GROWTH STUNTING IN NEPALESE CHllDREN 265
5. CONCLUSION
This paper examined linear growth retardation or stunting among a particular group
of children residing in peri-urban Kathmandu, Nepal. The two main goals were: 1) to
examine linear growth faltering in relation to other types of body growth; and 2) to
evaluate whether growth among children in peri-urban Nepal is uniquely different
from that among children in both urban and rural Nepal.
It appears that the process of stunting is very different from being stunted.
Decreased growth velocity occurs from infancy into the second year of life; however,
the children do not appear to be stunted (as defined by the -2 SD cut-off) until they
are over 24 months of age. This underscores the importance of examining factors
that affect stunting - such as macro and micro-nutrient deficiencies as well as
infectious disease - early on in infancy, before the signs of stunting, in terms of
cross-sectional growth measures, actually surface. Other anthropometrics such as
weight and arm measurements shed light on linear growth faltering. Although for the
most part among the sample wasting is not an issue, there is a marked increase in
low mean WHZ among 12 to 24 month old children, suggesting that acute weight
loss may precede or accompany linear growth retardation. Low mean upper arm
muscle mass (ZAM) mirrors the state of being stunted in this sample, indicating
chronic nutrient deprivation. Thus it is important to examine more than just arm
circumference to get a sense of growth deficits among children experiencing growth
stunting.
Comparisons between this peri-urban sample of children and other rural and urban
samples in Nepal show that the prevalence of stunting, although very high among
the peri-urban children, is more in line with urban than rural children. Both urban
and peri-urban children from six to 36 months of age have a significantly lower
prevalence of stunting than rural children. Despite problems of peri-urban
settlements such as lack of sanitation and poverty, there are still some benefits
accrued to children in terms of growth and development when living in an urban
environment compared to a rural village. Interestingly, despite differences in food
procurement between rural and peri-urban people, there are still seasonal differences
in growth among this peri-urban sample of children. Like in rural populations, mean
WHZ was lower in the hot and monsoon seasons compared to the cold season.
Seasonal differences in weight gain and loss are probably more related to infectious
disease than to seasonal differences in food consumption.
Finally, despite the fact that this sample of children is a unique sub-population
living in a particular peri-urban location in Nepal, it is striking how similar the
phenomenon of stunting is among children living in different locations around the
world. The common denominator among all of these groups is poverty.
6. REFERENCES
Allen, L.H., 1994, Nutritional influences on linear growth: a general review. European Journal of
Clinical Nutrition, 48 (Supp!. I), S75-S89.
Anderson, M., 1979, Comparison of anthropometric measures of nutritional status in preschool children
in five developing countries. American Journal of Clinical Nutrition, 32, 2339-2345.
Baumgartner, R.N., Roche, A.F. and Himes, J.H., 1986, Incremental growth tables: supplementary to
previously published charts. The American Journal of Clinical Nutrition, 43, 711-722.
266 CHAPTER 20
Beaton, G.H., 1992, The Raymond Pearl Memorial Lecture, 1990: Nutrition research in human biology:
changing perspectives and interpretations. American Journal of Human Biology, 4, 159-177.
Brink, E.W., Khan, I.H., Splitter, J.L., Staehling, N.W., Lane, J.M. and Nichaman, M.Z, 1976,
Nutritional status of children in Nepal, 1975. Bulletin of the World Health Organization, 54, 311-
318.
Brockerhoff, M., 1995, Child survival in big cities: the disadvantages of migrants. Social Science and
Medicine, 40, 1371-1383.
Chen, L.C., Chowdhury, A. and Huffman, S.L., 1980, Anthropometric assessment of energy-protein
malnutrition and subsequent risk of mortality among preschool aged children. The American
Journal of Clinical Nutrition, 33,1836-1845.
Costello, A.M., 1989, Growth velocity and stunting in rural Nepal. Archives of Disease in Childhood,
64, 1478-1482.
Dettwyler, K.A., 1991, Growth status of children in rural Mali: implications for nutrition education
programs. American Journal of Human Biology, 3, 447-462.
Eveleth, P.B. and Tanner, J.M., 1990,Worldwide Variation in Human Growth (2nd ed.) (Cambridge:
Gittelsohn, J., Thapa, M. and Landman, L.T., 1997, Cultural factors, caloric intake and micronutrient
sufficiency in rural Nepali households. Social Science and Medicine, 44,1739-1749.
Harpham, T., Lusty, T. and Vaughan, P., 1988, In the Shadow of the City. Community Health and the
Urban Poor (Oxford: Oxford University Press).
Himrnelgreen, D.A., Dannenhoffer, R., Baht, I. and Lee, R.V., 1991, Anthropometric assessment of
nutritional status among highland Kashmiri children: reevaluating the assumption of female
nutritional disadvantage. American Journal of Human Biology, 3, 239-249.
Frisancho, A.R., 1990, Anthropometric Standards for the Assessment of Growth and Nutritional Status
(Ann Arbor: University of Michigan Press).
Gorstein, J., Sullivan, K., Yip, R., de Onis, M., Trowbridge, F., Fajans, P. and Clugston, G., 1994, Issues
in the assessment of nutritional status using anthropometry. Bulletin of the World Health
Organization, 72, 273-283.
Keller, W., 1988, The epidemiology of stunting In Linear Growth Retardation in Less Developed
Countries. Nestle Nutrition Workshop Series Volume 14, edited by J.C. Waterlow (New York:
Raven Press) pp. 17-40.
Leonard, W.R., 1991, Household-level strategies for protecting children from seasonal food scarcity.
Social Science and Medicine, 33, 1127-1133.
Leonard, W.R., DeWalt, K.M., Stansbury, J. P. and McCaston, M.K., 1995, Growth differences
between children of highland and coastal Ecuador. American Journal of Physical Anthropology,
98,47-57.
Lohman, T.G., Roche, A.F.and Martorell, R., 1988, Anthropometric Standardization Reference Manual
(Campaign, Illinois: Human Kinetics Books).
Martorell, R., Khan, L.K. and Schroeder, D.G., 1994, Reversibility of stunting: epidemiological findings
in children from developing countries. European Journal of Clinical Nutrition, 48 (Suppl. 1), S45-
S57.
Martorell, R., Leslie, J. and Moock, P.R., 1987, Characteristics and determinants of child nutritional
status in Nepal. The American Journal of Clinical Nutrition, 39, 74-86.
Naborro, D., Howard, P., Cassels, e., Pant, M, Wijga, A. and Padfield, N., 1988, The importance of
infections and environmental factors as possible determinants of growth retardation in children. In
Linear Growth Retardation in Less Developed Countries. Nestle Nutrition Workshop Series
Volume 14, edited by J.C. Waterlow (New York: Raven Press) pp. 165-184.
Neumann, e.G. and Harrison, G.G., 1994, Onset and evolution of stunting in infants and children.
Examples form the Human Nutrition Collaborative Research Support Program. Kenya and Egypt
studies. European Journal of Clinical Nutrition, 48 (Suppl. 1), S90-S102.
Panter-Brick, C., 1997, Seasonal growth patterns in rural Nepali children. Annals of Human Biology,
24,1-18.
Panter-Brick, C., Todd, A. and Baker, R., 1996, Growth status of homeless Nepali boys: Do they differ
from rural and urban controls? Social Science and Medicine, 43, 441-451.
Ricci, J.A. and Becker, S., 1996, Risk factors for wasting and stunting among children in Metro Cebu,
Philippines. American Journal of Clinical Nutrition, 63, 966-975.
Rowland, M.G.M., 1983, Epidemiology of childhood diarrhea in the Gambia. In Diarrhea and
Malnutrition edited by L.C. Chen and N.C. Scrimshaw (New York: Plenum Press) pp. 87-98.
Rowland, M.G.M. and Barrell, R.A.E., 1980, Ecological factors in gastroenteritis. In Disease and
Urbanization, edited by EJ. Clegg and J. P. Garlic (London: Taylor and Francis Ltd) pp. 21-35.
Sastry, J. G., Vijayaraghavan, K. and Rao, N. P., 1989, Indian preschool children - a profile of stunting
and wasting. Journal of Tropical Pediatrics, 35, 237-240.
UNICEF, 1992, Children and Women of Nepal. A Situation Analysis (Kathmandu: UNICEF - Nepal).
UNICEF, 1996, Nepal Multiple Indicator Surveillance Health and Nutrition - Cycle 1, January to March
1995. (Kathmandu: UNICEF - Nepal).
GROWTH STUNTING IN NEPALESE CHll.DREN 267
Victora, CG., 1992, The association between wasting and stunting: an international perspective. Journal
of Nutrition, 122, 1105-1110.
Waterlow, J.C., 1988,Observations on the natural history of stunting. In Linear Growth Retardation in
Less Developed Countries. Nestle Nutrition Workshop Series Volume 14, edited by J.C. Waterlow
(New York: Raven Press) pp. 1-16.
Waterlow, J.C., 1992, Protein Energy Malnutrition (London: Edward Arnold).
Waterlow, J.C, 1994,Causes and mechanisms of linear growth retardation (stunting). European Journal
of Clinical Nutrition, 48 (Suppl. I), S I-S4.
Watts, T., Ng'andu, N. and Wray, J., 1990, Children in an Urban Township in Zambia. A prospective
study of children during their first year of life. Journal of Tropical Pediatrics, 36, 287-293.
WHO, 1988, Urbanization and its Implications for Child Health (Geneva: World Health Organization in
collaboration with the United Nations Environment Program).
Williams, B.T., 1990,Assessing the health impact of urbanization. World Health Statistics Quarterly, 43,
145-150.
CHAPTER 21
ETHNIC AND SEX DIFFERENCES IN THE SKELIC

INDEX AMONG FInAN AND SAMOAN CHILDREN
T. SATAKEt, K. HATTORI~, E. KANAZAWAt
tDepartment of Anatomy, Nihon University School of Dentistry at Matsuda, Matsuda,

Chiba, Japan, jDepartment of Health and Physical Education, Ibaraki University,
Milo, Ibaraki, Japan
1. INTRODUCTION
Populations of the South Seas are broadly classified as Polynesian, Melanesian and
Micronesian (Campbell 1989). However, their origin is relatively homogeneous in
an archaeological sense (Gravelle 1979). The distance between Fiji (Melanesia) and
Samoa (Polynesia) is not great. However, the populations have been subjected to
very different processes of socio-cultural change in the past generation. This, in turn,
has affected the growth status of the children. Complex interactions between genetic
and environmental factors contribute to variation between populations. Comparisons
among South Seas populations provide an opportunity to consider this variation
especially in children.
The growth status of children in the South Seas area has been documented,
including studies in Fiji (Hawley and Jansen 1971, Lourie 1972, Clegg 1989) and
Samoa (Wigg 1978, Baker et al. 1986, Bindon and Zansky 1986). However,
anthropometric data on these ethnic groups are still quite limited.
This paper focuses on the Skelic Index, which assesses the relationship of lower
limb length to sitting height. The Skelic Index changes during prepubertal growth
(greater contribution of lower limb length) and during pubertal growth (greater
contribution of sitting height) (Cameron et al. 1982, Malina and Bouchard 1991).
This study contrasts the pattern of growth in the Skelic Index among Fijian and
Samoan children during childhood and adolescence, and examines ethnic and sex
differences in body size and the index in cross-sectional sample of Fijian school
children resident in an urban area and Samoan school children resident in a rural
area.
2. MATERIALS AND METHODS

The study was done in Fiji (Suva on the island of Viti Levu: an urban area) and
Samoa (Vaisala and Asau on the island of Savaii: rural areas) in 1994 and 1995,
respectively (Satake 1996, Satake et al. 1996). The Fijian sample excluded children
oflndian origin. The sample included 251 Fijian school children (128 boys and 123
269
270 CHAPTER 21
girls) and 192 Samoan school children (95 boys and 97 girls) 6+ to 17+ years of
age. Chronological age (decimal age) was computed from birth dates and the date of
observation. Stature was measured in the standard erect posture on barefoot subjects,
and sitting height was measured in the standard sitting posture on a table top. A
field anthropometer was used and measurements were recorded to the nearest
millimetre. All measurements were done by the same individual (TS). Lower limb
length was calculated by subtracting sitting height from stature, and the Skelic Index
was calculated as lower limb length/sitting height.
Comparisons of within ethnic groups and between sexes were carried out using
analysis of covariance with age as the covariate. Analyses focused on two specific
comparisons: first, comparisons between and within Fijian and Samoan children in
each sex: second, comparisons between boys and girls in each ethnic group. All
statistics were calculated using the SAS statistical package.
3. RESULTS
Tables are shown in the Appendix. Descriptive statistics for Fijian and Samoan
children are given in Tables 1 and 2, respectively. There are few significant (p<0.05)
mean differences between boys and girls in both Fijian and Samoan children (Tables
1 and 2). There is only one significant difference in the Skelic Index in Samoan (at
7+ years). On the other hand, Fijian boys and girls significantly differ at 7+, 8+,
12+, 14+, 15+, 16+, and 17+ years of age.
Mean differences between Fijian and Samoan children within each sex (Table 1)
are small and very few are significant in stature, sitting height and lower limb
length. However, ethnic differences within sex in the Skelic Index are often
significant. The Skelic Indices of Samoan boys are significantly greater than those of
Fijian boys at 6+, 7+, 8+, 9+ and 13+ years of age. The same trend is evident in
girls, 6+, 7+, 8+, 9+, 13+ and 16+ years of age.
Figures 1 to 4 are scatter plots of stature, sitting height, and the Skelic Index by
chronological age. As expected, stature and sitting height increase with age (Figures
1 and 2). The Skelic Index shows ethnic variation by ethnicity and sex, respectively
(Figures 3 and 4).
Results of the analyses of covariance, with age as the covariate, are summarised
in Table 3. There are no significant interactions between ethnicity and sex in the
anthropometric dimensions and the Skelic Index. There are also no significant ethnic
differences in stature and lower limb length. Sitting height and the Skelic Index, in
contrast, show significant ethnic differences. Sex differences are significant for all
variables, except the sitting height.
Directions of the differences for Fijian and Samoan children are shown in Table
4, which gives age-adjusted means. Fijian children have a significantly longer
sitting height than Samoan children, whereas Samoan children have significantly
greater Skelic Index. And comparisons of the age-adjusted means between boys and
girls show that boys are significantly taller and have a longer lower limb length and
greater Skelic Index than girls. There are no significant sex differences in sitting
height.
SKELIC INDEX IN F IJIAN AND S AMOAN CHILDREN 271
200
~
oa
150 f-
.. . 1 .:
if~*~iof. 0
•
0
,.~ t4
•.
0
Stature
6
(,)
I
Q)~ 100
N
U5 t ~--j
50 f-
MIJI" ~ Sitting height
o Boys
• Girls
o t
4 6
I I
8
I
10
I
12
I
14
I
16 18
I
20
Age, years
Figure 1. Age changes and sex differences in stature and sitting height in Fijian school
children
200
~
150 f-
0.. ~ o ~~
o ~ ~ •• ~"oooot • , •
•
.1. 0 •
0
au
,. • r- • 0 ~ Stature
~~
..... 100 ~
V)
.. ~••~ i""'~~
S· .
.~ro
Ithng height
50 f- o Boys
• Girls
o t
4 6
I I
8
I
10
I
12
I
14
I
16 18
I
20
Age, years
Figure 2. Age changes and sex differences in stature and sitting height in Samoan
school children
272 CHAPTER 21
110
.g
.S
~
u
:..::l
100
90
~~,
o ~~ ~o~
~.
..
••
~
~_
tIP.
••• 0. Z' J
J .
~ 80 8 0
U)
Fijian children
70
120
o Boys
~ 110
t ..,.o.~t9.~~~~ i9.~oo~8e ~.
11)
• Girls 0 0
]
u
100 • 0 0 ~;.~
:..::l 90 •• lj 0 " • • 0
~
U)
• • o. • \
80 Samoan children
70
4 6 8 10 12 14 16 18 20
Age, years
Figure 3. Age changes and sex differences in the Skelic Index in Fijian and Samoan
school children
110
° eO •
.g • JI'~~' o ~. ~ ~
~
100 .' .,.1....
.S
u
:..::l
90
t80':·~i~ I~· ~~~.~.o.JI, fiJ
~ 80 8 0 o
° 00
U)
Boys
70
120
o Fijian
110
•• •.01,
~
.g • Samoan
.S 100 • c\t.. it~ ..• ~
~~ co'·~~ ~ O ~g"~ _D
u · 0
:..::l 90
~
U) ou
- lj 0 0 ~ 0Q:l 0 ~ ~ ~
80 Girls 0
70 I~~_ _~_ _~_ _~~_ _~_ _~_ _~~_ _- L_ _~_ _~~_ _~_ _~~
4 6 8 10 12 14 16 18 20
Age, years
Figure 4. Age changes and ethnic differences in the Skelic Index in boys and girls
SKELIC INDEX IN FIJIAN AND SAMOAN CHILDREN 273
Comparison of anthropometric dimensions and the Skelic Index by sex within

the Fijian and Samoan samples are summarised in Table 5. Sex differences between
age-adjusted means in the Skelic Index are greater in Fijian than in Samoan children
(Table 6).
4. DISCUSSION
Although numbers of children in each age group are limited, the data suggest several
trends in the comparison of growth in the Fijian and Samoan populations. Fijian
and Samoan children have similar statures, but Fijian children, especially in girls,
tend to exceed Samoan children in sitting height and Samoan children tend to
exceed Fijian children in lower limb length (Tables 1 and 2). The Fijian sample is
from an urban area and Samoan sample is from a rural area. It is reasonably well
documented that an urban children are larger in body size than the children living in
a rural area (Dasgupta and Das 1997, Bogin 1999). Compared with both the same
ethnic samples of different areas from the present study, results may differ from this
study.
The data of Samoan children are from rural villages on the island of Savaii.
Bindon and Zansky (1986) reported on growth changes of stature and body weight
on Western Samoa for similar rural villages, while Clegg (1989) reported several
physical characteristics of children in Suva, Fiji. Compared to these earlier studies,
statures in the present sample are, on average, greater in children of both ethnic
groups at almost ages. The secular increase in stature has been widely reported in
many different ethnic groups (Eveleth and Tanner 1990, Ulijaszek 1993). This
results thus suggest a similar secular change in stature in both ethnic groups.
The Skelic Index is, on average, greater in Samoan than in Fijian children in
each sex (Tables 1 and 2). Growth changes of the Index are, however, different in
both ethnic groups. Differences in relative leg length are evident among other ethnic
groups (Krogman 1970, Malina et al. 1974, Meredith and Spurgeon 1976, Meredith
1979, Buschang et al. 1986, Eveleth and Tanner 1990, Nath et al. 1991, Satake et
al. 1995, Spurgeon et al. 1997).
The Skelic Indices of boys are greater than those of girls in each ethnic group.
Absolute differences in the Skelic Index of boys in Fijian children increase in the
prepubertal period. And after that Fijian children, but not Samoan children, keep
differences between both sexes (Tables 1 and 2, and Figure 3). The sex difference in
the Skelic Index is significant (Tables 3 and 4) and the sex difference in age-adjusted
means is greater in Fijian than in Samoan children (Table 6). The results suggest
that during adolescence, Samoan children are proportionately similar, while Fijian
girls have relatively shorter lower limb length than boys. In both ethnic groups,
however, girls have a proportionately shorter lower limb length than boys as
Meredith and Knott (1938) and Buckler (1990) have also shown. Satake, Oh, Boo
and Hattori (1995) also showed sexual dimorphism in age changes of the Skelic
Index among American, Japanese and Korean children. The relationship between
boys and girls in age change of the Skelic Index differed within each ethnic group.
Lower limb length experiences its adolescent growth spurt before sitting height
(Tanner et al. 1982). And, the timing of prepubertal and pubertal growth in
anthropometric dimensions varies among different ethnic groups and between both
sexes (Malina et al. 1988). As a result, it is anticipated that age change in the Skelic
274 CHAPTER 21
Index would show different trends among ethnic groups and between sexes. Sexual
dimorphism in age changes of the Skelic Index may be useful in understanding
growth differences among ethnic groups.
The lack of significant interactions between ethnicity and sex in any of the
anthropometric dimensions or the Skelic index is of interest (Table 3). It suggests
that the ethnic influence on the anthropometric characteristics and the index is
independent of sex differences, and also that sex differences influence individuals from
both ethnic groups in the same way.
Ethnic comparisons within sex show that Fijian children were a significantly
longer sitting height than Samoan children, and that Samoan children have a
significantly greater Skelic Index (Table 3 and 4). The presence of significant ethnic
differences in sitting height and the lack of significant differences in stature and lower
limb length contribute to the significant difference in the Skelic Index.
Regardless of ethnicity, sex differences in stature, lower limb length and the
Skelic Index are significant. Sitting height, on the other hand, dose not differ
between boys and girls (Tables 3 and 4). The lack of significant sex differences in
sitting height and the presence of significant differences in stature and lower limb
length contribute to the significant sex difference in the Skelic Index. As such,
growth of sitting height is an important variable both in ethnic differences between
Fijian and Samoan children and in sexual dimorphism of linear body proportions.
Acknowledgement. The authors express their sincere gratitude to Dr. R.M. Malina
of Michigan State University for variable comments and helpful advice in preparation
of the English manuscript. The present study was financially supported by Nihon
University.
5. REFERENCES
Baker, P.T., Hanna, J.M., and Baker, T.S., 1986, The changing Samoan, (New York: Oxford University
Press).
Bindon, J.R., and Zansky, S.M., 1986, Growth patterns of height and weight among three groups of
Samoan preadolescents. Annals of Human Biology, 13, 171-178.
Bogin, B., 1999, Patterns of Human Growth (Cambridge: Cambridge University Press)
Buckler, J.M., 1990, A longitudinal study of adolescent growth. (London: Springer Verlag).
Buschang, P.H., Malina, R.M., and Little, B.B., 1986, Linear growth ofZapotec school children: growth
status and yearly velocity for leg length and sitting height. Annals of Human Biology, 13,225-234.
Cameron, N., Tanner, J.M., and Whitehouse, R.H., 1982, A longitudinal analysis of the growth of limb
segments in adolescence. Annals of Human Biology, 9, 211-220.
Campbell I.C., 1989, A History of the Pacific Island. (New Zealand: Canterbury University Press).
Clegg, EJ., 1989, The growth of Melanesian and Indian children in Fiji. Annals of Human Biology, 16,
507-528.
Dasgupta, P., and Das, S.R., 1997, A cross-sectional growth of trunk and limb segments of the Bengali
boys of Calcutta. Annals of Human Biology, 24, 363-369.
Eveleth, P.B., and Tanner, J.M., 1990, Worldwide variation in human growth. (Cambridge: Cambridge
University Press).
Gravelle, K., 1979, Fiji's Times A History of Fiji. (Fiji: The Fiji Times Limited)
Hawley, T.G. and Jansen, A.AJ., 1971, Height and weight of Fijians in coastal areas from one year to
adulthood. New Zealand Medical Journal, 73, 346-349.
Krogman, W.M., 1970, Growth of head, face trunk, and limbs in Philadelphia white and negro children
of elementary and high school age. Monographs of the Society of Research in Child Development,
35: serial No. 136, 1-80.
Lourie, lA., 1972, Anthropometry of Lau Islanders, Fiji with a note on their color vision. Human
Biology in Oceania, 1, 273-277.
Malina, RM., and Bouchard, C., 1991, Growth, Maturation, and Physical Activity. (Champaign, IL:
Human Kinetics).
Malina, R.M., Bouchard, c., Beunen, G., 1988, Human growth: Selected aspects of current research on
well-nourished children. Annual Reviews Anthropology, 17, 187-219.
Malina, RM., Hamil, P.V.V., and Lemeshow, S., 1974, Body dimensions and proportions, White and
Negro children 6-11 years, United States. Vital and Health Statistics, Series 11, Number 143, 66p.
Washington, D.C.: National Center for Health Statistics.
Meredith, H.V., 1979, Relationships oflower limb height to sitting height in black populations of Africa
and United States. American Journal of Physical Anthropology, 51, 63-66.
Meredith, H.V., and Knott, V.B., 1938, Changes in body proportions during infancy and the preschool
years III. The Skelic index. Child Development, 9, 49-62.
Meredith, H.V., and Spurgeon, lH., 1976, Comparative findings on the skelic index of black and white
children and youths residing in south Carolina. Growth, 40, 75-81.
Nath. S., French, K.E., and Spurgeon, J.H., 1991, Somatic comparisons: Baiga and Gond males of
Madhya Pradesh, India. American Journal of Human Biology, 3, 281-287.
Satake T., Yoshida K., Kanazawa E., Hattori K., 1996, Ethnic difference in the Skelic Index among
school children of Western Samoa and Fiji. Auxology 3, 28-30 (in Japanese).
Satake, T., 1996, Anthropological Study on Food intake and Oral Morphology of Polynesians in Special
Oceanic Environment. Final Report of the Interdisciplinary General Joint Research Grant for
Nihon University 1994-1996, pp. 106-118.
Satake, T., Oh, S.I., Boo, K.W., and Hattori, K., 1995, Growth of Korean schoolchildren using Skelic
Index. Bulletin of College of General Education Ibaraki University, 28,257-268(in Japanese with
English summary).
Spurgeon, lH., Kim, K-B., French, K.E., Giese, W.K., 1997, Somatic comparisons at three ages of
south Korean males and males of other Asian groups. American Journal of Human Biology, 9, 493-
503.
Tanner, J.M., Hayashi, T., Preece, M.A., Cameron, N., 1982, Increase in length ofieg relative to trunk
in Japanese children. Annals of Human Biology, 9, 411-423.
Ulijaszek, S.J., 1993, Evidence for a secular trend in heights and weights of adults in Papua New
Guinea. Annals of Human Biology, 20, 349-355.
Wigg, N .R, 1978, The anthropometry of Western Samoan pre-school children. New Zealand Medical
Journal, 87, 172-175.
276 CHAPTER 21
APPENDIX
Table 1. Descriptive statistics for stature, sitting height, lower limb length and the
Skelic Index by age and sex in Fijian school children (significances of t-Student test)
Boy"s (]V = 1282

Stature, cm Sitting height, Lower Limb Skelic Index
cm length, cm
Age, n Mean SD Mean SD Mean SD Mean SD
years
6+ 11 121.0 4.42 65.6 2.93 55.4 2.20# 84.5 3.64##
7+ 10 126.3 5.17 66.1 3.40 60.2 2.21 91.1 3.47*#
8+ 13 132.6 5.91 68.7 3.16 63.9 3.15 93.0 3.09*#
9+ 8 137.2 7.60 73.1 4.99 64.1 3.56 88.0 5.23##
10+ 2 140.3 1.41 72.3 0.78 68.1 0.64 94.2 0.13
11 + 10 146.1 4.67 75.0 2.39 71.1 2.94 94.9 3.44
12+ 20 151.3 5.90* 77.9 3.45* * 73.4 3.34 94.4 4.15*
13+ 9 157.6 10.90 81.1 4.90 76.5 6.59 94.3 4.88#
14+ 20 166.1 9.89** 84.4 4.71 81.7 5.61** 96.8 3.69**
15+ 12 169.7 8.35** 86.1 4.46 83.6 5.05** 97.2 5.4 7* *
16+ 8 174.1 5.54* 89.3 1.96* 84.8 4.53* 95.0 4.70*
17+ 5 173.2 4.89* 88.4 1.90 84.8 3.94* 96.0 4.25*
Girls (N = 1232
em length, cm
years
6+ 8 119.4 2.91 64.2 1.43# 55.2 1.95 86.0 2.75#
7+ 11 125.6 3.63 66.9 1.97 58.7 2.03 87.7 2.40##
8+ 9 129.8 4.11 68.2 2.25## 61.6 1.96 90.3 1.33##
9+ 10 136.1 5.58 71.0 3.32 65.0 2.65 91.6 2.94##
10+ 2 145.1 0.35 75.5 1.70 69.6 1.34 92.2 3.85
11+ 12 146.1 5.37 76.0 3.75 70.1 2.46 92.5 4.00
12+ 16 156.3 5.97 81.9 3.71 74.5 3.22 91.1 4.15
13+ 13 160.6 8.70 83.6 5.21 77.0 4.48 92.2 4.77#
14+ 17 158.4 5.19# 82.3 3.66 76.0 2.31 92.4 3.66
15+ 11 161.6 3.92 84.8 1.87 76.8 2.89 90.5 3.34
16+ 6 163.2 8.70 86.1 3.35 77.2 5.59 89.6 3.74#
17+ 8 166.0 5.13 87.2 1.78 78.8 3.95 90.3 3.88
* P < 0.05 (Sex differences)

** p < 0.01
# P < 0.05 (Ethnic differences)
## p < 0.01
Table 2. Descriptive statistics for stature, sitting height, lower limb length and the
Skelic Index by age and sex in Samoan school children (significances of t-Student test)
Bo.£s (]'I = 951

cm length, cm
years
6+ 9 121.9 4.31 * 63.6 2.41 58.3 2.41 * 91.8 3.20
7+ 7 125.3 3.71 64.2 1.53 61.2 2.83 95.3 4.10*
8+ 9 133.5 6.00 68.1 2.96** 65.4 3.29 96.0 2.43
9+ 11 l35.9 4.23 69.2 2.57 66.7 1.98 96.4 2.43
10+ 4 l38.2 3.93 70.9 3.48 67.4 0.73* 95.2 4.36
11+ 9 144.4 5.11 73.3 2.63 71.1 2.88 97.0 2.77
12+ 5 150.1 6.44 77.1 3.74 73.0 3.13 94.8 3.10
l3+ 5 157.1 8.30 78.6 4.00 78.5 4.66 99.8 3.30
14+ 7 169.3 4.60 86.2 3.09 83.2 3.41 * 96.6 5.36
15+ 7 169.0 7.65* 86.9 7.62* 82.2 4.73 94.6 4.70
16+ 16 172.3 3.51** 87.8 2.53** 84.5 2.28** 96.4 3.74
17+ 6 171.2 5.43 88.6 3.07 82.6 3.53 93.3 4.29
Girls (N = 972
cm length, cm
years
6+ 9 117.0 4.10 61.5 2.29 55.5 2.51 90.3 4.07
7+ 9 125.0 6.52 65.2 2.97 59.8 3.60 91.6 1.92
8+ 5 128.0 2.10 64.4 0.86 63.6 2.56 98.6 2.56
9+ 11 136.9 6.76 70.0 3.46 66.9 3.60 95.6 2.86
10+ 6 143.7 4.21 73.3 2.72 70.4 2.03 96.1 3.04
11+ 4 145.1 l3.42 73.5 6.41 71.6 7.31 97.3 4.23
12+ 3 156.3 9.49 80.4 5.73 75.9 4.03 94.5 3.11
13+ 9 157.2 5.80 80.1 2.98 77.1 3.50 96.3 3.59
14+ 5 163.7 4.07 85.3 3.20 78.4 3.14 92.0 3.14
15+ 13 162.4 5.15 83.9 2.52 78.5 3.87 93.7 4.68
16+ 19 165.6 4.49 85.1 2.61 80.5 2.91 94.6 3.70
17+ 4 164.0 7.06 85.9 3.23 78.1 5.46 90.1 6.42
* P < 0.05
** P < 0.01
278 CHAPTER 21
Table 3. Results of analysis of covariance, with age as the covariate, of stature, sitting
height, lower limb length and the Skelic Index of Fijian and Samoan school children
MS F p
Stature Ethnie(E)a 108.07 2.52 NS
Sex(S)a 682.55 15.92 <0.01
E*Sa 3.68 0.09 NS
Residual(R)b 42.86
Sitting height Ethnie(E)a 263.89 22.39 <0.01
Sex(S)a 12.05 1.02 NS
E*Sa 17.45 1.48 NS
Residual(R)b 11.79
Lower limb Ethnie(E)a 34.21 2.22 NS
length Sex(S)a 513.24 33.32 <0.01
E*Sa 5.10 0.33 NS
Residual(R)b 15.40
Skelie Index Ethnie(E)a 824.52 45.19 <0.01

Sex(S)a 573.36 31.43 <0.01
E*Sa 46.03 2.52 NS
Residual{R2 b 18.24
adf= 1
bdf= 438
Table 4. Age-adjusted means of stature, sitting height, lower limb length and the Skelic
Index of Fijian and Samoan school children
Ethnicity Sex
Fijian Samoan Boys Girls
Stature, em 149.6 148.6 150.4 147.8
Sitting height, em 77.8 76.2 77.2 76.8
Lower limb length, em 71.8 72.3 73.1 71.0
Skelie Index 92.1 94.9 94.6 92.4
Table 5. Results of analyses of covariance, with age as covariate
Fiji Samoa
MS F P MS F P
Stature Sex(S)a 339.09 7.00 <0.01 346.87 9.70 <0.01
Residual(R)b 48.56 35.77
Sitting height Sex(S)a 0.05 0.00 NS 31.20 3.16 NS
Lower limb length Sex(S)a 347.74 20.42 <0.01 170.02 12.94 <0.01
Skelie Index Sex(S)a 519.69 28.31 <0.01 103.23 6.27 <0.01
Residual~R2b 18.36 16.47
adf= 1
bdf= 248 Fijian, 189 Samoan
Table 6. Age-adjusted means of stature, sitting height, lower limb length and the Skelic
Index of Fijian and Samoan boys and girls
Fijian Samoan
Boys Girls Boys Girls
Stature, em 150.5 148.2 150.3 147.6
Sitting height, em 77.7 77.7 76.9 76.0
Lower limb length, em 72.8 70.5 73.5 71.6
Skelie Index 93.6 90.7 95.6 94.2
CHAPTER 22
SKELETAL MATURITY IN CHILDREN OF MIXED

AMERICAN AND JAPANESE PARENTAGE AS
ASSESSED BY THE TW2-METHOD
K. KIMURA
Kimura Auxological Institute, Sayama, Japan
1. INTRODUCTION
The present study forms a part of comprehensive anthropological investigation on
children of mixed American and Japanese parentage, carried out from 1949 to 1965
longitudinally. They were all the offspring of American Servicemen of the
Occupation Forces and Japanese women, born after the World War II. They were
cared for at the Elizabeth Sanders Home in Ohiso and the Boy's Town in Yamato,
Kanagawa. For the study, according to the father's racial characteristics, they were
divided into two groups of mixed Afro- and Euro-American and Japanese parentage.
It was rather difficult to obtain accurate information about the father's racial
background. They were placed into each group mainly on the basis of the mother's
statement, and also on their outstanding physical traits, when obvious.
The late Professor Suda has stated a general introduction (Suda 1968). Some co-
workers charged of the survey of their deciduous teeth (Hanihara 1954-57, 1963,
1965, 1968), oral zone (Hojo 1964), hair, iris and skin colors (Goto 1961, Honda
1958, Michibe 1961, Omoto 1968), facial muscles (Kondo et al. 1968) and finger
and palm prints (Kimura 1974a, 1974b). The others examined the urinary BAIB
excretion (Ishimoto 1968), blood groups (Matsuzawa and Kitamura 1956, Suzuki
and Matsuzawa 1956, Kitamura and Suzuki 1956), and subcutaneous fat (Kohara
1968). Physical measurements were measured longitudinally (Hoshi 1959, 1970,
Suda et at. 1965, 1968, 1973, 1975, 1976a, 1976b). I had taken radiographs ofthe
hand and wrist since 1958, and reported preliminarily on the skeletal maturity of the
children of only Euro-American and Japanese parentage, using the Oxford method
(Acheson 1954) and the TWI and TW2 method (Tanner et al. 1962, 1983, Kimura
1971, 1972, 1976a, 1994). I have also studied on the non-epiphyseal notch (Kimura
et at. 1982) and growth of the second metacarpal of both groups of mixed American
and Japanese parentage (Kimura 1996). The present paper examines skeletal maturity
of the children of mixed Afro- and Euro-American and Japanese parentage, compared
with the Japanese, using the TW2 method (Tanner et al. 1983).
281
© 200t Kluwer Academic Publishers.
282 CHAPTER 22

Radiographs of the right hand and wrist were taken in the children of mixed
American and Japanese parentage at the Elizabeth Sanders Home and the Boy's
Town twice a year, in each spring and autumn from 1958 to 1965, longitudinally.
The majority of children entered the homes under 12 months of age (about 30%),
and at the latest, under 3 years of age (about 70%). In general, they could stay at the
home until 15 years of age and occasionally until 18 years of age depending on the
situation. They were possibly cared under almost a similar living condition as the
average in Japan. The series was occasionally interrupted for some of the children for
their personal reasons. Some left the home in mid-series, because some of the
mothers later became able to take over their children, or childless American couples
offered to adopt suitable boys and girls as their children. Therefore, the sample could
not be accurately dealt with longitudinally. At the same time, Japanese children
orphaned in the war were also surveyed in the same way at the Children's Town at
Kasukabe, Saitama.
The mixed longitudinal data were obtained from 300 boys and 157 girls of
mixed Afro-American and Japanese parentage, 447 boys and 206 girls of mixed
Euro-American and Japanese parentage, and from 234 boys and 106 girls of the war
orphan, from 3 to 18 years of age. Only the data of samples taken in spring were
used for the present paper. In addition, as a control, the data obtained from 240 boys
and 332 girls of the Japanese, from 7 to 18 years of age, were used. They were
investigated in a lowland area of Tokyo in early summer, 1962. The annual income
is surely high in Tokyo (158.l% in 1965) compared with the average in Japan. The
annual income rate tends to be higher in the hilly area than in the lowland of Tokyo.
For example, it was 164.0% in the hilly section and 114.9% in the lowland of
Tokyo (143.7%) by the data in 1982.
The skeletal age was assessed for all children based on the 20-bone, RUS and
carpal age systems of the TW2 method (Tanner et al. 1983). All the data were
statistically analysed with a Facom M400 computer using SAS software (1985).
3. RESULTS
Figure 1 shows the pattern of differences between the means of skeletal age (20-bone,
RUS and carpal) and chronological ages (SA-CA) plotted against the means of each
chronological age in the children of mixed Afro- and Euro-American and Japanese
parentage and the war orphans, including the Japanese control. The mid-age method
was used fro chronological age grouping: for example, the 7 years group contains the
children aged 6.5 to 7.4 years at the date of investigation.
Common to each skeletal age, the control was somewhat delayed in maturity
before 12 years in boys and 9 years, respectively, and then rather advanced in both
sexes, compared with the British standards. The war orphan became gradually
delayed in maturity after 10 years of age in the boys, especially in the carpal age, but
followed roughly a similar pace of maturity in the girls, compared with the control.
The children of mixed Afro-American and Japanese parentage were slightly delayed
until 9 years of age and then advanced in the boys, and always advanced after 6 years
of age, markedly at 10 years, in the girls, with the British standards. The children of
mixed Euro-American and Japanese parentage were delayed until 10 years of age and
SKELETAL MATURITY IN CHILDREN OF MIXED PARENTAGE 283
then markedly advanced in the boys, and comparable at 6-9 years of age and then
markedly advanced in the girls, with the standards. They all arrived at full maturity
l
for each skeletal age at the same age for each TW2 age, as the British standards.
BOYS
§2 rO- bone
age ~!~
}t
,0 ~ _-o "
~ -~.,
~- .. ..
~~~
'7o~~
l RUS age
~ G- - ,,-' ". .•.• • 0 •
-2
2 , Carpal age
«
C? 0 I ~ ......-- __ .~o ~ ' A' " ' i '~"'"
~
-2 4 5 6 l' 8 9 1 '0 1 '1 1'21'3141'5 1'6 171819
l
Chronological age, years
GIRLS
§2r~
..
«W,0 0- -
'6--' .
" .,
l
o'
••
§-~t ~
~
'
.. 0.'
-2
, , -- .. '
2 , Carpal age _______ • Am. blacll - Japo

- - e - - Am. wh ite - Jap.
«0 I
"'";' -.....""'-- o. ~.r=
. . __ eo' ___ >;;;0.
ft---- Japoorphans
0 • 0
q.". • .
«
Ir
C/) .... • .... Jap. Tokyo
-2 ' p
4
Figure 1. Age changes of the difference between TW2 skeletal and chronological ages
in the children of mixed American and Japanese parentage and the Japanese war-
orphans and controls
284 CHAPTER 22
The children of mixed Afro-American and Japanese parentage were significantly

more advanced than the Japanese at 8-13 years for all skeletal ages in the boys, and
at 7-12 years for the 20-bone and RUS ages and at 7-8 years for the carpal age in the
girls, at p<0.05. The children of mixed Euro-American and Japanese parentage were
so for the 20-bone and RUS ages at 9-13 years in the boys and at 7-12 years in the
girls significantly at p<0.05. The children of mixed Afro-American and Japanese
parentage were significantly more advanced than those of mixed Euro-American and
Japanese only for the carpal age at 10-13 years of age in the boys and at 9-10 years of
age in the girls at p<0.05.
4. DISCUSSION
Most variations in skeletal maturation are genetically determined (Pryor 1907, 1936,
1939, Hess and Abramson 1933, Buschke 1935, Key 1936, Rigler 1938, Reynolds
1943, Gam et al. 1961a, 1961b, 1962a, 1962b, 1963, 1969, Hertzog et al. 1969,
Kimura 1962, 1981, 1983, Sklad 1973). However population differences in physique
and growth may reflect not only genetic factors, but also environmental ones.
4.1 Assessment of skeletal maturity

The standards by Todd (1937) and Greulich and Pyle (1950, 1959) were derived
from the Euro-American children of high socio-economic status living in Cleveland
(Brush Foundation Series) born in 1917-30 and -42, respectively. The TW2
standards were derived from the British children of lower and middle socio-economic
status living in various parts of England in the 1950s. As for the TW2 method, it
has been pointed out that the RUS system had little problems to assess skeletal ages
compared with the carpal (Ashizawa 1974, Borns and Moshang 1977, Wenzel and
Melsen 1982). Carpal bones should not be considered in assessing skeletal age
(Johnston and Jahina 1965) or be assessed only from 7 to 13 years in boys and from
4 to 10 years in girls (Roche 1989). The 20-bone system may be the next best way
as the representative of TW2 method at the present time (Tatara et at. 1985). All of
these skeletal ages were assessed and examined in the present paper.
Reproducibility of ratings by the same observer ranged from 72% to 87%, but
differences did not exceed one stage except in rare instances (Medicus et al. 1971).
Mean intra-assessor differences were 0.1 years for the RUS skeletal age and 0.3 years
for the carpal (Beunen and Cameron 1980). In Japanese samples, including inter- and
intra-assessor errors, the average coincidence of the RUS estimation was 81.6% and
of the carpal 77.7% (Satoh et al. 1993). However, according to Snedecor and
Cochran (1980), the absolute inter-assessor differences (0.20 years) were small as the
intra-assessor differences (0.08 and 0.13 years), and the coefficients of reliability were
high (inter-assessor, 99.7%; intra-assessor, 99.96%). Recently, Tanner and Gibson
(1994) developed a computer-assisted skeletal age (CASA) system, and Tanner et
al. (1994) and Yoshida (1997) have reported the reliability and validity of the new
system compared with the manual method. The short system, based on the third
phalanges and metacarpal and the radius and ulna, is useful and reliable for
assessment of skeletal maturity, as it is simple and not influenced by x-ray
conditions (Yoshida 1997). In the present study, the skeletal maturity was all
assessed by the author alone on each radiograph of all samples twice and more at
intervals of a few weeks and more. In the longitudinal samples, each individual
series was assessed successively.
4.2 Extrinsic factors

Among the more important environmental influences on growth are nutrition,
disease, socio-economic status, urbanisation, physical activity, psychological stress,
season of year, and climate (Eveleth 1978a). It was suggested that long bones react
more sensitively to extrinsic disturbances than carpal bones (Dreizen et al. 1954)
and that certain carpal bones are greatly influenced by genetic factors (Gam et al.
1963). The retarding influence on the growing skeleton is most pronounced before
menarche (Dreizen et al. 1967). Malnutrition and severe disease exert possibly
wrong influence on skeletal maturation (Abbott et al. 1950, Dreizen et al. 1954,
1957, 1958, 1964, 1967, Jones and Dean 1956, Douglas and Blomfield 1958,
Acheson 1960, Masse and Hunt, jr. 1963, Prader et al. 1963, Gam and Clark 1975).
The samples used for this study were all from healthy children based on a periodic
medical examination and investigated at places in and near Tokyo in spring or early
summer.
In general, growth of infants from a variety of ethnic and socio-economic classes
is remarkably similar during the first 6 months of age. After 6 months children from
the higher and lower socio-economic status become divergent in size. For the boys,
the difference in rate of growth between both higher and lower socio-economic
groups generally increase with age and becomes greatest at the time of the adolescent
spurt in height and weight. But, for the girls, only growth increments in height
show an increasing disparity with age between groups (Habicht et al. 1974, Van
Loon et al. 1986). Skeletal maturation also is generally slower in lower socio-
economic groups compared with upper and middle socio-economic groups (Eveleth
and Tanner 1976). The same fact was observed in the British (Kristmundsdottir et al
1984), Danish (Andersen 1968), Baghdad Arab (Shakir and Zaini 1974), Delhi
Indian (Maniar et at. 1974), Hong Kong Chinese (Low 1972) and Guatemalan
(Bogin et at. 1989). The Japanese series does not necessary show the same skeletal
maturity in details in various regions (Kimura 1976b, 1977, Takai and Akiyoshi
1983, Takai et al. 1984, Ashizawa 1993). Significant differences were found in
maturity of the TW2 ages and growth of the second metacarpal between the children
in the hilly and lowland sections of Tokyo in the 1960's. Skeletal maturity was
more advanced in early childhood, and the second metacarpal was longer and more
slender at almost ages, in children of the hilly section than of the lowland (Kimura
1999). Although there is no universal pattern, skeletal development is more retarded
in boys than in girls under conditions of environmental stress (Stinson 1985).
Recently, however, no socio-economic influences could be confirmed for growth of
children in European countries and the United States (Backstrom-Jarvien 1964,
Valaoras and Laros 1969, Goldstein 1971, Hamill et al. 1972, Jones et al. 1973,
Roche et al. 1975, 1978, Lindgren 1976, Otto 1976, Meredith 1979, Anzai et al.
1981, Laska-Mierzejewska et al. 1982, Bogin 1991, Guilfored et al. 1991,
Mackenbach 1991, Lindgren and Cernerud 1992). No relationships were also
revealed between skeletal maturity and socio-economic status for children in
Denmark (1965-66) and Austria (1974-75) (Wenzel and Melsen 1982, Wenzel et at.
286 CHAPTER 22
1984), except for those in developing and poor countries (Meredith 1984, Prado et
al. 1986, Preece 1989, Rosique and Rebato 1995).
Comparison of the data of the British by Poland (1898) and Stuart et al. (1962)
indicated a secular increase in the rates of skeletal maturation of approximately 0.22-
0.66 year/decade (Himes 1984). But no changes were found in skeletal maturity
corresponding to the secular trend in the United States (Roche and Davila 1976,
Roche et al. 1978). In Japan, it seems that the secular trend to toward greater stature
since 1900 arrived at a final stage approximately after 1970 in students and 1980 in
the general public, and that the secular trend proceeded rapidly in a population under
conditions of vigorous socio-economic changes (Kimura 1984). Few differences have
been already evident in the height and body weight of children between the hilly and
the lowland sections of Tokyo in 1995 (Kimura 1999). However, an acceleration of
the RUS maturation was observed in Japanese children between 1968-69 and 1986
(Ashizawa 1994).
4.3 Differences in populations

Differences among populations are certainly found in adult body size and shape, but
also in rate of growth and development in children.
Japanese children are somewhat delayed (Ashizawa 1970, Eto 1971, Kawashima
et al. 1972, Kimura 1972, 1976b, 1977, Kondo and Eto 1972: Takai and Akiyoshi
1983, Takai et al. 1984, Takai 1990, 1993) or comparable (Eto and Ashizawa 1992,
Ashizawa 1993, Ashizawa et al. 1996) in skeletal maturity during childhood with
the British standards. Then, they are markedly advanced during adolescence, and
arrive at full maturity a little earlier or at the same age with the standards. Chinese
children also show a similar pattern of skeletal maturation as the Japanese (Low et
al. 1964, 1976a, 1976b, Chang et al. 1967, Waldmann et al. 1977, Zhen and
Baolin 1986, Ye et al. 1992). The war orphan followed roughly the same pattern of
skeletal maturity to the control at the early ages, approximately until 10 years in the
boys and 11 years in the girls. Then, they became gradually delayed in maturity to
the control, especially in the boys. Such adolescent retardation was not appeared for
the children of mixed American and Japanese parentage, whether in the boys or in
the girls. Its certain reasons are unknown, but it may be due to their living
conditions, possibly emotional ones during preadolescence.
European children were delayed in skeletal maturity with the Greulich-Pyle
standards, in Zurich (Budlinger and Prader 1972), Copenhagen (Andersen 1968,
Mathiasen 1973) and Helsinki (Koski et al. 1961). A lag of skeletal maturity was
also found during adolescence in Danish (Andersen 1971, Helm 1979, Wenzel and
Melsen 1982), Dutch (Borms et al. 1973, Prahl-Andersen et al. 1979, Roede et al.
1976, Venrooij-Ysselmuiden and Ipenburg 1978), Austrian (Wenzel et al. 1984) and
Arabian children (Shakir and Zaini 1974), with the British standards. Whereas the
majority of recent studies has reported good agreement or somewhat advancement
with the British standards in Swedish (Karlberg et al. 1976, Taranger et al. 1987),
Finnish (Tiisala, et al. 1969, 1971), Belgian (Beunen et al. 1990) and Italian
children (Vignolo et al. 1992). This fact possibly involves a secular trend in skeletal
maturity (Wenzel et al. 1984).
East Africans were more and more delayed after 5 years of age and 1.5 to 2 years
behind by 10 years of age (Mackay 1952), with the Todd's standards (Todd 1937).
West African boys were delayed by an average 1.5 years after 9 years of age (Weiner
and Thambipillai 1952), and South Africans in Pretoria by an average 0.7 years in
boys and 1.5 years in girls in 7-12 years of age (Levine 1972), with the Greulich and
Pyle standards. According to a longitudinal study in West Africans (TWl), the girls
and boys both were retarded at 11 years of age but caught up during puberty. The
girls actually reached and the boys were only 0.5 years retarded the standards at 15
years of age (Michaut et al. 1972). African Coloured children showed skeletal
advancement in the early years of life over Europeans (Dommisse and Leipoldt
1936). Pretoria Black and Coloured children did not differ significantly from each
other (Levine 1972). The mean TW2 ages are always less in children ofKhoi Khoi
and Rehoboth Basters than the British standards by one or two years in both sexes
(Singer and Kimura 1981). Thus, in general, African children were advanced
skeletally in infancy (Hess and Weinstock 1925, Beresowski and Lundie 1952,
Falkner 1958), followed by deceleration and finally retardation in middle and late
childhood (Masse and Hunt 1963), compared with European children. Africans
apparently have a different maturational pattern from Europeans as far as timing and
duration of some stages is concerned (Wingerd et al. 1974). The rates of skeletal
maturation for African children in the United States (Roche et al. 1975) are more
rapid than in Africa (Levine 1972) and also in Jamaica (Marshall et al. 1970).
European descendants in Pretoria (Levine 1972) and Melbourne (Roche 1967)
were closely comparable with the Greulich-Pyle standards. The Euro-American
children indicated a trend to lesser retardation in skeletal maturity with increasing
age in Atlanta (Gray and Lamons 1959), Philadelphia (Johnston 1962) and Nebraska
(Fry 1966). The same trend was observed in the French Canadians (Baughan et al.
1979). The National Health Survey (NHS) in the United States in 1963-70 found
that the Euro-American boys and girls are retarded between 9 and 13 years of age and
that the definite tendency increases retardation (Roche et al. 1974, 1976). In
Philadelphia, however, the boys were first skeletally retarded, but soon more ahead
of the Greulich-Pyle standards to a peak advancement at 11 years of age. But the
girls did not move ahead in the standards until more later, beginning this spurt at
10 years, and levelling off at 14 years of age (Johnston 1963). The Euro-Americans
tended to be slightly advanced in skeletal maturity status with the British standards
(Malina 1970). The British descendants in Australia also accelerated at about 11
years in the boys and 9 years in the girls, after 12 years of age they decelerated in the
girls (Roche 1967), relative to the British standards.
First, Todd (1931) noted that the Afro-American girls are often in advance
skeletally to the Euro-American girls. A similar precocity in ossification of the
skeleton was demonstrated for the African and Afro-American new-borns and infants
compared with the Euro-Americans (Kelly and Reynolds 1947, Christie 1949,
Tompkins and Wiehl 1954). But no significant differences were found in skeletal
maturity between Afro- and Euro-American children 4.5 to 14.5 years of age in
Philadelphia (Platt 1956, Bass 1960). The TWI skeletal ages showed little
consistent difference between the Afro- and Euro-American boys, but the Afro-
American girls tended to be somewhat advanced over the Euro-American girls 9 to
12 years of age (Malina 1969, 1970). The NHS data (Roche et al. 1975, 1978)
indicated that the mean skeletal ages tend to be more advanced in the Afro-American
boys than in the Euro-American boys at most ages up to 10 years but not later. The
mean skeletal ages tend to be more advanced in the Afro-American girls than in the
288 CHAPTER 22
Euro-American girls at all ages except 8, 16 and 17 years. Low-income Afro-

Americans were advanced in timing of ossification centres over Euro-Americans of
somewhat higher incomes, suggesting that genetic factors also have an importance
on skeletal maturation (Gam et al. 1972: Gam and Clark 1975).
Thus, Asiatic, European and African populations apparently have a different
maturational pattern as far as timing and duration of stages is concerned. These
findings possibly involve more or less a secular trend and socio-economic influences
in skeletal maturity in every population. In comparison of skeletal maturation and
growth among populations, their differences should not be attributed to genetic
factors until extrinsic influences have been excluded as a cause for such differences.
However, some extrinsic factors may be surely unavoidable in these population
studies. In the present paper, the skeletal maturity has been examined on the
children of mixed American and Japanese parentage and the Japanese, living at
almost similar living conditions in the 1960' s, with the British standards.
4.4 Children of mixed racial origins
The Rehobth Basters (Khoi Khoi x European children) showed the intermediate
TW2 ages between the Kohi Khoi and British children at almost all ages (Singer
and Kimura 1981). According to the preliminary study by the Greulich-Pyle and
TWI methods, the children of mixed Euro-American and Japanese parentage showed
the intermediate maturity between the Japanese in Sapporo and British descendants
in Melbourne (Roche et al. 1971) after pre-adolescence (Kimura 1971, 1972). The
present study confirmed the followings.
1 The children of mixed American and Japanese parentage are followed roughly
a similar pattern of the skeletal maturity with the Japanese.
2 The children of mixed Afro-American and Japanese parentage are always more
advanced in skeletal maturity, followed by those of mixed Euro-American and
Japanese parentage, than the Japanese do.
3 Differences of skeletal maturity between the children of mixed American and
Japanese parentage and the Japanese could be especially due to those for the
RUS bones during preadolescence in both sexes.
4 The children of mixed Afro-American and Japanese parentage are advanced in
skeletal maturity as compared to those of mixed Euro-American and Japanese
parentage, especially for the carpal in the boys and each skeletal age in the
girls.
The Afro-American is advanced in skeletal maturity during adolescence as
compared to the Euro-American, especially in girls (Malina 1969, 1970, Roche et
al. 1975, 1978, Robson et al. 1975, Gam 1976). As the Afro-American is a
European-African admixture, the flow of genes into them has been primarily
undirected with respect to the sex chromosome (European male to African female),
and a greater percentage of the Y chromosome in the Afro-American gene pool are
from European sources (Malina 1970). In this series of mixed American and Japanese
parentage also, a greater percentage of the Y chromosome come from Afro- and Euro-
American males, that is, from European sources.
Examining ossification in children with Klinefelter's and Turner's syndromes,
Tanner et al. (1959) suggested that Y chromosome slows the rate of skeletal
maturation in males. A large majority of studies generally indicated higher
correlation for sister-sister pairs than for brother-brother or brother-sister pairs in

postnatal ossification. It seemed to be consists the hypothesis of partial X-linkage
mediated by the autosomal gene in skeletal maturation (Gam and Rohmann 1962b,
Gam et al. 1963, 1969). However, the fmdings, that relative, conception-corrected
variability is significantly larger in girls, was suggestive of a "dosage" effect and
inconsistent with the hypothesis of selective inactivation of either the parental or the
maternal X chromosome (Gam and McCreery 1970). In the study of twins, the
heritability index (H 2 , Wetherington and Rutenberg 1978) in the TWI ages was
also slightly higher in the girls (0.978) than in the boys (0.928) (Sklad 1973). It
was certainly higher in the girls (0.919) than in the boys (0.881) for the RUS age,
though in the boys (0.922) than in the girls (0.864) for the 20-bone age, at 12-15
years of age, just before full maturity (recalculated from Kimura 1981). According to
mixed longitudinal data of siblings, Xi et al. (1989) found that the correlation for
the brother-brother and sister-sister pairs was similar from about 1 thorough 12 years
after which the sister-sister values were markedly higher due to decreases in the
coefficients for the brother-brother pairs.
Adolescence appears to begin earlier and occupies a shorter period among East
Asiatics than in children of European ancestry, and this more rapid ossification,
combined with the fact that they are shorter at the initiation of the spurt, may result
in smaller adult height (Eveleth 1978b). This hypothesis was not supported from
the fact that no significant correlation was found between the adult height and the age
at the full RUS maturity (r = 0.02 for boys and 0.01 for girls) (Takai 1996). The
children of mixed Euro-American and Japanese parentage were as tall as the Japanese
children were at 6 years of age, and then, became gradually with age placed in an
intermediate position for height between their parental populations until 15 years of
age (Suda et al. 1965). However, both the children of mixed Afro- and Euro-
American and Japanese parentage followed roughly a similar pattern of growth of the
second metacarpal length to their paternal Americans, being the values usually
placed in almost the intermediate position between those of their parental
populations (Kimura 1996). For regression relationship of the bone length to
skeletal maturity, significant differences were found between the children of mixed
American-Japanese parentage and the Japanese, regardless sex, and between both
children of mixed Afro- and Euro-American and Japanese parentage in the girls. The
bone length increased more rapidly in the children of American and Japanese
parentage than the Japanese as it matures. The genetic control could become
dominant gradually for the bone length, as skeletal maturity progresses from the start
(Kimura 1996).
5. REFERENCES
Abbott, E.D., Townsen, R.O., French, RB., and Armann, C.F., 1950, Carpal and epiphyseal
development: Another index of nutritional status of rural school children. American Journal of
Disease of Children, 79, 69-81.
Acheson, RM., 1954, A method of assessing skeletal maturity from radiographs. A report from the
Oxford health survey. Journal of Anatomy, 88,498-508.
Acheson, RM., 1960, Effects of nutrition and disease on human growth. In: Tanner, 1.M.(Ed.) Human
Growth. Symposium of the Society for the Study of Human Biology. Vol. 3, pp. 73-92, Pergamon
Press, Oxford.
290 CHAPTER 22
Andersen, E., 1968, Skeletal maturation of Danish school children in relation to height, sexual
development, and social conditions. Acta Paediatrica Scandinavia, Supplement 185.
Andersen, E., 1971, Comparison of Tanner-Whitehouse and Greulich-Pyle methods in a large scale
Danish survey. American Journal of Physical Anthropology, 35, 373-376.
Anzai, I., Sakamoto, K., Togo, M., and Katsunuma, H., 1981, A comparative study of Body composition
of urban and rural Japanese boys 12 to 14 year old. Annals of Human Biology, 8, 109-117.
Ashizawa, K., 1970, Maturation osseuse des enfants japonais de 6 a 18 ans, estimee par la methode de
Tanner-Whitehouse. Bulletins et Memoires de la Societe d'Anthropologie de Paris, 6, 265-280.
Ashizawa, K., 1974, Les erreurs d'interpretation dans I'etude de I'estimation des stades de la
maturation osseuse par un observateur utilisant la methode de Tanner-Whitehouse. Bulletins et
Memoires de la Societe d'Anthropologie de Paris, 13,335-350.
Ashizawa, K., 1993, Skeletal and sexual maturation and growth in Tokyo girls: longitudinal
observations. Clinical Pediatric Endocrinology, 2 (Supplement 1), 5-8.
Ashizawa, K., 1994, Secular trends and regional variations of Japanese skeletal maturation.
Anthropological Science, 102, 193.
Ashizawa, K., Asami, T., Anzo, M., Matsuo, N., Matsuoka, H., Murata, M., Ohtsuki, F., Satoh, M.,
Tanaka, T., Tatara, H., and tsukagoshi, K., 1996, Standard RUS skeletal maturation of Tokyo
children. Annals of Human Biology, 23, 457-469.
Backstrom-Jarvien, L, 1964, Height and weight of Finnish children and young adults. Annals of
Paediartics,23 (Supplement), 1-116.
Bass, W.M., 1960, A comparative study of the sequence and time of appearance of maturity indicators
in the hand in white, Negro and Chinese children in the U.S. American Journal of Physical
Anthropology, n.s.18, 339-340 (abstract).
Baughan, B., Demirjian, A., and Levesque, G-Y., 1979, Skeletal maturity standards for French-
Canadian children of school-age with a discussion of the reliability and validity of such measures.
Beresowski, A., and Lundie, J.K., 1952, Sequence in the time of ossification of the carpal bones in 705
African children from birth to six years of age. South African Journal of Medical Science, 17, 25-
31.
Beunen, G., and Cameron, N., 1980, The reproducibility of TW2 skeletal age assessment by a self-
taught assessor. Annals of Human Biology, 7,155-162.
Beunen, G., Lefevre, J., Ostyn, M., Renson, R., Simons, J., and Van Gerven, D., 1990, Skeletal maturity
in Belgian youths assessed by the Tanner-Whitehouse method (TW2). Annals of Human Biology,
17,355-376.
Bogin, B, 1991, Measurement of growth variability and environment quality in Guatemalan children.
Bogin, B., Sullivan, T., Hauspie, R., and MacVean, R.B., 1989, Longitudinal growth in height, weight,
and bone age of Guatemalan Ladino and Indian schoolchildren. American Journal of Human
Biology, 1, 103-113.
Borms, J., Broekhoff, J., and Beunen, G., 1973, Vergelijkende studie tussen drie methoden voor het
bepalen van de skeletale leeftijd. Archives belges de Medecine sociale, hygiene, Medecine du
Travail et Medecine legale, 3, 343- 353. (cited from Wenzel et al., 1984).
Borns, P.F., and Moshang, T., 1977, Weekly radiology science update. Biometrica, 37, 1-5. Budlinger,
H., and Prader, A, 1972, Unpublished data from Zurich Longitudinal Growth Study, May 1972.
(cited from Eveleth and Tanner, 1976).
Buschke, F., 1935, The radiological examination of skeletons of triplets. Journal of Heredity, 26, 391-
410.
Chang, K.S.F., Chan, S.T., and Low, W.D., 1967, Skeletal maturation of Chinese preschool children.
Far East Medical Journal, 3, 289-293.
Christie, A., 1949, Prevalence and distribution of ossification centers in the newborn infant. American
Journal of Disease of Children, 77, 355-361.
Dommisse, F.H., and Leipoldt, C.L., 1936, The ossification of the carpal centres in colored children.
South African Medical Journal, 10,713-715.
Douglas, J.W.B., and Blomfield, J.M., 1958, Children Under Five. London, George Allen and Unwin.
Dreizen, S., Snodgrasse, R.M., Parker, a.s., Currie, c., and Spies, T.D., 1954, Maturation of bone
centers in hand and wrist of children with chronic nutritive failure. American Journal of Disease of
Children, 87,429-439.
Dreizen, S., Snodgrasse, R.M., Webb-Peploe, H., and Spies, T.D., 1957, The effect of prolonged
nutritive failure on epiphyseal fusion in the human hand skeleton. American Journal of
Roentgenolgy, Radium Therapy and Nuclear Medicine, 78,461-470.
Dreizen, S., Snodgrasse, R.M., Webb-Peploe, H., and Spies, T.D., 1958, The retarding effect of
protracted undernutrition on the appearance of the postnatal ossification centers in the hand and
wrist. Human Biology, 30, 253-264.
Dreizen, S., Spirakis, C.N., and Stone, R.E., 1964, Chronic undernutrition and postnatal ossification.
American Journal of Disease of Children, 108,44-52.
Dreizen, S., Spirakis, C.N., and Stone, R.E., 1967, A comparison of skeletal growth and maturation in
undernourished and well-nourished girls before and after menarche. Journal of Pediatrics, 70, 156-
263.
Eto, M., 1971, The skeletal development of the children in Tokyo -Comparative study Through two
kinds of the skeletal age assessed by Japanese and British Methods -. Journal of Anthropological
Society of Nippon, 79, 9-20.
Eto, M., and Ashizawa, K., 1992, Longitudinal Observations on Physical Growth and TW2 Skeletal
Maturation of Girls in Tokyo. Radiographic Atlas of Hand and Wrist. Tokyo, Terpeia.
Eveleth, P.B., 1978a, Population differences in growth: environmental and genetic factors. In: Falkner,
F. and Tanner, lM. (Eds.), Human Growth, Vol. 3 (Rev. Ed.), New York, Plenum Press, pp. 373-
394.
Eveleth, P.B., 1978b, Differences between populations in body shape of children and adolescents.
Eveleth, P.B., and Tanner, lM., 1976, Worldwide Variations in Human Growth. Cambridge, Cambridge
University Press, pp. 199-206.
Falkner, F., 1958, Skeletal maturation; an appraisal of concept and method. American Journal of
Physical Anthropology, 16,381-396.
Fry, E.I., 1966, Skeletal age by guess and by Tanner II. Large sample analysis. American Journal of
Physical Anthropology, 25, 202. (quoted from Roche et aI., 1974).
Garn, S.M., 1976, Metacarpal Lengths, Cortical Diameters and Areas from the 10-State Nutrition
Survey, Including Estimated Skeletal Weights, Height, and Stature For Whites, Blacks and
Mexican-Americans. Publication No. (HSM) 21-72-522. U.S. Department of Health, Education and
Welfare, Washington, D.C.
Garn, S.M., and Clark, D.C., 1975, Nutrition, growth, development, and maturation findings from the
ten-state nutrition survey of 1968-1970. Pediatrics, 56, 306- 319.
Garn, S.M., and McCreery, L.D., 1970, Variability of postnatal ossification timing and evidence for a
"dosage" effect. American Journal of Physical Anthropology, 32, 139-144.
Garn, S.M., and Rohmann, C.G., 1962a, Parent-child similarities in hand-wrist ossification. American
Journal of Disease of Children, 103,603-607.
Gam, S.M., and Rohmann, C.G., 1962b, X-linked inheritance of developmental timing in man. Nature,
196, 695-696.
Garn, S.M., Rohmann, C.G., and Apfelbaum, B., 1961a, Complete epiphyseal union of the hand.
American Journal of Physical Anthropology, 19,365-372.
Garn, S.M., Rohmann, C.G., and Davis, A., 1963, Genetics of hand-wrist ossification. American Journal
Garn, S.M., Rohmann, C.G., and Hertzog, K.P., 1969, Apparent influence of the X chromosome on
timing of73 ossification centers. American Journal of Physical Anthropology, 30, 123-128.
Gam, S.M., Rohmann, C.G., and Robinow, M., 1961b, Increments in hand-wrist ossification. American
Journal of Physical Anthropology, 19,45-53.
Goldstein, H., 1971, Factors influencing the height of seven-year-old children. Results from the
National Child Development Study. Human Biology, 43, 92-111.
Goto, A., 1961, Studies on the color tone and the histological structure of the skin of mixed blood.
Journal of Nippon Medical School, 28, 145-158. (in Japanese).
Gray, S., and Lamons, F., 1959, Skeletal development and tooth eruption in Atlanta children. American
Journal ofOrthod., 45, 272-277.
Greulich, W.W., and Pyle, S.L., 1950, Radiographic Atlas of Skeletal Development of the Hand and
Wrist. Stanford, Stanford University Press.
Greulich, W.W., and Pyle, S.L., 1959, Radiographic Atlas of Skeletal Development of the Hand and
Wrist. 2nd Ed. Stanford, Stanford University Press.
292 CHAPTER 22
Guilfored, M.C., Chinn, S., and Rona, R.I., 1991, Social environment and height: England and Scotland
1987 and 1988. Archives of Diseases in Childhood, 66, 235-240.
Habicht, J-P., Yarbrough, C., Martorell, R., Malina, R.M., and Klein, R.E., 1974, Height and weight
standards for preschool children. Lancet, I, 611-615.
Hamill, P.V.V., Johnston, F.E., and Lemeshow, S, 1972, Height and Weight of Children; Socioeconomic
Status. Department of Health, Education and Welfare, Pub. No. (HSM) 73-1601. Vital Health
Statitics Series 11, No. 11. Washington, D.C.
Hanihara, K., 1954-57, Studies on the deciduous dentition of the Japanese and the Japanese-American
hybrids. I. Journal of Anthropological Society of Nippon, 63, 165-185; 64, 63-82, 95-116; 65, 67-78,
151-164.
Hanihara, K., 1963, Crown characters of the deciduous dentition of the Japanese and the Japanese-
American hybrids. Dental Anthropology, pp. 105-124, Pergamon, Oxford.
Hanihara, K., 1965, Some crown characteristics of the deciduous incisors and canines in Japanese-
American hybrids. Journal of Anthropological Society of Nippon, 72, 135-145.
Hanihara, K., 1968, Morphological patterns of the deciduous dentition in the Japanese- American
hybrids. Journal of Anthropological Society of Nippon, 76, 114-121.
Helm, S., 1979, Skeletal maturity in Danish schoolchildren assessed by TW2 method. American Journal
Hertzog, K.P., Falkner, F., and Garn, S.M., 1969, The genetic determination of Ossification sequence
polymorphism. American Journal of Physical Anthropology, 30, 141-144.
Hess, A.F., and Abramson, H, 1933, Familial retardation in ossification of the carpal centers. Journal of
Hess, A.F., and Weinstock, M., 1925, A comparison of the evolution of carpal centres in white and
Negro newborn infants. American Journal of Disease of Children, 29, 347-354.
Himes, J.H., 1984, An early hand-wrist atlas and its implications for secular change in bone age. Annals
of Human Biology, 11,71-75.
Hojo, T., 1964, Untersuchungen tiber die Mundzone bei den Japaner-Americaner- Mischlingen.
Journal of Anthropological Society of Nippon, 71, 143-152. (in Japanese with German summary).
Honda, A., 1958, On the iris and head-hair of the half-breed children. Journal of Nippon Medical
School, 25, 297-322.
Hoshi, H., 1959, Brief note on some cephalometrical results from the longitudinal study of the Japanese-
American hybrids. Journal of Anthropological Society of Nippon, 67, 73-81.
Hoshi, H., 1970, Physical growth of Japanese-American hybrids from 6 to 15 years of age with special
reference to genetic-environmental relationships. Monograph of Society of Research of Child
Development, 35, 49-54.
lshimoto, G., 1968, Urinary excretion of Ei,-aminoisobutyric acid in Japanese- American hybrids.
Journal of Anthropological Society of Nippon, 68, 141-144.
Johnston, F.E., 1962, Longitudinal Study of Skeletal Maturation and Its Relationship to Growth in
Philadelphia White Children; a Dissertation in Anthropology, Unpublished doctoral dissertation,
University of Pennsylvania. (quoted from Roche et at., 1974).
Johnston, F.E., 1963, Skeletal age and its prediction in Philadelphia children. Human Biology, 35, 192-
202.
Johnston, F.E., and Jahina, S.B., 1965, The contribution of the carpal bones to the assessment of skeletal
age. American Journal of Physical Anthropology, 23, 349-354.
Jones, D.L., Hempill, W., and Meyer, E.SA, 1973, Height, Weight and Other Physical Characteristics
of New South Wales Children. Part I. Children Aged Five Years and Over. Sydney, N.S.W.
Department of Health. (quoted from Eveleth, 1978).
Jones, P.R.M., and Dean, R.F.A., 1956, The effects of kwashiorkor on the development of the bones of
the hand. Journal of Tropical Medicine, 2, 51-68.
Kariberg, P., Taranger, J., EngstrOm, I., Lichtenstein, H., and Svenberg-Redegren, I., 1976, The
somatic development of children in a Swedish urban community. A prospective longitudinal study.
Acta Paediatrica Scandinavica (Supplement 258).
Kawashima, M., Nanjo, F., Sakura, H., and Ohta, S., 1972, Skeletal age estimated by the Tanner-
Whitehouse-Heary method. Comparison with the Sugiura and Nakazawa method. Clinical
Orthopedic Surgery, 7, 350-358. (in Japanese).
Kelly, H.I. and Reynolds, L., 1947, Appearance and growth of ossification centers and increases in the
body dimensions of white and Negro infants. American Journal of Roentgenology and Radium
Therapy, 57,477-516.
Key, W.E., 1936, Radiological studies of the skeleton of twins, triplets and quadruplets. Journal of
Heredity, 27,86-88.
Kimura, K., 1962, Radiological studies of the hand and wrist bones in twins. In: Fujita, T. (Ed.), The
Study of Twins-III. Tokyo, Maruzen, pp. 44-56. (in Japanese).
Kimura, K., 1971, On the skeletal maturation of the Japanese-American hybrids- A preliminary report.
Journal of Anthropological Society of Nippon, 79, 21-29.
Kimura, K., 1972, Skeletal maturation in Japanese according to the Oxford and Tanner- Whitehouse
methods. Acta Anatomica Nipponica, 47, 358-372.
Kimura, K., 1974a, An application of the method of biological distance to the dermatoglyphic data of
Ainu-Japanese and Japanese American white hybrids. Journal of Anthropological Society of
Nippon, 13,37-50.
Kimura, K., 1974b, Dermatoglyphics of the Japanese-American hybrids. Bulletin of Faculty of Physical
Education, Tokyo University of Education, 13,37-50.
Kimura, K., 1976a, Skeletal maturation of Japanese-American white hybrids. American Journal of
Physical Anthropology, 44, 353-356.
Kimura, K., 1976b, Skeletal maturation of children in Okinawa. Annals of Human Biology, 3, 149-155.
Kimura, K., 1977, Skeletal maturity of the hand and wrist in Japanese children in Sapporo by the TW2
method. Annals of Human Biology, 4, 353-356.
Kimura, K., 1981, Skeletal maturity in twins. Journal of Anthropological Society of Nippon, 89, 457-
478.
Kimura, K., 1983, Skeletal maturity and bone growth in twins. American Journal of Physical
Kimura, K., 1984, Has the secular trend stopped in Japanese? Japanese Journal of Human Posture, 4,
17-27. (in Japanese with English summary).
Kimura, K., 1994, Skeletal maturation of American-Japanese hybrids by TW2 method. Anthropological
Science, 102, 194.
Kimura, K., 1996, Growth of second metacarpal of American-Japanese hybrids. Acta Anatomica
Nipponica, 71, 39-46.
Kimura, K., 1999, Skeletal maturity and growth of children in two sections of Tokyo in the 1960s,
Perspective in Human Biology, 4,181-187.
Kimura, K., and Takeuchi, K., 1982, Notches at the n for the carpal age on-epiphyseal end of the
second metacarpal in Japanese-American hybrids. Zeitschrift fUr Morphologie und Anthropoiogie,
73, 1-14.
Kitamura, Sh., and Suzuki, K., 1965, Blood groups of the hybrid children. III. subgroups of A and E-e
blood factor. Tokyo Journal of Medical Science, 64, 171-178.
Kohara, Y., 1968, Subcutaneous fat distribution in Japanese-American hybrid boys. Journal of
Anthropological Society of Nippon, 76, 105-113.
Kondo, S, and Eto, M., 1972, Physical growth studies on Japanese-American hybrids in comparison with
native Japanese. In: Proceedings of Meeting for Review and Seminar of the US-Japan Cooperative
Research on Human Adaptability, Kyoto, 1972, Japan Society for the Promotion of Science and
National Sciences Foundation, Kyoto.
Kondo, S., Sato, M., Kikuchi, Y., Tomita, M., Okada, M., and Hayami, A., 1968, Electoromyographic
studies on facial muscles of Japanese-American hybrids. Journal of Anthropological Society of
Nippon, 76, 135-140.
Koski, K., Haataja, 1., and Lappalanien, M., 1961, Skeletal development of hand and wrist in Finnish
children. American Journal of Physical Anthropology, 19,379- 382.
Kristmundsdottir, F., Burwell, R.G., Marshall, W.A., and Small, P., 1984, Cross- sectional study of
skeletal maturation in normal children from Nottingham and London. Annals of Human Biology, II,
133-139.
Laska-Mierzejewska, T., Milicer, H., and Piechaczek, H, 1982, Age at menarche and its secular trend
in urban and rural girls in Poland. Annals of Human Biology, 9, 284-292.
Levine, E., 1972, The skeletal development of children of four South African populations. Human
Biology, 44, 399-412.
Lindgren, G., 1976, Height, weight and menarche in Swedish urban school children in relation of
socioeconomic and regional factors. Annals of Human Biology, 3,501-528.
Lindgren, G., and Cernerud, L., 1992, Physical growth and socioeconomic background of Stockholm
schoolchildren born in 1933-63. Annals of Human Biology, 19, 1-16.
294 CHAPTER 22
Low, W.O., 1972, Relation between socio-economic status and skeletal maturation of Chinese Children.
Zeitschrift fur Morphologie und Anthropologie, 64, 1-11.
Low, W.O., Chan, S.T., Chang, K.S.F., and Lee, M.M.C., 1964, Skeletal maturation of Southern
Chinese. Child Development, 35, 1313-1336.
Low, W.O., Fung, S.H., and Cerny, E.B., 1976a, Biological age of Southern Chinese boys. Zeitschrift
fur Morphologie und Anthropologie, 67, 136-144.
Low, W.O., Fung, S.H., and Cerny, E.E., 1976b, Biological age of Southern Chinese girls. Zeitschriffilr
Morphologie und Anthropologie, 67, 264-274.
Mackay, D.H., 1952, Skeletal maturation in the hand: A study of development in East African children.
Trans. Royal Society of Tropical Medicine and Hygiene, 46,135-150.
Mackenbach, J.P., 1991, Narrowing inequalities in children's height, Lancet, 338, 764.
Malina, R.M., 1969, Skeletal maturation rate in North American Negro and White children. Nature,
223,1075.
Malina, R.M., 1970, Skeletal maturation studied longitudinally over one year in American whites and
Negroes six through thirteen years of age. Human Biology, 42, 377-390.
Maniar, B.M., Seevai, M.H., and Kapur, P.L., 1974, A study of ossification centers in the hand and wrist
ofIndian children. Indian Pediatrics, 11, 203-211.
Marshall, W.A., Ashcroft, M.T., and Bryan, G., 1970, Skeletal maturation of the hand and wrist of
Jamaican children. Human Biology, 42, 419-435.
Masse, G., and Hunt, E.B. Jr., 1963, Skeletal maturation of the hand and wrist in West African children.
Mathiasen, M.S., 1973, Determination of bone age and recording of minor skeletal hand anomalies in
normal children. Danish Medical Bulletin, 20, 80-85. (quoted from Eveleth and Tanner, 1976).
Matsuzawa, Sh., and Kitamura, Sh., 1956, Blood groups of the hybrid children. I. ABO Groups, MN
types, Rh types, secretors and nonsecretors. Tokyo Journal of Medical Science, 64, 83-88.
Medicus, H., Gron, A., and Moorrees, C.F.A., 1971, Reproducibility of rating stages of osseous
development (Tanner-Whitehouse system). American Journal of Physical Anthropology, 35, 359-
371.
Meredith, H.V., 1979, Comparative findings on body-size of children and youth living at urban centers
and rural areas. Growth, 43, 95-104.
Meredith, H.V., 1984, Body size of infants and children around the world in relation to socioeconomic
status. Archives of Child Development and Behavior, 18,81- 145.
Michaut, E., Niang, J., and Dan, V., 1972, La maturation osseous pendant la peri ode pubertaire.
Annales de Radiologie, 15,767-779. (quoted from Eveleth and Tanner, 1976).
Michibe, T., 1961, On the skin color of the half-breed children. Journal of Nippon Medical School, 28,
1196-1223. (in Japanese).
Omoto, K., 1968, Studies of skin color among hybrids subjects in Japan with special reference to
measurements of Sun-tanning. Journal of Anthropological Society of Nippon, 76,122-134.
Otto, U, 1976, Male youth. A socio-psychiatric study of a total annual population of Swedish adolescent
boys. Acta Psyciatrica Scandinavia, 264 (Supplement), 1- 312.
Platt, R.A., 1956, The Skeletal Maturation of Negro School Children. Unpublished master's thesis.
University of Pennsylvania, Philadelphia. (quoted from Malina, 1970).
Poland, J., 1898, Skiagraphic Atlas Showing the Development of the Bones of the Wrist and Hand.
London, Smith, Elder & Co., pp. 41. (quoted from Himes, 1984).
Prader, A., Tanner, lM., and von Harnack, G.A., 1963, Catch-up growth following illness or starvation.
An example of developmental canalization in man. Journal of Pediatrics. 62, 646-659.
Prado, C., Andersen, A.H., and Martinez, R., 1986, The relationship obesity-blood pressure: Attending
to ontogenesis and socio-environmental variation. International Journal of Anthropology (Spain), 1,
153-164.
Prahl-Andersen, B., Kowalski C.J., and Heydendael, P.H.lM., 1979, A Mixed- Longitudinal
Interdisciplinary Study of Growth and Development. London, Academic Press. (quoted from
Wenzel et al., 1984).
Preece, M.A., 1989, Standardization of growth. Acta Paediatrica Scandinavica, 349, 57- 64.
Pryor, J.W., 1907, The hereditary nature of variation in the ossification of bones. Anatomical Record, 1,
84-88.
Pryor, J.W., 1936, Ossification as additional evidence in differentiating identicals and fraternals in
mUltiple births. American Journal of Anatomy, 59, 403-409.
Pryor, lW., 1939, Normal variations of in the ossification of bones. Journal of Heredity, 30, 249-255.
Reynolds, E.L., 1943, Degree of kinship and pattern of ossification. American Journal of Physical
Anthropology, 1,405-416.
Rigler, L.G., 1938, Roentgen studies of twins and triplets. Radiology, 30, 461-470.
Robson, J.R.K., Larkin, F.A., Bursick, J.H., and Perri, K.P., 1975, Growth standards for Infants and
children: a cross-sectional study. Pediatrics, 56, 1014-1020.
Roche, A.F., 1967, A study of skeletal maturation in a group of Melbourne children. Australian
Paediatric Journal, 3,123-127.
Roche, A.F., 1989, Relative utility of carpal skeletal age. American Journal of Human Biology, 1,479-
482.
Roche, A.F., and Davila, G.H., 1976, The reliability of assessments of the maturity of individual hand-
wrist bones. Human Biology, 48, 585-597.
Roche, A.F., Davila, G.H., and Eyman, S.L., 1971, A comparison between Greulich- Pyle and Tanner-
Whitehouse assessments of skeletal maturity. Radiology, 98, 273-280.
Roche, A.F., Roberts, J., and Hamill, P.V.V., 1974, Skeletal Maturity of Children 6-11 Years. Vital
Health Statistics-Series II-No. 140, DHEW Publication No. (HRA) 75-1622. U.S. Department of
Health, Education and Welfare, Washington, D.C., pp. 62.
Roche, A.F., Roberts, J., and Hamill, P.V.V., 1976, Skeletal Maturity of Youths 12-17 Years. Vital
Health Statistics-Series l1-No. 160, DHEW Publication No. (HRA) 77-1642. U.S. Department of
Health, Education and Welfare, Washington, D.C., pp. 90.
Roche, A.F., Roberts, J., and Hamill, P.V.V., 1975, Skeletal maturity of youths 6-11 years: racial,
geographic area and socioeconomic differentials. Vital Health Statistics 11 - No. 149, DHEW
Publication No. (PHS) 76-1631. U.S. Department of Health Education and Welfare, Washington,
D.C., pp. 81.
Roche, A.F., Roberts, J., and Hamill, P.V.V., 1978, Skeletal Maturity of Youths 12-17 Years: Racial,
Geographic Area and Socioeconomic Differentials. Vital Health Statistics-Series II-No. 167,
DHEW Publication No. (PHS) 79-1654. U.S. Department of Health, Education and Welfare,
Washington, D.C., pp. 98.
Roede, M., Stoelinga, G., De Jong-van der Kar, M., and van 't Hof, M., 1976, Het Nijmeegse
Groeionderzoek. Antropometrische gegevens betreffende gezonde Nederlandse kinderen van 5
112, 8 1/2 en 10 1/2 jaar. Nederlands Tijdschrift voor Geneeskunde, 120, 133-140. (quoted from
Wenzel et al., 1984).
Rosique, J., and Rebato, E., 1995, Comparative study of statural growth in Spanish population. American
Journal of Human Biology, 7, 553-564.
Sahkir, A., and Zini, S., 1974, Skeletal maturation of the hand and wrist of young children in Baghdad.
SAS Institute Incorporation, 1958, SAS User's Guide: Statistics. 5th Ed. Cary, North Carolina, SAS
Institute Incorporation.
Satoh, M., Tanaka, 1., Ashizawa, K., Ohtuski, F., Anzo, M., Matsuo, N., Tatara, H., and Murata, M.,
1993, Inter- and intra-observer errors on estimation of the bone age by Tanner-Whitehouse II
method. Clinical Pediatric Endocrinology, 2,1055-1059.
Singer, R., and Kimura, K., 1981, Body height, weigh, and skeletal maturation in Hottentot (Khoikhoi)
children. American Journal of Physical Anthropology, 54, 401-413.
Sklad, V.M., 1973, Genetische Grundlagen einiger mit dem Heranreifen verbundener Erscheinungen
(nach Awillingsuntersuchungen). Zeitschrift fOr Morphologie und Anthropolgie, 65, 192-211.
Snedecor, G.W., and Cochran, W.G., 1980, Statistical Methods (9th Ed.). Ames, The Iowa State
University Press, pp. 243-248. (quoted from Roche and Guo, 1989).
Stinson, S., 1985, Sex differences in environmental sensitivity during growth and development.
Yearbook of Physical Anthropology, 28, 123-147.
Stuart, H.C., Pyle, S.l., Cornoni, I., and Reed, R.B., 1962, Onsets completions and spans of ossification in
the 29 bone-growth centers of the hand and wrist. Pediatrics, 29, 237-249.
Suda, A., 1968, Introduction to our anthropological study on the Japanese-American hybrids. Journal of
Suda, A., Hoshi, H., Eto, M., Ashizawa, K., 1975, Longitudinal observation on the pectoral growth of
Japanese-American hybrids with special reference to the difference in growth pattern between
thoracal and extrathoracal parts. Journal of Anthropological Society of Nippon, 83, 95-104.
Suda, A., Hoshi, H., Eto, M., Ashizawa, K., 1976a, Longitudinal studies on the Japanese-American
hybrids. Anthropological and Genetical Studies On the Japanese. JIBP Synthesis-Human
Adaptability, Vol. 2. Tokyo, University of Tokyo Press, pp. 33-72.
296 CHAPTER 22
Suda, A, Hoshi, H., Eto, M., Ashizawa, K., 1976b, Longitudinal observation on the limb growth of
Japanese-American hybrids with special regard to seasonal variation. Journal of Anthropological
Society of Nippon, 84, 15-30.
Suda, A., Hoshi, H., Eto, M., Ashizawa, K., and Hojo, Teruyuki, 1973, Longitudinal observation on
anterior trunk height, iliospinal height, biacromial breadth and bicristal breadth of Japanese-
American hybrids from 6 to 15 years of age. Journal of Anthropological Society of Nippon, 81,
185-194.
Suda, A, Hoshi, H., Sato, M., Eto, M., and Ashizawa, K., 1968, Longitudinal observation on the chest
circumference and sitting height of Japanese-American hybrids from 6 to 15 years of age. Journal
of Anthropological Society of Nippon, 76, 95-104.
Suda, A, Yamaguchi, 8., Hoshi, H., Endo, B., and Eto, M., 1965, Longitudinal observation on the
stature and body weight of Japanese-American hybrids from 6 to 15 years of age. Journal of
Suzuki, K., and Matsuzawa, Sh., 1956, Blood groups of the hybrid children. II. P (Q) groups. Tokyo
Journal of Medical Science, 64, 105-108.
Takai, S., 1990, Smoothed skeletal maturity curve of Japanese children by Tanner- Whitehouse (TW2)
method and its application. Acta Anatomica Nipponica, 65, 436-447.(in Japanese with English
summary).
Takai, S., 1993, Velocities for the Tanner-Whitehouse 2 skeletal maturity in northwestern Japanese
children. Okajimas Folia Anatomica Japonica, 70, 119- 126.
Takai, S., 1996, Does early skeletal maturer tend to be a smaller adult? Bulletin of Institute of Health
and Sport Sciences, University of Tsukuba, 19,91-97.
Takai, S., Akiyoshi, T., and Fuchigami, A, 1984, Skeletal maturity of Japanese children in Amami-
Oshima Island. Annals of Human Biology, 11,571-575.
Takai, S., and Akiyoshi, T., 1983, Skeletal maturity of Japanese children in western Kyushu. American
Tanner, J.M., and Gibson, R.D., 1994, A computerized image analysis system for estimating Tanner-
Whitehouse 2 bone age. Hormone Research, 42, 282-287.
Tanner, J.M., Oshman, D., Lindgren, G., Grunbaum, J.A., Elsouki, R., and Labarthe, D., 1994,
Reliability and validity of computer-assessed estimates of Tanner- Whitehouse skeletal maturity
(CASAS): comparison with manual method. Hormone Research, 42, 288-294.
Tanner, J.M., Prader, A., Habicht, H., and Ferguson-Smith, M.A, 1959, Genes on the Y chromosome
influencing rate of maturation in man. Lancet, 2, 141-144.
Tanner, J.M., Whitehouse, R.H., and Healy, M.J.R., 1962, A New System for Estimating Skeletal
Maturity from the Hand and Wrist, with Standards derived from a Study of 2,600 Healthy British
Children. Paris, International Children's Center.
Tanner, J.M., Whitehouse, R.H., Cameron, N., Marshall, W.A, Healy, MJ.R., and Goldstein, H., 1983,
Assessment of Skeletal Maturity and Prediction of Adult Height (TW2 method). 2nd Ed. London,
Academic Press.
Taranger, J., Karlberg, J., Bruning, B., and EngstrOm, I., 1987, Standard deviation score chart of
skeletal maturity and its velocity in Swedish children assessed by the Tanner-Whitehouse method
(TW2). Annals of Human Biology, 14,357-365.
Tatara, H., Murata, M., Takaishi, M., and Funakawa, H., 1985, Estimation of bone age of Japanese
children, especially based on longitudinal data. Pediatrics, 26, 455- 459. (in Japanese).
Tiisala, R., Kantero, R., and Tamminen, T., 1971, A mixed longitudinal study of skeletal maturation in
healthy Finnish children aged 5 to 10 years. Human Biology, 43, 224-236.
Tiisala, R., Kantero, R., BackstrOm and Hallman, K., and Hallman, N., 1969, A mixed longitudinal study
on skeletal maturation in healthy Finnish children aged 5 to 10 years. Human Biology, 41, 560-579.
Todd, T.W., 1931, Differential skeletal maturation in relation to sex, race, variability and disease. Child
Development, 2, 49-65.
Todd, T.W., 1937, Atlas of Skeletal Maturation (Hand). St. Louis, C.V. Mosby Co. Tompkins, W.T., and
Wiehl, D.G., 1954, Epiphyseal maturation in the newborn as related to maternal nutritional status.
American Journal of Obstetrics and Gynecology, 68, 1366-1377.
Valaoras, V., and Laros, K., 1969, Biometric characteristics of Greek pupils in elementary Schools.
IATPKH, 15,266-276. (quoted from Eveleth, 1979).
Van Loon, H., Saverys, V., Vuylsteke, J.P., Vlietinck, R.F., and Eeckels, R., 1986, Local versus
universal growth standards: The effect of using NCHS as a universal reference. Annals of Human
Biology, 13,347-357.
Venrooij-Ysselmuiden, M.E., van, and Ipenburg, A.V., 1978, Mixed longitudinal data on skeletal age
from a group of Dutch children living in Utrecht and surroundings. Annals of Human Biology, 5,
359-380.
Vignolo, M., Milani, S., Cerbello, G., Coroli, P., Di Batista, E., and Aicardi, G., 1992, Fels, Greulich-
Pyle, and Tanner-Whitehouse bone age assessments in a group of Italian children and adolescents.
American Journal of Human Biology, 4, 493-500.
Waldmann, E., Baber, F.M., Field, C.E., Billewicz, W.l., and Thomson, A.M., 1977, Skeletal maturation
of Hong Kong Chinese children in the first five years oflife. Annals of Human Biology, 4, 343-352.
Weiner, J.S., and Thambipillai, V., 1952, Skeletal maturation of West African Negroes. American
Journal of Physical Anthropology, 10,407-418.
Wenzel, A., and Melsen, B., 1982, Skeletal maturity in 6-16-year-old Danish children assessed by the
Tanner-Whitehouse method. Annals of Human Biology, 9, 277- 281.
Wenzel, A., Dorschl, H., and Melsen, B., 1984, Skeletal maturity in Austrian children assessed by the
GP and TW-2 methods. Annals of Human Biology, II, 173- 177.
Wetherington, R., and Rutenber, G.W., 1978, Contribution to variance in phalangeal growth: estimates
from twins. American Journal of Physical Anthropology, 48,83-88.
Wingerd, J., Peritz, E., and Sproul, A., 1974, Race and stature differences in the skeletal maturation of
the hand and wrist. Annals of Human Biology, 1: 201-209.
Xi, H., Roche, A.F., and Guo, S., 1989, Sibling correlations for skeletal age assessments by the Fels
method. American Journal of Human Biology, 1,613-619.
Ye, Y., Wang, C., and Cao, L., 1992, Skeletal maturity of the hand and wrist in Chinese children in
Changsha assessed by TW2 method. Annals of Human Biology, 19,427-430.
Yoshida, N., 1997, Bone age assessed by computer-assessed skeletal age (CASA) system: comparison
with the manual method. Journal of Osaka Medical College, 55, 43-54. (in Japanese with English
summary).
Zhen, 0., and Baolin, L., 1986, Skeletal maturity of the hand and wrist in Chinese school children in
Harbin assessed by the TW2 method. Annals of Human Biology, 13, 183-187.
BIOLOGICAL ASPECTS OF GROWTH
CHAPTER 23
SOMATOTYPES OF BUDAPEST CHILDREN
O. G. EIBEN, A. NEMETH
Department of Biological Anthropology, E6tv6s Lorimd University, Budapest,
Hungary
1. INTRODUCTION
Growth studies in Hungary have been carried out since the 1870s (Eiben 1988a).
The authors of that time were usually content with measuring stature. Systematic
growth studies with remarkable anthropometric programmes started after World War
II, and included amongst others, in the 1980s the Hungarian National Growth Study
(thereafter HNGS, Eiben et al. 1991). Before and after this, during the 20th Century,
there were many regional studies, too.
Growth of children and youth has been recommended as one of the best indices of
health and nutritional status of a community (WHO 1976). Starting out from the
scientifically established fact that growth and development of children and youth
monitors the biological status of the whole population (Tanner 1978, Bielicki 1986,
Eiben 1988b), and having a sense of responsibility for Hungarian children's welfare,
Eiben, with his colleagues A. Barabas and E. Pant6, organised and carried out the
above mentioned study.
The HNGS was a nation-wide, representative cross-sectional growth and physical
fitness study. The aim of it was to analyse growth and maturation as well as
physical fitness of Hungarian children and youth, taking into consideration all the
ecological factors existing in Hungary in the 1980s, especially regrouping of the
population, urbanisation, the urban and rural mode of life, etc. In the last third of our
20th Century, namely, these social factors were the most important environmental
ones influencing growth and maturation process.
The first HNGS was planned in 1981 and was designed to meet two important
needs:
• To provide normative standards to assess and monitor individual growth and
physical fitness, and
• to establish a baseline for successive sampling studies to analyse changes in the
Hungarian population.
Field studies commenced in January 1982 and were completed by March 1985 (see
Eiben et al. 1991).
The investigators of the HNGS visited 113 settlements in Hungary, more than
350 nurseries and schools. Their sample contained 39,035 healthy boys and girls
301
302 CHAPTER 23
between 3 and 18 years of age. This sample represents 1.5% of the same aged
Hungarian youth.
The anthropometric programme contained 18 body measurements. Seven
motoric tests were taken and socio-demographic data of the family background were
collected. Tools and techniques of the investigations were according to the
internationally standards (Martin and Saller 1957, Tanner et al. 1969, AAHPER
1965, Matthews 1973, Haag and Dassel 1975, Simon and Renson 1982)
The fIrst Hungarian national growth standards for 18 body measurements and 7
motoric tests were published in the middle 1980s (Eiben and Pant6 1986, 1987/88,
Eiben et al. 1991). This reference data serve as an 'etalon' in Hungary for growth
and physical fItness studies. Ten years after initial the HNGS, we were able to repeat
our study in Budapest, in the capital of Hungary.

The sample was based on the educational institutions in Budapest which were
selected for the HNGS. Some changes happened in the meantime: certain nurseries
and schools were wound up or closed down or altered. This was important,
however, for representativity: new institutions were involved which did not exist ten
years before (e.g. certain specialised schools). Thus 15 nurseries, 15 primary
schools, 10 secondary schools and 12 vocational training schools were involved in
our recent investigation. One group or class per cohorts was selected randomly in
each educational institution. All voluntary children were measured in each group.
The fInal sample contains 5076 healthy 3-18 year-old Budapest children. This
sample represents 1.5% of the Budapest youth of these age groups. Measurements
were carried out in the school-year 1994/95.
The anthropometric programme corresponded to that of the HNGS (18 body
measurements, see Eiben and Pant6 1986, 1987/88, Eiben et al. 1991), for
comparability. In this study, only those measurements were used from which the
anthropometric somatotype components were calculated: height, body mass, upper
arm and calf girths, bicondylar width of the humerus and femur, skinfold thickness
on the triceps, subscapular, suprailiac and medial calf (Heath 1977, Carter and Heath
1990). Measurements were taken according to the IBP recommendations (Tanner et
al. 1969), with children wearing only minimal closing. Basic statistical values were
calculated for all measurements for both sexes and for all age groups. Age was
calculated in decimal system for completed year plus/minus 6 months (i.e. 10 year-
old are children between 9.51 and 10.50 year). In this study these values (number of
cases: N, mean, M, standard deviation: SD, standard error SE, and the range: Ware
given only for height and body mass as informative data about the Budapest
children. Somatotypes were estimated using the Heath and Carter anthropometric
somatotyping method modifIed for children (Carter and Heath 1990).
3. RESULTS
Tables are shown in the Appendix. Budapest children in the middle of the 1990s
show a slight positive secular trend compared to the HNGS (Nemeth and Eiben
1997). Mean height and body mass of boys' and girls' age groups investigated are
SOMATOTYPES OF BUDAPEST CHILDREN 303
given in Table 1, 2, 3, and 4 to see what absolute measurements are connected to

somatotypes.
The means of the first somatotype component, endomorphy, are higher in girls
than in boys in all age groups (Table 5 and 6). The degree of difference is not
constant, it increases with the onset of the girls' puberty. It is less than 1 unit in age
groups below 13 years and it is between 1.5 and 2.5 in age groups of 13 year and
above. Endomorphy decreases between ages 3 and 7 years in both sexes, then
increases up to the age of 12 years in boys and 15 years in girls. There is a
remarkable decrease between 12 and 13 years of age and a slight decrease between the
ages of 13 and 15 in boys. There are practically no changes in the age groups older
than 15 years in both sexes.
The second component, mesomorphy, is higher in boys in all age groups than in
girls (Table 5 and 6). This difference is not greater than half a unit until the age of
11 years. In the older age groups, differences are between 0.5 and 1 unit, except the
14-15 year-old groups. Mean mesomorphy decreases until the age of lOin boys and
until the age of 14 in girls. Then, in the older age groups, it fluctuates around 4 in
boys and around 3.5 in girls.
The third component, ectomorphy, is very similar in the two genders until the
age of 10 (Table 5 and 6). Means are higher in the 11 and 12 year-old age groups in
girls and in the older age groups in boys. This component increases until the age of
10 in boys and until the age of 12 in girls, then decreases a bit until the age of 12 in
boys and until the age of 15 in girls. There is a great increase between 12 and 15
and slight decrease in the older age groups in boys. In girls, the mean is slightly
higher in the 16 year old group than in the 15 year-old one, and it does not change
in the older age groups.
Somatotypes are graphically presented in somatocharts (Figures 1 and 2).
Comparing boys' and girls' somatocharts, one can see that somatotypes are in the
endo-mesomorphic field of the chart (where mesomorphy dominates and
endomorphy is higher than ectomorphy) in the early childhood and move to the
central field with increasing age in both sexes. Boys are, however, more
mesomorphic than girls. According to age groups, the boys are in the en do-
mesomorphic field between ages of 3 and 6. Mean somatotypes of the 7-8 year-old
ones are in the axis of balanced mesomorphy (where mesomorphy dominates and
endomorphy and ectomorphy are about equal). Somatotypes of older age groups are
more or less in the central field (where the three components move around 3 and 4),
but they are in all cases in the field of somatochart where mesomorphy dominates.
Within this, 12 year-old boys are on the mesomorph-endomorph axis, while 15
year-old ones are on the mesomorph-ectomorph axis. Values of the oldest age groups
return through the central field to the axis of balanced mesomorphy.
In girls, only the means of 7-11 year-old groups are in the central field. The girls
between ages 3 and 4 years are in the en do-mesomorphic field. The 5-6 year-old
ones are rather on the mesomorph-endomorph axis (where mesomorphy and
endomorphy do not differ more than half a unit and ectomorphy is lower than the
two former types). The mean somatotype of 12 year-old girls is on the balanced
endomorphy axis (where endomorphy dominates and mesomorphy and ectomorphy
are equal). The means of older age groups move first along this axis to the apex of
endomorphy then to the meso-endomorphic field (where endomorphy dominates and
mesomorphy is higher than ectomorphy).
304 CHAPTER 23
~ud.pest boys
I
MESOVORPHY
~
Z
D
o
~
cr:
3: o
o::0 :::<:
o
/~ t> .........
UJ
Figure 1. Changes in somatotype of Budapest boys between the age of 3 and 18 years.
Budapest 01 rlS MESO'.1OAPHY
~cr:
Z
o
g,
o
3:
o o
:;0 :::<:
/~ vL.LI
Figure 2. Changes in somatotype of Budapest girls between the age of 3 and 18 years.
4. DISCUSSION
The results of anthropometric somatotyping in the case of children must be handled

cautiously and those using this method consider it reliable above the age of 10
(Carter and Heath 1990).
Analysing somatotype components one by one, we can conclude that differences
in the first component, endomorphy, between boys and girls show that sexual
dimorphism in the relative fat content of the body already exists in the early
childhood. The skinfold thickness data corroborate this finding. Another difference
between genders is a decrease in endomorphy during puberty in boys, but not in
girls. This implies that during puberty there is an absolute fat loss in boys, while
there is only a relative fat loss in girls (an analysis of fatness indicators was carried
out and published elsewhere, Nemeth et al. 1999).
Regarding the second component, mesomorphy, it can be stated that there exists
a sexual dimorphism in the development of the musculo-skeletal system from early
childhood just as for body fat. Though sexual dimorphism in body fat increases
considerably with age, the means of mesomorphy show only moderate increase in
sex difference of the musculo-skeletal system.
Changes in mesomorphy during puberty are very moderate in boys, while it
decreases during this period in girls. This is surprising, because it is well-known
that the musculo-skeletal system during puberty develops significantly, especially in
boys. Presenting our results, one must take into consideration that the values
obtained from age groups younger than 10 years are to be interpreted with caution
because of the relative insufficiency of the method used (as it was mentioned above,
see Carter and Heath 1990). On the other hand, mesomorphy is a relative value and
the relative girth measurements and indices of girth as well as the length
measurements in this sample (published elsewhere, Nemeth: Thesis 1999) increase
with age, too; the physique becomes more slender during puberty.
Changes of the third component, ectomorphy, reflect the lengthening of the body
as a result of the pubertal growth spurt in girls and the same phenomenon in boys
some two years later. Even more these changes indicate that growth is more
intensive in boys than in girls.
Changes in somatotype as a whole indicate that during the growth process, boys
become more muscular and linear and less "roundish". This is more or less
characteristic also of girls but only until onset of puberty and then they become
fatter, i.e. more feminine. In boys, the considerable increase in ectomorphy between
the ages of 12 and 15 show the intensive pubertal growth spurt.
Comparing our data with previous studies, one can see that single components
had changed with age in a similar direction, there are, however, also remarkable
differences in somatotypes. Carter and Heath (1990) in "Growth and ageing" chapter
of their handbook wrote that "In general, boys at young ages move from endo-
mesomorphy to ecto-mesomorphy and balanced ectomorphy-mesomorphy. During
adolescence, with increased muscle mass and complete ossification, mesomorphy
increases and ectomorphy decreases. In general, in girls, like boys, move from endo-
mesomorphy and balanced endomorphy-mesomorphy toward central somatotypes. In
adolescence and early maturity they move toward balanced endomorphy-
mesomorphy and meso-endomorphy" (Carter and Heath 1990, page 180).
306 CHAPTER 23
Carter and Heath based their statement on several somatotype studies, carried out
all over the world, in which the authors investigated children and adolescents and in
some cases also followed their changes in somatotype with age. It is good form to
mention first the Atlas of Tanner and Whitehouse (1982), based on their famous
Harpenden Growth Study. Further papers using Heath - Carter method are at our
disposal (in Belgium: Duquet 1980; in Bohemia and Moravia: Stepnicka 1976; in
Hungary: Farmosi 1982, Eiben (Kormend) 1985, Bodzsar (Bakony) 1982, 1991; in
Finland: Holopainen et al. 1984; in India (Gaddi, Raj put): Singh and Sidhu 1980,
(Jat Sikh Punjab) Kansal 1981, (Bania, Punjab) Kansal 1981, (Bangalore) Rangan
1982; in Venezuela (Caracas) Perez et al. 1985; in Nigeria (lle-Ife) Toriola and
Igbokwe 1985; in Brazil (Londrina) Guedes 1983; in the U.S.A. (Illinois) Slaughter
et al. 1980, etc. - see all these: Carter and Heath 1990 pages 160-164).
It is really illuminating to follow the wandering pattern of the mean changes in
somatotypes with age. We can demonstrate this with some Hungarian examples.
Eiben (1977) found in his Western Hungarian "Kormend Growth Study" of 1968, in
contrast to the present sample, that mean somatotypes in early childhood in both
sexes could be found in the ecto-mesomorphic field. In boys, mean somatotypes
shifted with age through the central field to the meso-ectomorphy, while in girls,
these means moved through the central field to the meso-endomorphic region. Ten
years later, in his 1978 Kormend Growth Study, Eiben found a striking difference:
endomorphy was far higher in both genders than in his previous study (Eiben 1985,
1988b). These later findings are more similar to those of our recent Budapest
sample.
Szmodis (1977) examined Hungarian boys and girls aged 5-17 years, among
them young athletes as well. Those boys were more ectomorphic and girls were less
endomorphic. Even mean somatotypes of the older girls in his sample were in the
central field of the somatochart.
Bodzsar (1982) investigated 10-14 year old Hungarian girls, and she found
central mean somatotypes, and increase of endomorphy and decrease of mesomorphy
with age. The same author published data of 6.5-14.5 year-old Hungarian boys and
girls in her other study (Bodzsar 1991). Mean somatotypes in this study differ
significantly from those of our recent sample. Somatotypes of boys in Bodzsar's
study moved from the ecto-mesomorphic field to the meso-ectomorphic one with
increasing age. Values of girls shifted from the ecto-mesomorphic filed through the
central field to the ecto-endomorphic region. According to this, endomorphy in
general was lower and ectomorphy in general was higher in Bodzsar's sample. The
only exception was that in 13-14 year-old girls in endomorphy were more or less
equal with that of our current sample.
Our Budapest study, presented here, provides convincing evidence that
somatotype analysis results in adequate information about changes in physique from
early childhood to adolescence. Changes in proportions and body composition
during puberty are mirrored in somatotypes.
Acknowledgements. This research was funded by the Hungarian National

Foundation for Scientific Research (OTKA grant No T 13098 and T 22599).
5. REFERENCES
AAHPER, 1965, Youth Fitness Test Manual. Washington D.C.
Bielicki, T., 1986, Physical growth as a measure of the economic well-being of population in twentieth
century. In Human Orowth (2nd ed.), edited by F. Falkner and l.M. Tanner (New York: Plenum
Press) Vol. 3 p. 283-305.
Bodzsar, B.E., 1982, The indices of the physique and the socio-economic factors based on a growth
study in Bakony girls. Anthropologiai KOzlemenyek (Budapest), 26, 129-134.
Bodzsar, B.E., 1991, The Bakony Orowth Study. Humanbiologia Budapestinensis, 22, pp. 210.
Carter, J.E.L., and Heath, H.B., 1990, Somatotyping. Development and Applications (Cambridge:
Duquet, W., 1980, Applicability of the Heath-Carter somatotype method to 6 to 13 year-old children [in
Flemish]. PhD Dissertation, Vrije Universiteit Brussel.
Eiben, 0.0., 1977, Changes in body measurements and proportions of children, based on the KOrmend
Orowth Study (in Hungarian with an English summary). Anthropologiai KOzlemenyek (Budapest),
21,43-58.
Eiben, 0.0., 1985, The KOrmend Orowth Study: Somatotypes. Humanbiologia Budapestinensis, 16,37-
52.
Eiben, 0.0., 1988a, History of Human Biology in Hungary. - Occasional Papers 2/4 pp 75. International
Association of Human Biologists, Newcastle upon Tyne, UK.
Eiben, 0.0., 1988b, Secular growth changes in Hungary. Secular trend of growth and biological
maturation of children in Hungary, based on the KOrmend Orowth Study (in Hungarian).
Humanbiologia Budapestinensis Suppl. 6., pp. 133.
Eiben, 0.0., and Panto, E., 1986, The Hungarian national growth standards. Anthropologiai
KOzlemenyek (Budapest), 30, 5-23.
Eiben, 0.0., and Panto, E., 1987/88, Body measurements in the Hungarian youth at the 1980s, based on
the Hungarian National Orowth Study. Anthropologiai KOzlemenyek (Budapest), 31,49-68.
Eiben, 0.0., Barabas, A., and Panto, E., 1991, The Hungarian National Orowth Study I. Reference data
on the biological developmental status and physical fitness of 3-18 year old Hungarian youth in the
1980s. Humanbiologia Budapestinensis, 1, pp. 123.
Farmosi, I., 1982, Results of constitutional and motor examinations of male students. Olasnik
AntropoloOkog Druotova Jugoslavije, 19,35-51.
Ouedes, D.P., 1983, Estudos antropometricos entre escolares. Revista Brazileire de Educacao Fisica e
Desportos, out/mar., 12-17.
Haag, H., and Dassel, H., 1975, Fitness Test. (Stuttgart: Verlag Karl Hofmann, Schorndorf bei
Stuttgart).
Heath, B.H., 1977, Applying the Heath-Carter somatotype method. - in Orowth and Development:
Physique edited by 0.0. Eiben (Budapest: Akademiai Kiado), p. 335-347.
Holopainen, S., Lumiaho, P., and Telarna, R., 1984, Level and rate of development of motor fitness,
motor abilities and skills by somatotype. Scandinavian Journal of Sport Sciences, 6/2, 67-75.
Kansal, D.K., 1981, A study of age changes in physique and body composition in males of two
communities in Punjab. PhD Thesis, Punjabi University, Patiala, India.
Martin, R, and Saller, K., 1957, Lehrbuch der Anthropologie, (Stuttgart: 1.-0. Fischer).
Matthews, D.R., 1973, Measurements in Physical Education. (Philadelphia: Saunders Co).
Nemeth, A. and Eiben, 0.0., 1997, Secular growth changes in Budapest in the 20th century. Acta
Medica Auxologica, 29, 5-12.
Nemeth, A., 1999, Somatic development of Budapest children and adolescents at the end of the XXth
Century and secular trend in Budapest. Ph.D. Thesis, Eotvos Lonind University, Budapest.
Nemeth, A., Bodzsar, B.E., and Eiben, 0.0., 1999, Comparison of fatness indicators in Budapest
children. Anthropologische Anzeiger, 57, 325-337.
Perez, B., Castillo, T.L., and Brief, F .K., 1985, Caracteristicas somatotipicas asociadas con la edad en
un grupo de escolares Venezolanos. Instituto de Investigaciones Economicas y Sociales. FACES
Universidad Central de Venezuela, Caracas, Venezuela.
Rangan, S.C.B., 1982, Validity of age, socio-economic belonging and dietary type as somatotype
determinants in boys of secondary schools. PhD Thesis, Bangalore University, Bangalore, India.
Simon, 1. ,and Renson, R., 1982, Evaluation of Motor Fitness. Report of the European Seminar on the
Evaluation of Motor Fitness. Leuven University Press, Leuven.
308 CHAPTER 23
Singh, S.P., and Sidhu, L.S., 1980, Changes in somatotypes during 4 to 20 years in Gaddi Rajput boys.
Zeitschrift fur Morphologie und Anthropologie, 71, 285-293.
Slaughter, M., Lohman, T.G., and Misner, J.E., 1980, Association of somatotype and body composition
to physical performance in 7-12 year old girls. Journal of Sports Medicine and Physical Fitness, 20,
189-198.
Stepnicka, J., 1976, Somatotypes of Bohemian and Moravian youth. acta Facultatis Medicinae
Universitatis Brunensis, Brno, 57,233-242.
Szmodis, I., 1977, Physique and growth estimated by Conrad's and Heath-Carter's somatocharts in
athletic children. In Growth and Development: Physique edited by O.G. Eiben (Budapest:
Akademiai Kiad6), p. 407-415.
Tanner, J.M, Hiernaux, J., and Jarman, S., 1969, Growth and physique studies. In Human Biology. A
Guide to Field Methods: IBP Handbook 9 edited by J.S. Weiner, J.A. and Lourie
(OxfordlEdinburgh: Blackwell Scientific Publishers) p. 1-76.
Tanner, J.M., and Whitehouse, R.H., 1982, Atlas of Children's Growth. Normal Variation and Growth
Disorders (New York: Academic Press).
Tanner, JM., 1978, Human growth standards: Contribution and use. In Auxology: Human Growth in
Health and Disorders, edited by I. Gedda, and P. Parisi (London: Academic Press), p. 109-122.
Toriola, A.L., and Igbokwe, N.U., 1985. Relationship between perceived physique and somatotype
characteristics of 10 to 18 year old boys and girls. Perceptual and Motor Skills, 60, 878.
WHO, 1976, New Trends and Approaches in the Delivery of Maternal and Child Care in Health
Services. Sixth report of the WHO Expert Committee on Maternal and Child Care, WHO Technical
Report Series no 600. World Health Organisation, Geneva.
APPENDIX
Table 1. Height of 3 - 18 year old Budapest boys (em)
Age, years N mean SD SE W

3± 38 99.26 4.15 0.673 90.8 - 108.1
4± 124 103.28 5.28 0.474 91.7-117.1
5± 144 110.32 4.97 0.414 99.0 - 125.6
6± 160 117.26 5.49 0.434 102.6 - 131.2
7± 142 122.49 6.10 0.512 107.1 - 136.7
8± 154 128.85 6.23 0.502 112.3 - 147.4
9± 164 134.95 6.35 0.496 114.5 - 151.0
10± 158 140.83 6.53 0.520 122.7 - 160.7
11± 162 145.99 7.53 0.592 125.3 - 168.7
12± 135 151.79 7.82 0.673 134.4 - 179.0
13± 181 158.21 8.60 0.639 138.3 - 182.2
14± 167 166.63 7.58 0.587 145.5 - 187.0
15± 209 171.34 8.25 0.571 144.9 - 189.8
16± 207 174.38 7.21 0.502 150.5 - 197.7
17± 276 175.70 6.97 0.419 156.9 - 197.4
18± 185 176.21 6.07 0.447 158.7 - 194.5
Table 2. Height of 3 - 18 year old Budapest girls (em)

3± 38 96.33 3.94 0.638 85.2 - 107.5
4± 138 102.08 4.60 0.392 91.2 - 113.9
5± 129 109.52 4.72 0.415 99.9 - 129.8
6± 135 116.60 5.05 0.434 105.8 - 129.8
7± 140 122.19 5.47 0.462 102.6 - 136.3
8± 169 129.07 5.58 0.429 112.3 - 144.5
9± 165 135.19 6.64 0.517 119.7 - 152.6
10± 188 140.03 6.82 0.497 125.2 - 163.4
11± 172 147.57 7.46 0.569 131.1 - 169.0
12± 153 152.79 7.72 0.624 130.1 - 172.5
13± 164 158.02 6.24 0.487 142.7 - 174.3
14± 136 161.32 6.33 0.542 137.3 - 176.3
15± 178 162.21 5.98 0.448 148.1 - 176.5
16± 214 163.73 6.26 0.428 148.6 - 180.2
17± 205 164.77 6.50 0.454 150.3 - 186.2
18± 147 164.34 6.31 0.520 149.7 - 183.8
310 CHAPTER 23
Table 3. Weight of 3 - 18 year old Budapest boys (kg)

3± 38 15.43 1.79 0.290 12.5 - 20.0
4± 124 16.43 2.34 0.210 12.0 - 27.0
5± 144 18.63 2.95 0.246 14.0 - 33.0
6± 160 2l.04 3.27 0.258 15.0 - 32.5
7± 142 23.55 4.40 0.369 16.0 - 43.0
8± 154 26.78 4.96 0.400 17.5 - 42.5
9± 164 30.63 6.12 0.478 18.0 - 49.0
10± 158 35.25 9.08 0.723 22.0 - 7l.5
11± 162 39.52 10.44 0.820 2l.5 - 74.0
12± 135 44.25 9.89 0.851 29.5 - 78.0
13± 181 48.01 1l.98 0.890 28.5 - 93.0
14± 167 55.06 10.72 0.830 34.0 - 95.0
15± 209 59.90 12.76 0.883 30.0 - 108.0
16± 207 65.47 1l.80 0.820 42.0 - 106.0
17± 276 67.12 12.99 0.782 41.0 - 145.0
18± 185 68.86 11.43 0.840 39.0 - 111.0
Table 4. Weight of 3 - 18 year old Budapest girls (kg)

3± 38 14.30 1.54 0.250 11.0 - 18.0
4± 138 15.69 2.23 0.190 9.5 - 21.0
5± 129 18.50 2.97 0.261 13.0 - 37.0
6± 135 20.94 3.59 0.309 14.0 - 39.0
7± 140 22.85 3.86 0.326 15.0 - 42.5
8± 169 26.92 5.41 0.416 17.5 - 45.0
9± 165 30.54 6.88 0.536 20.5 - 53.5
10± 188 33.99 7.50 0.547 22.0 - 56.5
ll± 172 39.43 9.35 0.713 23.5 - 83.5
12± 153 43.43 10.43 0.843 26.5 - 79.0
I3± 164 49.27 10.56 0.825 30.0 - 84.0
14± 136 52.93 9.74 0.836 37.0 - 89.5
15± 178 56.63 10.06 0.754 40.0 - 91.0
I6± 214 56.50 9.72 0.664 39.5 - 110.0
I7± 205 57.57 10.28 0.718 38.0 - 108.0
18± 147 57.06 9.92 0.815 41.0 - 111.0
Table 5. Somatotype components in 3 - 18 year old Budapest boys

EndomorphJ'.
3± 38 4.13 0.86 0.139 3 - 6.5
4± 124 3.84 0.97 0.087 2 - 6.5
5± 144 3.48 1.16 0.096 1.5-8
6± 160 3.23 1.11 0.088 1.5 - 7
7± 142 3.09 1.55 0.130 1 - 11
8± 154 3.17 1.58 0.128 1 - 10
9± 164 3.35 1.88 0.147 1 - 9.5
10± 158 3.64 2.12 0.169 1.5 - 9.5
11± 162 3.94 2.31 0.181 1 - 9.5
12± 135 4.26 2.35 0.202 1 - 10
13± 181 3.39 2.00 0.148 1 - 9.5
14± 167 3.33 1.89 0.146 1 - 9.5
15± 209 2.84 1.74 0.121 1-9
16± 207 3.17 1.73 0.120 1 - 8.5
17± 276 2.90 1.58 0.095 1-9
18± 185 3.23 1.68 0.123 1- 8
MesomorphJ'. - - - _.... _ . _ - - -
3± 38 5.62 0.41 0.066 5 - 6.5

4± 124 5.23 0.61 0.055 4 - 6.5
5± 144 4.86 0.68 0.056 3.5 - 6.5
6± 160 4.46 0.68 0.054 3 - 6.5
7± 142 4.47 0.81 0.068 2-8
8± 154 4.36 0.90 0.073 2.5 - 8
9± 164 4.27 1.04 0.081 2.5 - 7.5
10± 158 4.25 1.16 0.092 1.5 - 8.5
11± 162 4.38 1.31 0.103 2 - 8.5
12± 135 4.33 1.26 0.108 1.5 - 8.5
13± 181 4.00 1.37 0.101 1 - 8.5
14± 167 3.88 1.23 0.095 0-7
15± 209 3.73 1.2 0.083 1 - 7.5
16± 207 3.93 1.39 0.097 1- 8
17± 276 3.94 1.42 0.086 0.5 - 9.5
18± 185 4.03 1.35 0.099 0-8.5
EctomorphJ'.
3± 38 0.97 0.46 0.Q75 0.5 - 2
4± 124 1.38 0.76 0.068 0.5 - 4
5± 144 2.06 0.97 0.081 0.5 - 5
6± 160 2.63 1.00 0.079 0.5 - 5.5
7± 142 2.90 1.05 0.088 0.5 - 6.5
8± 154 3.15 1.25 0.101 0.5 - 6.5
9± 164 3.23 1.34 0.104 0.5 - 5.5
10± 158 3.26 1.46 0.116 0.5 -7
ll± 162 3.24 1.61 0.127 0.5 - 7.5
12± 135 3.17 1.50 0.129 0.5 - 6.5
13± 181 3.64 1.67 0.124 0.5 - 8.5
14± 167 3.71 1.52 0.118 0.5 - 7.5
15± 209 3.74 1.53 0.106 0.5 -7
16± 207 3.33 1.49 0.104 0.5 - 6.5
17± 276 3.30 1.46 0.088 0.5 - 7.5
18± 185 3.12 1.41 0.104 0.5 - 7.5
312 CHAPTER 23
Table 6. Somatotype components in 3 - 18 year old Budapest girls

EndomoTehr.
3± 38 4.79 1.15 0.186 3 - 7.5
4± 138 4.37 1.17 0.100 2.5 - 8.5
5± 129 4.33 1.32 0.117 2 - 10
6± 135 4.28 1.55 0.133 1.5 - 10.5
7± 140 3.87 1.41 0.119 1.5 - 10
8± 169 3.97 1.62 0.125 1.5 - 9
9± 165 4.13 1.77 0.138 1.5 - 9.5
101 188 4.36 1.86 0.136 1.5-10
11± 172 4.47 1.91 0.146 1.5 - 10.5
12± 153 4.68 1.65 0.133 1.5 - 9.5
13± 164 4.90 1.72 0.134 1.5 - 9.5
14± 136 5.17 1.82 0.156 1.5 - 9.5
15± 178 5.44 1.62 0.121 2 - 9.5
16± 214 5.33 1.59 0.109 2 - 9.5
17± 205 5.11 1.65 0.115 1.5 - 9.5
18± 147 5.30 1.40 0.116 2.5 - 9.5
MesomOTehp"
3± 38 5.34 0.66 0.107 4 - 6.5
4± 138 4.88 0.63 0.053 3.5 - 7
5± 129 4.70 0.76 0.067 3 - 8.5
6± 135 4.29 0.81 0.070 2.5 - 8.5
7± 140 3.98 0.82 0.069 2.5 - 7
8± 169 4.00 0.87 0.067 2.5 - 7.5
9± 165 3.90 0.97 0.075 1.5 - 7
101 188 3.90 1.09 0.080 1.5 - 8
11± 172 3.62 1.19 0.091 1.5 - 7.5
12± 153 3.45 1.17 0.095 1- 7
13± 164 3.43 1.30 0.101 1 - 8.5
14± 136 3.41 1.43 0.123 0.5 - 8.5
15± 178 3.71 1.36 0.102 0.5 - 8
16± 214 3.36 1.39 0.095 0.5 - 9
17± 205 3.40 1.35 0.095 0-8.5
18± 147 3.31 1.23 0.101 0-8
EctomoTehp"
3± 38 0.95 0.65 0.105 0.5 - 3.5
4± 138 1.53 0.87 0.074 0.5 - 4
5± 129 1.90 0.87 0.077 0.5 - 4.5
6± 135 2.52 1.03 0.089 0.5 - 4.5
7± 140 3.11 1.13 0.095 0.5 - 6
8± 169 3.15 1.21 0.093 0.5 - 6
9± 165 3.36 1.35 0.105 0.5 - 7.5
101 188 3.36 1.52 0.111 0.5 - 6.5
11± 172 3.48 1.48 0.113 0.5 - 6.5
12± 153 3.57 1.61 0.130 0.5 - 6.5
13± 164 3.32 1.54 0.120 0.5 - 6.5
14± 136 3.14 1.58 0.135 0.5 - 6.5
15± 178 2.62 1.33 0.100 0.5 - 6
16± 214 2.91 1.41 0.097 0.5 - 7
17± 205 2.89 1.30 0.091 0.5 - 6.5
18± 147 2.89 1.26 0.104 0.5 - 6
CHAPTER 24
DIFFERENTIAL RATE OF GROWTH OF THE HUMAN

BODY PARTS
M. PROKOPEC
National Institute of Public Health, Prague, Czech Republic
l. INTRODUCTION
Growth of the human body from birth to adulthood does not proceed as a simple
enlargement of the body tissues, organs and parts. It is obvious that body
proportions change considerably from the new-born to the grown-up male or female.
If we take the sizes of individual body parts of a child at the age of one month as a
starting point (zero) and the sizes of the same body parts at the age of 20 years as
final (100 percent), we get different rates of growth of individual body measurements
at each chosen age.
The body of the growing child and youth changes its proportions year by year.
Lengths, widths, circumferences of the body and of its parts and body weight set out
for their individual journeys from one month of age towards their goal in early
adulthood at the age of 20. They proceed at different speeds reflecting thus the
changing morphology and individuality of the child in the course of its pre-school
and school years, and throughout its teens until each of them reaches its 100 percent
fInal size at the age of 20 where they all meet (Marshall 1977).
Ages at which the CUIVes of individual measurements are most wide apart in their
paths may indicate a period of unbalanced, unconsolidated, disproportionate state of
the body which may be associated with greater demand on energy regulating body
integrity and thus resulting in an increased health risk. It may be useful to know at
which age this time period comes in a child population as a whole or in an
individual. The fIrst decade of life is a period at which the quickly developed head
(and brain) governs the yet much less developed body parts which may be an
advantage in learning.
The body proportionality of the human being which has been reached at the age
of 20 developed hundreds of thousands of years and may be considered as stabilised,
as a fInal product of human evolution and any stage preceding it may be thus viewed
as immature, unfinished, as a state of transition. Naturally, some individuals may
reach the mature stage earlier than at 20 years, say already at the age of 16 in females
and 18 in males. It is obvious that further changes in the human body take place
after the age of 20 years, but they need not be considered here as important. On the
313
314 CHAPTER 24
other hand the body proportions of a physiologically normal new-born are the
product of its prenatal development "in utero" and must be taken as granted.
The aim of this paper is:
• To show the widely different rates of growth of various body parts
(measurements) in boys and girls followed-up from one month to 20 years of
age.
• To study changes in body proportions throughout the entire growing period,
and to enable to determine specific body proportions (mutual relations of
various body parts) at any given age.
• To point out ages at which the differences in the rate of growth in individual
measurements are greatest and when they seem unimportant.
• To detect mutual similarities and diversities in the pattern of growth between
individual parts (measurements) of the body.
• To determine differences in boys and girls in this respect.

Fourteen body measurements (heights, widths, circumferences and weight) were
ascertained in a group of 300 Prague children (half of them boys and half girls),
followed-up longitudinally from birth to the age of 20 years from 1956 through
1980. Only healthy children born on Wednesdays from one Prague district were
involved in the study and measured five times in the first year of life (at the age of 1
month, 3, 6, 9 and 12 months) and semiannually thereafter (at each birthday and 6
months later). (Kapalin et al. 1969, Prokopec 1980). The initial number of children
diminished throughout the smvey by about 30 percent (the sex ratio remained
unchanged). The methodology was the same as designed by Falkner (1955) for the
teams of the Centre International de l'Enfance in Paris and by Tanner (1947).
Internationally agreed instruments and methods of taking the measurements were
used throughout the survey (anthropometer, large and small spreading callipers,
gliding calliper and tape measure). The measurements were selected as to present the
weight of the body, its heights and lengths, widths of the shoulders, pelvis,
humerus and femur, circumferences of the head, chest, abdomen, upper arm and calf,
and the breadth and length of the head. Skinfolds were also measured, but not
involved in this study. The length of the body in babies was measured in a lying
position, starting at the age of 18 months (at 2 years in some instances) in standing
erect position. This was also the age, at which the sitting height was measured
instead of crown-rump distance, which had been measured in babies in lying
position.
The measurements and measuring units were as follows:
1. Body weight (kg)
2. Body height -length (em)
3. Sitting height - crown-rump length (cm)
4. Chest circumference (cm)
5. Head circumference (cm)
6. Upper ann circumference (cm)
7. Calf circumference (cm)
8. Biepicondylar femur (mm)
9. Biepicondylar humerus (mm)
PREPARATION OF A CAMERA-READY MANUSCRIPT 315
10. Bicristal (biiliocristal) width (cm)

11. Biacromial width (cm)
12. Abdominal circumference (cm)
13. Head breadth (cm)
14. Head length (cm)
Means and SD's of each measurement were calculated separately for each sex and
age group. Distance growth curves were drawn from the means for each
measurement. The gross increment (difference) from 1 month to 20 years of each
measurement was given as 100 percent and ages (in years and decimals of the year),
corresponding to selected percent values were put into a table (Table 1).
Graphical presentation of growth curves of individual measurements was used to
follow the differential rate of growth of body parts throughout the period of growth.
To maintain a clear picture of the curves, only typical curves are shown in the
figures. Measurements, the curves of which are similar to those already depicted in
the figures, were not included into the figures and are only mentioned in the text.
3. RESULTS
The results are given in Table 1 for both sexes and in Figure 1 for males and in
Figure 2 for females.
The following measurements: bicristal diameter, stature, sitting height and
biacromial diameter had similar patterns of growth. Thus, the curve of the bicristal
diameter represents the rest of these measurements in the figures. The type of the
curve of the chest circumference is similar to upper arm circumference. The latter is
thus not shown in the figures.
An extremely wide variability in the rate of growth of individual measurements
appear from Table 1 and from Figures 1 and 2. The majority of curves follow the
typical trend of stature (a steep start from one month to 3 years, followed by smaller
and regular increments until pubertal spurt after which a transition to a steady state
follows). This is not the case in head measurements (head breadth, head length and
head circumference) the CUIVes of which are most advanced of all others straight from
the beginning. Their CUIVes are followed by those of the femur and humerus, which
intersect the former CUIVes at the age from 12 to 14 years in boys. Only the CUIVe of
the humerus does the same between 9 and 12 years in girls.
The CUIVe of the bicristal diameter (which represents also stature, sitting height
and biacromial diameter) occupies the medium position among other growth CUIVes
in both sexes. The curve of the calf intersects that of the bicristal width at the age of
about 10 years both in boys and girls. The calf CUIVe in boys runs alone before the
age often, whereas in girls it twists with that of abdomen circumference in the same
period of time. The curves of abdominal and chest circumferences follow the same
path until the age of 13 in boys, whereas in girls they run separately but cross before
the age of 11 years.
The CUIVe of weight runs wide apart from the curves of other body parts in both
sexes, showing thus how much weight lagged behind the other measurements. It
remains separate until the age of 20 in girls, but it has been crossed by the CUIVe of
chest circumference shortly after 14 years of age in boys.
316 CHAPTER 24
100
90 L head br.
head c.
80
70 r h ". tI 1 / ~ ~~d( // 'abdomen. c.
~ 60
~
Q.. 50
"-humerus
40 biiliocrist. d.
30
20 Boys
10
0
0 2 4 6 8 10 12 14 16 18 20
Age, years
Figure 1. Percent of total increment from 1 month to 20 years of age attained by

selected measurements in boys from the City of Prague
\00
90
80
70
....
5u 60
[)
Q.. 50
40 biiliocrist. d.
30
20 Girls
10
0
0 2 4 6 8 10 12 \4 16 18 20
Age, years
Figure 2. Percent of total increment from 1 month to 20 years of age attained by

selected measurements in girls from the City of Prague
Table 1. Age (in years) at which children attain the given percent of total increment
from 1 month to 20 years of age (M: male; F: female)
Percent 10 20 30 40 50
M F M F M F M F M F
Weight 1.0 0.8 4.0 3.3 6.8 5.9 9.2 8.0 11.3 9.5
Stature 0.5 0.6 1.3 1.2 2.5 2.2 4.0 3.5 6.0 5.3
Sitting h. 0.3 OA 0.8 0.9 2.0 1.8 3.6 3.2 6.0 5.3
Chest eire. 0.3 0.3 0.8 0.7 3.3 2.5 6.1 5.6 8.5 7.6
Head eire. 0.1 0.1 0.2 0.2 OA 0.5 0.7 0.6 1.2 1.1
Upp. arm e. 0.3 0.2 0.5 0.5 2.5 lA 6A 5.0 9.0 7.3
Calf eire. 0.2 0.2 0.5 0.6 1.6 l.6 4.8 4.1 7.0 6.6
Femur 0.2 0.3 OA 0.5 l.0 0.9 2.9 1.7 3A 3.0
Humerus 0.3 0.3 0.7 0.5 1.5 l.0 3.2 2.1 5.2 3.6
Bieristal w. 0.2 0.3 0.8 0.8 1.8 1.6 3.2 3.0 5.2 5.1
Biaerom. w. 0.3 OA 1.1 l.0 2.3 l.9 4.2 3A 6.2 5.1
Abdom. eire. 0.2 0.2 0.7 0.6 3.5 l.6 6.0 4.2 8.3 6.5
Head breadth 0.1 0.1 0.3 0.2 OA 0.3 0.6 0.6 0.9 0.8
Head length 0.2 0.2 OA OA 0.6 0.6 0.9 0.8 1.5 1.3
Percent 60 70 80 90 100
M F M F M F M F M F
Weight 12.6 10.9 13.7 12.0 15.0 13.2 l6A 15.0 20.0 20.0
Stature 8.0 7.1 lOA 9.0 12.7 10.9 14.3 12.6 19.0 18.0
Sitting h. 8.6 7.5 11.5 9.7 13.3 11.7 15.0 13.3 20.0 19.0
Chest eire. 10.8 9.3 12.9 10.7 14.3 12.2 16.0 13.9 20.0 19.0
Head eire. 2A 2.0 5.3 4.5 9.0 7.2 14.0 11.5 20.0 20.0
Upp. arm e. 11.0 9.0 12.9 10.7 14.6 12.9 16.3 14.6 20.0 18.6
Calf eire. 9.1 8.6 10.6 lOA 12A 11.9 14A 13.8 19.0 18.0
Femur 5.5 4.7 7.9 7.0 10.8 9.0 13.3 11.8 19.0 19.0
Humerus 7.3 5.1 9.7 6.7 12A 8.6 13.9 10.7 19.0 19.0
Bieristal w. 8.0 7.6 10.7 9.8 12.8 11.8 14.7 13.7 19.0 19.0
Biaerom. w. 8.6 7.2 11A 9.2 13A 11.2 15.0 13.2 20.0 19.0
Abdom. eire. 10.5 8.5 12.9 10.6 14.9 12.7 16.8 14.7 20.0 19.9
Head breadth 1.6 1.3 3.2 2.2 6.0 4.1 13.3 8.2 20.0 20.0
Head length 2.7 2.2 5.1 3.9 9.7 6.8 15.0 11.3 20.0 20.0
The greatest difference in years between the curves of the head breadth and weight
(extremes), was found to be in boys at the level of 60 percent and amounts to 12
years of time space. The same was found in girls at the level of 70 percent and
amounts to 10.5 years of time space.
The most rapid rate of growth show the head measurements. The head breadth
reached the level of 80 percent of the total increment from 1 month to 20 years at the
age of 6 years in boys and at the age of 4.1 years in girls. The head circumference
and the head length reached the level of 80 percent in boys at 9 and 9.7 years,
respectively; and in girls at 7.2 and at 6.8 years, respectively.
318 CHAPTER 24
The post-pubertal spurt in head measurements (only in boys after the age of 14
years) may be explained by the development of frontal sinuses and by thickening of
the skull bones and skin at that age. This is a visibly clear difference between boys
and girls in Figures 1 and 2.
The following picture may be seen in boys at the level of 50 percent of the total
increment from 1 month to 20 years of age:
The head breadth reached this level at 0.9 years
head circumference at 1.2 years
head length at 1.5 years
biepicondylar femur at 3.4 years
biepicondylar humerus and bicristal diameter at 5.2 years
stature and sitting height at 6.0 years
biacromial diameter at 6.2 years
calf circumferen::e at 7.0 years
chest circumference at 8,5 years
upper ann circmnference at 9.0 years
abdominal circmnferences at 8.3 years
and weight at 11.3 years
The 50th percent level was reached by individual measurements in girls at

following ages:
Head breadth at 0.8 years
head circmnference at 1.1 years
head length at 1.3 years
biepicondylar femur at 3.0 years
biepicondylar humerus at 3.6 years
bicristal diameter and biacromial diameter at 5.1 years
stature and sitting height at 5.3 years
abdominal circmnferences at 6.5 years
calf circumferen::e at 6.6 years
upper ann circmnference at 7.3 years
chest circmnference at 7.6 years
and weight at 9.5 years
Similarly, ages may be easily read from Table 1 and/or from Figures 1 and 2 for
any body dimension at the aibitraIy percent level in boys and in girls.
Different rates of growth in the measurements followed up in this study may be
also detected from Figures 1 an 2 when we look at aibitraIy percent levels, reached
by individual measurements at given ages. The greatest span is between the curves
of head breadth and body weight. In boys at the ages from 3 to 6 years it amounts to
53 percent of the total increment from 1 month to 20 years. Whereas weight attained
at the age of three years represents only 17 percent of the total increment, head
breadth attained at the same age equals 71 percent of the total increment.
At 6 years the body weight reaches in boys 27 percent of the total increment,
circumferences of the thorax (also upper arm) and abdomen reach 40 percent; calf
circumference 45 percent; bicristal diameter (and also stature, sitting height and
biacromial diameter), 54 percent; biepicondylar humerus, 55 percent; biepicondylar
femur 63 percent; head circumference and head length, 73 percent; and head breadth,
81 percent of the total increment from 1 month to 20 years.
The greatest span between weight curve and head breadth curve in girls is at the
age of 4 years. It amounts to 56 percent of the total increment. At this age the
weight reaches only 22 percent and head breadth, 79 percent of the total increment.
Weight reached in girls at the age of 6 years, 30 percent of the total increment;
thorax circumference (and the upper arm circumference), 42 percent; abdominal and
calf circumferences, 47 percent; bicristal (and also stature, sitting height and
biacromial diameter), 53 percent; biepicondylar humerus, 65 and biepicondylar
femur, 66 percent; head circumference and head length, 76 and 77 percent,
respectively; and head breadth, 86 percent.
4. DISCUSSION
We may postulate, that ideal proportions of the human body are attained at early
adulthood (say, 18 to 20 years), which is the result of hundreds of thousands years
long development of man on the Earth. Thus we may consider human proportions
between birth and 20 years as tmnsitional, unbalanced, unfinished, and suspect the
time period of greatest diversity in the rate of grow of the body parts to be a health
risk. On the other hand, we cannot deny beauty in children at any given age, or
body efficiency in sports to teenagers with still evident disproportions between trunk
and extremities.
We may also deduce from the amazingly great variability in the rate of growth of
individual body parts that the development of an individual works stepwise,
concentrating on one structure at a time, and on another one when its time comes.
Throughout the period of child development from birth to the teens the body is
dominated by the growth of the head. The necessary body structures (skeleton,
muscles of the extremities) develop less quickly, later the chest and trunk develop,
but weight is minimal and the capacity of neurones in the brain may serve learning
much more effectively than later in life. Starting at puberty the body comes to power
and the reign of the brain is diminishing.
There is much beauty in the unfinished human with a big head and big eyes.
The will to learn and the dependence on society makes the young preferred, beloved
and well treated. Their relative rareness in the society helps to support such an
attitude (children under 14 years make up for a fraction of the population in an
industrialised country). Once one enters the domain of adults, one is lost in
triviality. There are too many adults in an ageing society.
After all, the period of life during which the man and woman are in the making is
miraculous.
5. CONCLUSION
The ages at which boys and girls attain a certain percentage of the total increment
from 1 month to 20 years of age showed that girls are ahead of boys in all
measurements. At the level of 50 percent of the total increment from 1 month to 20
years of age the difference between the sexes amounts to 0.1 years in the head
320 CHAPTER 24
measurements as mlmmum, and 1.8 years in weight, chest and abdomen

circumferences as maximum in favour of girls.
At the level of 80 percent of the total increment from I month to 20 years of age
the difference between the sexes was from 0.5 year in calf circumference to 3.8 years
in biepicondylar humerus in favour of girls.
The boys differ from the girls quite distinctly in late (after l3-14 years of age)
increments of the head measurements under follow-up.
The age of about 6 years in both sexes corresponds with the greatest diversity in
the attained percentages of "maturation" of the body parts followed up in the study.
The age of about 6 years is thus for the children critical because of the greatest
disproportion between the measurements under study.
The levels at about 60 percent of the total increment from one month to 20 years
of age in boys and at about 70 percent in girls are critical for the children because of
the greatest age variation in dynamics of the measurements.
Extreme growth curves were represented on the one hand by the head
measurements (quick development) and by body weight (slow development) on the
other. Some of the measurements followed the pattern of growth in stature (sitting
height, biacromial diameter and bicristal diameter). Another group with similar
trends of growth were abdominal circumference, chest circumference, upper arm
circumference and calf circumference.
The head grows the quickest. It is followed by the growth of skeletal structures,
as well as by increments of heights, lengths, widths and circumferences, first of the
extremities and then of the chest and abdomen. Body weight showed the slowest
increment rate.
Acknowledgement. The author thanks Marcela Havlinova for drawing the figures
and Dr. Pavel Chyle for the revision of the English text.
6. REFERENCES
Falkner, F., 1955, A baseline of investigations for longitudinal growth studies in the child. Centre
International de l'Enfance, Chateau de Longchamp, Bois de Boulogne, Paris XVIo.
Kapalin, V., Kotaskova, J., and Prokopec, M., 1969, Physical and Mental Development of the
Contemporary Generation of Our Children. [In Czech, English Summary] (Prague: Academia) 300
pp.
Marshall, W.A., 1977, Human Growth and its Disorders (London, New York, San Francisco: Academic
Press) p.179.
Prokopec, M., 1980, Longitudinal Follow-up of Physical and Mental Development and Health State of a
Selected Group of Children from Prague: Final Report [In Czech] (Prague: National Institute of
Public Health) 126 pp.
Tanner, 1M., 1947, A Guide to American Growth Studies. Yearbook of Physical Anthropology, 3, 28-
33.
CHAPTER 25
SHORT-TERM GROWTH
M. HERMANUSSEN
Aschauhof, Altenhof, Germany
1. INTRODUCTION
Child growth is a dynamic process that is usually measured at infrequent intervals.

Though at this level, body stature often appears to follow certain growth channels,
the majority of children grows non-linearly.
Regular measurements of body stature for scientific purpose have been published
since more than 200 years (Boyd 1980). Most of these historic studies were based on
measurements taken at annual, half-annual and even monthly intervals, and this
regimen has been kept since. Growth velocity is unevenly distributed throughout the
year with seasonal variations of stature increment both in healthy children, and in
children during growth hormone therapy (Hermanussen 1998). Yet, most of the
within-year variance of stature increment remained unexplained, and was commonly
attributed to measurement error. For many years, it was common knowledge that "a
satisfactory assessment of a child's growth cannot be made over a period of less than
year" (Marshall 1971). Tanner added in 1979 that "increments of growth at a cellular
level are discontinuous; but at the level of bodily measurements. we can only discern
complete continuity, with a velocity that changes gradually from one day to the
next" (Tanner 1978). This was published even in view of obvious circadian postural
changes in height that were popularly known since Medieval times, and scientifically
documented since more than 100 years (Camerer 1880, MaIling-Hansen 1886). But
circadian variation in stature was not considered a true growth process as the morning
maximum was regularly followed by shrinkage in height during the day, similar
diurnal variation of length were also seen in the extremities (Valk et al. 1983b).
In 1971, an ulnar length measuring (Valk 1971) device was developed that
significantly improved short-term growth measurements as it cut down the technical
error to 0.21 mm, compared to approximately 2 mm of body stature measurements,
and thus enabled visualising growth within few weeks. In 1983, an even more
attractive technique was developed for accurate non-invasive determinations of the
lower leg length (knemometry). This device is easier to handle, it has a measurement
error between 0.09 mm (Valk et al. 1983a) and 0.16 mm (Hermanussen et al.
1988a), and is observer independent (Wales and Milner 1987a). Initially, intervals of
three weeks were recommended for short-term growth measurements, but it soon
became obvious that intervals of one week were technically feasible, and provided
321
322 CHAPTER 25
evidence for non-linear progress of growth at short-term (Hermanussen et al. 1988b,

Wit et al. 1987). Besides the original device, similar techniques for lower leg
measurements were inaugurated, a modified Danish knemometer by Michaelsen et al.
(1991), a Polish knemometer (Hulanicka et al. 1999), a "knee height measuring
device" by Cronk et al. (1989), a portable knemometer by Davies et al. (1996) for
measuring sitting individuals, a miniaturised portable knemometer (1.5 kg) for field
work (Hermanussen, unpublished), and an Italian device (Benso, personal
communication 1992). Mikro-knemometry is available for lower leg length
measurements in rodents (Hermanussen et al. 1992, Hermanussen et al. 1995), and a
particularly small version (mini-knemometer) for the measurement of infants, or
neonates and premature infants within their incubators (Hermanussen, Seele 1997,
Kaempf et al. 1998). The latter devices operate with a precision that reduces the
amount of variance due to measurement error, to five percent of the total variance of
24 hour measurement series, and thus visualise length changes within periods of less
than one day.
High precision measurements of the lower leg are influenced by numerous factors
that are not related to growth, such as weight (Ahmed et al. 1996), leg position of
the subject, water uptake or subcutaneous fat (Hermanussen et al. 1988a), and even
reflect soft tissue changes after single injections of growth hormone (Hermanussen
and Sippell 1985). Knemometry may also be affected by observer bias. Routine
measurements are usually performed by one investigator who is aware of the
previously measured leg length. But this problem may be overcome by "blinding"
the observer - random zero method (Ahmed et al. 1995). This technique involves
measuring from a baseline value of which the operator is not aware.
2. THE PHYSIOLOGY OF SHORT-TERM GROWTH

When body stature is measured quarterly or monthly, an irregular incremental pattern
becomes obvious with a number of large scale endogenous components such as
series of pre-pubertal and pubertal growth spurts and exogenous components such as
seasonal influences on height gain, and influences of the psychosocial and economic
background reflecting living standards and economic conditions of the respective
subjects. When measuring body stature at shorter intervals, incremental patterns
appear even more irregular, and a number of short scale components become apparent
that are distinct from measurement error. In 1982, Togo and Togo published time-
series analysis of monthly measurements of five children, and stated literally that
"body stature fluctuated enormously, instead of being smooth, even after removal of
seasonal and irregular components from the original series" (Togo and Togo 1982).
We investigated the predictive quality of short-term incremental rates obtained at
various intervals between less than one week and several months both by stadiometer
and knemometer, and found little predictability of short-term length increments on 6-
month growth rates, quite regardless of the technique used. We concluded that the
irregularity of short-term incremental patterns was largely responsible for the poor
predictability of long-term increments, and not the measurement error. Different
opinions however, existed on the proportionality of short-term growth. Whereas
Wolthers et al. found identical annual ratios between height velocity and lower leg
growth velocity in 27 school children, Dean et al. reported that there was no month-
to-month consistency in the ratio of lower leg and height growth. They found an up
SHORT-TERM GROWTH 323
to 4-fold variation in the intra- and inter-individual monthly lower leg growth rates,
and they failed to provide evidence for any correlation between one month lower leg
growth and annual height growth, when analysing short-term measurements of lower
leg length and body stature in 90 healthy children aged 3-16 years. Several authors
concluded that growth of different parts of the skeleton and variable interval growth
rates limit the ability of knemometry to predict long-term growth (literature reviewed
in Hermanussen 1998).
It is questionable whether short-term length increments should be used to predict
long-term growth, at all. From the very beginning, several authors have
demonstrated significant non-linearity of short-term growth by the use of weekly
measurements of lower leg length, and though most authors did not further
characterise the dynamics of short-term growth, they have made clear that linear
extrapolations of non-linear events do not appear to be very useful.
3. GROWTH AS A PERIODIC EVENT

Measurements of the lower leg at weekly intervals suggested repetitive "mini growth
spurts" once every 30 to 55 days that took place in both legs, and gave rise to the
idea of periodicity in child growth. This impression was further enforced by
analysing 23 daily individual series of lower leg measurements. Correlations were
calculated between the individual measurement series and given sine functions, with
suggestive evidence for periodicity once every 5 to 10 days. As circaseptan
(approximately-once-every-seven-day) periodicity would tend to interfere with weekly
measurements, speculations appeared justified that mini growth spurts once every 30
to 55 days might have resulted from aliasing between an endogenous rhythm of 5 to
10 days with the weekly measurement rhythm. Yet, investigations in series of leg
length measurements in rats and human neonates, applying time series analysis,
failed to support further evidence for strictly periodic behaviour of short-term growth
(Hermanussen et at. 1998).
4. GROWTH AS A SALTATORY EVENT
Measurements of body stature at weekly, semi-weekly and daily intervals led to a

mathematical model suggesting an aperiodic pattern in human growth, with
alternating episodes of stasis and saltations. Lampl et al. (1992), Lampl and Johnson
(2000) performed serial length measurements of normal infants. Ten infants were
measured weekly, 18 semi-weekly and three infants daily, during their first 21
months. Growth in length seemed to occur by discontinuous, aperiodic saltatory
spurts. The spurts appeared to be 0.5 to 2.5 centimetres in amplitude during intervals
separated by no measurable growth of two to 63 days duration. The authors
suggested that 90 to 95 percent of normal development during infancy was growth-
free and length accretion was a distinctly saltatory process of incremental bursts
punctuating background stasis. Similar growth characteristics were found in series of
daily measurements during adolescence and in data published previously by Ashizawa
and Kawabata (1990), Lampl et aI., (1998). This model became very popular, yet,
several authors failed to confirm the saltation and stasis hypothesis. Heinrichs et al.
determined the frequency distribution of daily growth velocities, and tried to compare
324 CHAPTER 25
the observed distributions with those predicted by the saltatory, and by continuous
growth models. The authors concluded that their data were incompatible with the
saltatory model. Similar findings were published by us, and further longitudinal
studies have shown that only a minority of gain in stature can be said to be truly
saltatory. It was suggested that alternate series of saltatory events and stasis simply
are a result of the assumptions of the mathematical model, as the model only allows
"increase" (saltation), or "no increase" (stasis). The model does not seem applicable
for truly linear growth (Herman us sen 1998).
5. GROWTH AS A CHAOTIC PROCESS

When body stature is measured at frequent intervals one can determine growth
velocity for annual, half-annual, quarterly, and shorter intervals. The more the
measuring interval decreases, the more increases the complexity of the height
velocity pattern. Even when possible artefacts of measurement errors are ignored, the
pattern shows attributes of fractal behaviour (Wales and Gibson 1994). Increasing
evidence obtained from daily series of high accuracy measurements suggests chaotic
behaviour of short-term growth (Hermanussen et at. 1998).
6. DAILY GROWTH MEASUREMENTS

A major step towards a better understanding of the short scale components of growth
took place, when series of several measurements per week or even daily
measurements were analysed. Daily measurements were first published by Ashizawa
and Kawabata who measured body stature of two Japanese children between June
1984 and May 1985 (1990) and provided a brilliant description of short-term growth.
Kobayashi and Togo reported similar data in two children and one adult (1993). Daily
measurements of stature were also reported by Lampl et at. (1992) who did not only
provide graphs of short-term individual growth, but developed a growth model
suggesting alternating episodes of stasis and saltations. We have presented daily
series of lower leg length measurements in 1989 with evidence for short growth
spurts once every 5 to 10 days, and since the introduction of mini-knemometry,
several authors have investigated daily growth in healthy neonates and infants
(Hermanussen and Seele 1997, Kaempf et at. 1998, Keller et at. 1998, 1999,2000b)
and confirmed the existence of mini growth spurts.
7. ANIMAL STUDIES ON SHORT-TERM GROWTH

In view of the increasing popularity of the study of short-term growth, we
established an animal models for experimental auxology, and measured the distance
between knee and heel of the rear lower leg in rabbits, using a digital goniometer
that discriminated at intervals of 0.00610. The goniometer is fixed to two measuring
arms with metallic holders. The animal's leg is placed between the holders. The
rotational angle between the measuring arms indicates the distance between the
holders, and has to be converted into micrometers (Hermanussen et at. 1992). Daily
measurements with this micro-knemometer revealed an irregular pattern of repetitive
mini growth spurts very similar to what has been shown in lower leg growth of
children. The technique was compared with X-ray stereophotogrammetry after

implanting tan tal bone markers close to the epiphyseal growth plates of animals.
There was significant correlation between X-ray stereophotogrammetry and micro-
knemometry within 24 hours (Hermanussen et at. 1992).
In contrast to the irregular sequence of growth spurts observed by us, Klein et at.
(1994) described linear leg increments in rabbits using daily X-ray measurements
after insertion of metal pins. Yet, these studies were based on short observation
periods of only nine days which may be too limited for detecting non linear growth
(Hermanussen et al. 1992).
Mikro-knemometry was also applied in laboratory rats that were measured at brief
intervals of less than 24 hours. We found significant circadian variation with
minimum length increments after midnight, and maximum increments between 0600
hand 0900 h (Hermanussen et al. 1998) that coincided with the diurnal variation of
mitotic activity of chondrocytes of the growth plate. Measurements at exactly 24-
hour intervals showed repetitive mini growth spurts, at various intervals between
two and eight days (Hermanussen 1998). Short-term growth studies in healthy male
and female rats, in castrated males, and in growth hormone deficient animals without
and during growth hormone replacement therapy, showed different growth patterns in
these groups. Healthy rats tended to produce spontaneous mini growth spurts at
intervals around 5.5 days, whereas the spurts appeared at shorter intervals in the
growth hormone deficient animals. Investigations of daily growth in weight and
lower leg length during and after food restriction (Hermanussen et al. 1996)
confirmed the existence of mini growth spurts in rats, and provided evidence for
severe disturbances of this pattern during starvation. Further investigations have been
published on sexual dimorphism in rat growth (Rol de Lama et at. 1998), and the
differential effects of growth hormone on peri-pubertal rats with no growth
stimulation in the male animals (Rol de Lama et at. 2000).
8. THE MATHEMATICAL ANALYSIS OF SHORT-TERM GROWTH

Numerous mathematical functions have been published to describe growth (Ledford
and Cole 1998, Milani 2000), but none of these have been applied to short-term
series of lower leg length measurements.
Mathematical analysis of weekly measurement series has usually been restricted
to the calculation of weekly increments. More sophisticated approaches besides the
saltation and stasis model mentioned above, are rare. Wit et at. (1987) utilised 2nd,
3rd, and 4th order polynomials for describing non-linearity of short-term growth.
This approach, however, does not provide characteristics of individual short-term
growth, i.e., in the sense of detecting periods of catch-up, growth arrest, periodicity
and other phenomena. We tried descriptions of short-term growth using techniques
similar to the moving average (Hermanussen et at. 1988b), but this approach
depends on smoothing intervals and might lead to false estimates on growth periods
due to aliasing (Hermanussen 1998). Thus, we currently favour a technique that
enables to describe those sections of an individual growth curve that are characterised
by multiple bursts of short-term growth. We introduced a numerical algorithm which
is suitable to analyse time series in the sense of detecting local structures. The
algorithm recognises single incremental bursts by their characteristic S-shape; in
other words, the numerical algorithm recognises the mathematical equivalent of what
326 CHAPTER 25
has been named an "incremental burst" or "mini growth spurt". These local
structures were represented by parameterised functions, in this case, by so-called
Gompertz functions, i.e. double exponential functions that contain a central S-shaped
structure.
Series of measurements were converted into series of time windows, containing
either exactly one Gompertz function or no Gompertz function. If a Gompertz
function was detected within a time window, the respective "mini growth spurt"
could be characterised by the three parameters of this function, i.e., amplitude,
inflection point (the day of peak growth velocity), and the slope. The slope
determines rapidity and, thus, alludes to the duration that one incremental burst needs
for completion. In healthy neonates (Herman us sen 1998), we found a mean
amplitude of mini growth spurts of 2057 ~m (SD 1132 ~m) and a mean peak
growth velocity of 85 ~m per hour. The spurts appeared once every 4.2 days on
average, but the standard deviation of 1.9 days indicated that the time interval
between subsequent spurts varied considerably. Time series analysis confirmed that
mini growth spurts did not occur periodically.
Similar characteristics of mini growth spurts were detected in growing rats. 5018
quadruplet daily measurements of lower leg length were performed in 62 female, and
81 male rats. We detected and analysed 684 local structures (mini growth spurts) and
found a mean amplitude of 2153 ~m (SD 1034 ~m) in female rats, and of 2958 ~m
(SD 1614 ~m) in males, with peak growth velocities that were lower than in
humans. Partial growth hormone deficiency led to a modification of mini growth
spurts in the rats with a reduction of spurt amplitude, and an increase in peak growth
velocity that was both reversed when exogenous growth hormone was administered.
The interval between subsequent mini growth spurts ranged between 4.2 and 4.6
days, with a large standard deviation (Hermanussen et at. 1998).
First in animal studies and later in neonates (Keller et af. 1998), we detected
growth periods that contained no mini growth spurts. The lower leg length increased
linearly, but the respective time windows appeared "empty", i.e., the algorithm was
unable to detect an S-shaped structure of the increments. Thus, evidence had occurred
that short-term growth may not always proceed in spurts. This must be taken into
consideration when applying mathematical models for the description of short-term
growth.
9. CLINICAL IMPLICATIONS OF SHORT-TERM GROWTH

MEASUREMENTS
Quite immediately after the first publications of short-term measurements, several
groups started to study the usefulness of these techniques for clinical purposes,
particularly in respect to the question whether accurate lower leg length
measurements might shorten the time for growth rate assessments. Short-term lower
leg increments were used to investigate the effects of corticosteroids on growth, renal
disease, they were correlated to parameters of bone turnover, and they were related to
urinary growth hormone excretion.
The first clinical study on short-term growth stimulation, monitored the effects
of sex steroids on 3-week ulnar length increments in 10 boys before and after a 4-day
intra-venous administration of testosterone, dihydrotestosterone and estradiol. The
study showed significant effects of hormone administration on ulnar length
increments, but it remained uncertain whether this effect was a brief soft tissue
reaction similar to that seen after growth hormone administration (Hermanussen,
Sippell 1985), or whether it presented the beginning of a long-term stimulation of
growth. In order to study the dynamics of short-term growth, we investigated the
growth pattern that followed the transfer from three times weekly intra-muscular to
daily sub-cutaneous administration of human growth hormone in growth hormone
deficient patients (literature reviewed in Hermanussen 1998). Though the
measurements were not done at strict weekly intervals, a significant catch-up growth
spurt was detected immediately after transfer of administration which lasted for
several weeks and was followed by an intermittent decrease of growth velocity.
Thereafter, growth velocity proceeded in a wave-like pattern, and stabilised on a
significantly higher level than before the transfer of application. Later studies of
short-term growth that were based on strict weekly series of lower leg measurements
have confirmed this pattern (Hermanussen 1998). It appeared logical to use short-
term growth as predictive criterion for successful growth promotion in growth
hormone therapy. Serial measurements of the lower leg during the first 10 weeks of
growth hormone administration provided useful information in 21 out of 24 children
on the growth response during the first year of treatment. Weekly measurement
regimens were used for optimising growth therapies (Hermanussen 1995), and we
investigated the possibility to synchronise the administration of growth hormone
with the spontaneous pattern of short-term growth (Hermanussen 1998).
Other authors used weekly measurements of the lower leg length for studying
growth promotion during treatments with growth hormone (Guzzaloni et al. 1997a,
Wales, Milner, 1989, Wit et al. 1987). But most of these studies lacked adequate
statistical treatment of the weekly measurement series. In view of the traditional
ideas of short-term growth being a linear process, growth was usually expressed in
terms of "linear growth rates", i.e. by length differences within intervals of 3 to 6
weeks. And since serial measurements at weekly intervals are tedious, many
investigators avoided the weekly series, and only obtained measurements once every
3 to 6 weeks, with no measurements in-between. In other words, many "short-term"
growth studies consist of subsequent 3 to 6-week length increments.
Ross et at. (1986) used four-week increments in children with Turner syndrome,
and compared mean leg length increment before and during treatment with growth
hormone, similar protocols were used by Gelato et al. (1985) who studied the effects
of pulsatile administration of growth hormone-releasing hormone in children with
growth hormone deficiency. Mortensen et at. (1991) published that lower leg growth
during the first three months of growth hormone treatment was significantly
associated with total height gain during the following nine months period, and when
comparing 4-week periods of growth hormone treatment in patients with Turner
Syndrome, Wales and Milner 1987b identified those children in whom growth
hormone treatment was not beneficial (literature reviewed in Hermanussen 1998).
When pre-treatment "linear growth rates" remained significantly smaller than "linear
growth rates" during treatment, it was assumed that growth stimulation had taken
place. Knemometry has also been applied in patients with constitutional growth
delay who were treated with oxandrolone (Guzzaloni et at. 1997b).
Since knemometry is not available for everybody, studies have started to replace
measurement accuracy by measurement frequency. In view of the brilliant
descriptions of growth by daily series of body height measurements, we inaugurated
328 CHAPTER 25
daily measurements of height for controlling height increments during growth

promoting therapies (Glock et at. 1999) with very promising results (Hermanussen,
unpublished).
Knemometry was also used for studying the effect of growth inhibition on short-
term growth. In 1981, van den Bosch et at. (1981) reported on low dose estrogens on
short-term growth in girls with tall stature. We further investigated this effects, and
found growth inhibition only in the leg whereas trunk growth appeared insensitive to
oestrogen treatment in tall girls. In 1987, we presented evidence for "negative
growth", with shrinkage of body stature and lower leg length in two patients with
anorexia nervosa (reviewed in Hermanussen 1998).
Inhibiting effects of glucocorticoids on short-term growth were described by
several authors, and reviewed by Wolthers (Heuck, Wolthers 1997, Wolthers 1996).
Visser et at. (1998) studied short-term growth in asthmatic children using fluticasone
proprionate, Wolthers and Pedersen (1990) investigated short-term growth of
asthmatic children during long-term treatment with prednisolone, and with
budesonide, and found a dose-related suppression of short-term growth. They also
investigated the effects of inhaled fluticasone propionate and beclomethasone
diproprionate, and short-term growth in children treated with oral antihistamine, and
found growth inhibition during treatments with depot and intranasal corticosteroids.
More recently, the group studied short-term growth and urine cortisol excretion in
children treated with corticosteroids (Agertoft and Pedersen 1997, 1999), and related
short-term growth and collagen turnover in children and adolescents treated with
corticosteroids (Heuck et at. 1998b, 2000, Wolthers et al. 1997), they studied growth
in atopic dermatitis treated with topical glucocorticoids (Heuck et at. 1998a), and the
growth suppressive effect of intra-articular glucocorticoids (Heuck et al. 1999).
Seidel et at. (1991) published the effect of corticosteroids in renal transplant patients
and presented some preliminary data on erythropoietin therapy in four patients
managed by regular haemodialysis, and in one girl with pre-terminal chronic renal
failure. We detected stunted growth during intermittent infectious illness, in growth
hormone deficient children during growth hormone therapy (Hermanussen et at.
1986). Several observations on short-term growth exist in patients who were treated
for acute lymphoblastic leukaemia (Ahmed et at. 1999, Crofton et at. 1999,2000).
10. SHORT-TERM GROWTH STUDIES IN NEONATES

Conventional knemometry is performed in children beyond an age of four years, for
technical reasons. Thus, short-term growth during early infancy was usually
investigated by standard anthropometric techniques. Since neonates grow very
rapidly, standard techniques appeared primarily satisfying. More accurate techniques
had to be developed for measurements at intervals of less than one week. First in
1991, Michaelsen published non-invasive lower leg length measurements in infants
using a small hand-held knemometer, with a technical error between 0.82 mm and
0.31 mm (Gibson et al. 1993, Michaelsen et al. 1991). This accuracy allowed
measurement intervals of 2 to 3 days. The technique showed non-linear incremental
patterns with episodes of growth rate acceleration, separated by intervals of reduced
growth velocity. Shrivastava et at. (2000) investigated 11 very low birth weight pre-
term infants with chronic lung disease, treated with dexamethasone, and found
significant decreases of lower leg length velocity compared to pre-treatment. Crofton
et al. (1996) examined the relationship of bone-specific alkaline phosphates, C-

terminal propeptide of type I collagen, N-terminal propeptide of type III procollagen,
C-terminal telopeptide of type I collagen, urinary pyridinoline and deoxypyridinoline,
with rates of gain in weight, length, and lower leg length in 25 pre-term infants, and
concluded that the markers of soft-tissue collagen formation, are good markers for
overall linear growth in pre-term infants. Cianfarani et al. (1998) studied serum
levels of growth hormone, IGFBP-l, IGF-l, IGFBP-3 and C-peptide in intrauterine
growth retarded new-borns, and demonstrated that baseline peptide levels did not
correlate with knee-heel length at 3 months of age. Knemometry was also used for
nutrition studies. Nicholl and Gamsu (1999) investigated the growth promoting
effect of a commercially prepared milk fortifier to human milk in very low birth
weight infants.
In all studies mentioned above, measurements were taken at weekly or more than
weekly intervals since the accuracy of the technique did not allow studies at shorter
intervals. We therefore, started to modify the hand-held knemometer used hitherto,
and introduced metallic holders for measurements in infants, as this was already
successfully applied in the animal studies. The holders are connected to an electronic
slide that discriminates intervals of 10 /-lm. Thus, the measurement error could be
reduced to 144 /-lm in neonates, and further reduced to 60 /-lm in pre-term infants as
they usually have less subcutaneous fat. The lower leg length measurements are
painless, and take place during breast-feeding, or while the infants are sleeping in
their beds or incubators (Hermanussen and Seele 1997, Kaempf et al. 1998, Keller et
al. 2000a).
Multiple studies are in progress or have recently been published documenting the
daily patterns of lower leg length increments in neonates and prematures, and relate
growth to illnesses and feeding regimes (Keller et at. 1998, 2000a,b,c). Keller et al.
(1999) were able to visualise the growth promoting effect of single transfusions of
erythrocyte concentrate. Particularly the latest investigations of infant and neonatal
growth have emphasised the practicability and usefulness of accurate short-term
growth measurements in various clinical situations.
Acknowledgements. This research was supported by Deutsche Gesellschaft fUr

Auxologie.
11. REFERENCES
Agertoft, L., and Pedersen, S., 1997, Short-term knemometry and urine cortisol excretion in children
treated with fluticasone propionate and budesonide: a dose response study. European Respiratory
Journal, 10, 1507-12.
Agertoft, L., and Pedersen, S., 1999, Short-term lower leg growth rate in children with rhinitis treated
with intranasal mometasone furoate and budesonide. Journal of Allergy and Clinical Immunology,
104, 948-52.
Ahmed, S.F., Wallace, W.H., Crofton, P.M., Wardhaugh, B., Magowa, R., and Kelnar, C.J., 1999,
Short-term changes in lower leg length in children treated for acute lymphoblastic leukaemia.
Journal of Pediatric Endocrinology & Metabolism, 12, 75-80.
Ahmed, S.F., Wallace, W.H.B., and Kelnar, C.J.H., 1995, Knemometry in childhood: a study to
compare the precision of two different techniques. Annals of Human Biology, 22, 247-252.
Ahmed, S.F., Wardhaugh, B.W., Duff, J., Wallace, W.H.B., and Kelnar, C.J.H., 1996, The relationship
between short-term changes in weight and lower leg length in children and young adults. Annals of
Human Biology, 23,159-162.
Ashizawa, K., and Kawabata, M., 1990, Daily measurements of the heights of two children from June
1984 to May 1985. Annals of Human Biology, 17,437-443.
330 CHAPTER 25
Boyd, E., 1980, Origins of the study of human growth. Eugene OR, University of Oregon Health
Science Foundation.
Camerer, W., 1880, Versuche tiber den Stoffwechsel, angestelit mit 5 Kindem im Alter von 2 bis 11
Jahren. Zeitschrift ftir Biologie, 16, 24-41.
Cianfarani, S., Germani, D., Rossi, P., Rossi, L., Germani, A., Ossicini, c., Zuppa, A., Argiro, G., Holly,
J.M., and Branca, F., 1998, Intrauterine growth retardation: evidence for the activation of the
insulin-like growth factor (IGF)-related growth-promoting machinery and the presence of a cation-
independent IGF binding protein-3 proteolytic activity by two months of life. Pediatric Research,
44,374-380.
Crofton, P.M., Ahmed, S.F., Wade, J.C., Elmlinger, M.W., Ranke M.B., Kelnar C.J., and Wallace W.,
1999, Effects of a third intensification block of chemotherapy on bone and collagen turnover,
insulin-like growth factor I, its binding proteins and short-term growth in children with acute
lymphoblastic leukaemia. European Journal of Cancer, 35, 960-967.
Crofton, P.M., Ahmed, S.F., Wade, J.C., Elmlinger, M.W., Ranke M.B., Kelnar, C.J., and Wallace, W.,
2000, Bone turnover and growth during and after continuing chemotherapy in children with acute
lymphoblastic leukemia. Pediatric Research, 48, 490-496.
Crofton, P.M., Shrivastava, A., Wade, J.C., Stephen, R., Kelnar, C.J.H., Lyon, A.J., and Mcintosh, N.,
1996, Bone and collagen markers in preterm infants: relationship with growth and bone mineral
content over the first 10 weeks of life. Pediatric Research, 46, 581-587.
Cronk, C.E., Stallings, V.A., Spender, Q.W., Ross, J.L., and Widdoes, H.D., 1989, Measurement of short
term growth with a new knee height measuring device (KHMD). American Journal of Human
Biology, 1,421-428.
Davies, H.A., Pickering, M., and Wales, J.K.H., 1996, A portable knemometer: a technique for
assessment of short-term growth. Annals of Human Biology, 23, 149-157.
Gelato, M.C., Ross, J.L., Malozowski, S., Pescovitz, O.H., Skerda, M., Cassoria, F., Loriaux, L., and
Merriam, G.R., 1985, Effects of pulsatile administration of growth hormone (GH)-releasing
hormone on short term linear growth in children with GH deficiency. Journal of Clinical
Endocrinology and Metabolism, 61, 444-450.
Gibson, A.T., Pearse, R.G., and Wales, J.K.H., 1993, Knemometry and the assessment of growth in
premature babies. Archives of Disease in Childhood, 69, 498-504.
Glock, M., Hermanussen, M., Keller, E., and Hartmann, K.K.P., 1999, Gulliver G-100 - a new device
to evaluate daily growth measurement in comparison with Harpenden stadiometer. Hormone
Research, 52, 287-290.
Guzzaloni, G., Grugni, G., Moro, D., Minocci, A., and Morabito, F., 1997a, Impiego della knemometria
nella valutazione degli effetti a breve termine della terapia con GH in bambini con diversi disordini
accrescitivi. Minerva Pediatrica, 49, 139-145.
Guzzaloni, G., Grugni, G., Moro, D., Minocci, A., and Morabito, F., 1997b, Usefullness of knemometry
in the assessment of oxandrolone effects and role of infections in short-term growth within
constitutional growth delay. Acta Medica Auxologica, 29, 21-27.
Heinrichs, c., Munson, P.J., Counts, D.R., Cutler, G.B., and Baron, J., 1995, Patterns of human growth.
Science, 268, 442-444.
Hermanussen, M., 1995, Growth promotion by steroid substitution in short stature treated with GH and
LHRH analogue. Acta Paediatrica, 84, 1207-1210.
Hermanussen, M., 1998, The analysis of short-term growth. Hormone Research, 49, 53-64.
Hermanussen, M., and Seele, K., 1997, Mini-knemometry: an accurate technique for lower leg length
measurements in early childhood. Annals of Human Biology, 24, 307-313.
Hermanussen, M., and Sippell, W.G., 1985, Knemometric monitoring of early effects of human growth
hormone on leg growth in children with growth hormone deficiency, Lancet, I, 1069-1071.
Hermanussen, M., Bugiel, S., Aronson, S., and Moell, C., 1992, A non-invasive technique for the
accurate measurement of leg length in animals. Growth, Development & Aging, 56,129-140.
Hermanussen, M., Geiger-Benoit, K., and Sippell, W.G., 1986, GH deficient children receiving GH
replacement do not grow during intermittent infectious illness. Acta Paediatrica Scandinavica, 75,
601-604.
Hermanussen, M., Geiger-Benoit, K., Burmeister, J., and Sippell, W.G., 1988a, Knemometry in
childhood: accuracy and standardization of a new technique of lower leg length measurement.
Hermanussen, M., Geiger-Benoit, K., Burmeister, J., and Sippell, W.G., 1988b, Periodical changes of
short term growth velocity ("mini growth spurts") in human growth. Annals of Hurnan Biology, 15,
103-109.
Hermanussen, M., Rol de Lama, M., Burmeister, J., and Fernandez-Tresguerres, J.A., 1995, Mikro-
knemometry: an accurate technique of growth measurement in rats. Physiology nd Behavior, 58,
347-352.
Hermanussen, M., Rol de Lama, M., Perez Romero, A., Ariznaverreta, C., Burmeister, J., and
Tresguerres, J.A.F., 1996, Differential catch-up in body weight and bone growth after short term
starvation in rats. Growth Regulation, 6, 230-237.
Hermanussen, M., Rol de Lama, M., Tresguerres, J.A.F., Grasedyck, L., and Burmeister, J., 1998, Short
term growth: evidence for chaotic series of mini growth spurts in rat growth. Physiology and
Behavior, 64, 7-13.
Heuck, C., and Wolthers, 0.0., 1997, Calculation of knemometric growth rates in group studies of
children treated with exogenous glucocorticoids. Annals of Human Biology, 24, 411-418.
Heuck, c., Heickendorff, L., and Wolthers, 0.0., 2000, A randomised controlled trial of short term
growth and collagen turnover in asthmatics treated with inhaled formoterol and budesonide.
Archives of Disease in Childhood, 83, 334-339.
Heuck, c., Ternowitz, T., Herlin, T., and Wolthers, 0.0., 1998a, Knemometry in children with atopic
dermatitis treated with topical glucocorticoids. Pediatric Dermatology, 15,7-11.
Heuck, C., Wolthers, 0.0., and Herlin, T., 1999, growth-suppressive effect of intra-articular
glucocorticoids detected by knemometry. Hormone Research, 52, 95-Q6.
Heuck, c., Wolthers, 0.0., Kollerup, G., Hansen, M., and Teisner, B., 1998b, Adverse effects of
inhaled budesonide (880f.lg) on growth and collagen turnover in children with asthma: a double-
blind comparison of once-daily versus twice-daily administration. Journal of Pediatrics, 133, 608-
612.
Hulanicka, B., Gronkiewicz, L., and Koziel, S., 1999, Wzrastanie dzieci badanie knemometryczne.
Monografie. Zakladu Antropologii Polskiej Akademii Nauk. Wroclaw.
Kaempf, D.E., Pfliiger, M.S., Thiele, A.M., Hermanussen, M., and Linderkamp, 0., 1998, Influence of
nutrition on growth in premature infants: assessment by knemometry. Annals of Human Biology, 25,
127-136.
Keller, A., Hermanussen, M., Vogtmann, C., Kiess, W., and Keller, E., 1999, Effect of erythrocyte
transfusion on longitudinal bone growth of premature infants assessed by mini-knemometry.
European Journal of Pediatrics, 158,871-872.
Keller, A., Hermanussen, M., Vogtmann, Ch., Burmeister, 1., and Keller, E., 2000a, Analysis of short-
term growth of premature infants during the first days of life assessed by mini-knemometry. Acta
Medica Auxologicaogica, 32, 40.
Keller, A., Keller, E., Burmeister, J., and Hermanussen, M., 1998, Daily knemometric measurements
can visualize the clinical situation of critically ill infants. Hormone Research 50, suppl 3, 37.
Keller, A., Vogtmann, Ch., Hermanussen, M., and Keller, E., 2000b, Short-term growth during the first
month of life of SGA and AGA infants. In: II. Aschauer Seminar: The impact of environmental
factors on individual development with special reference to nutrition and obesity. Monatschrift fiir
Kinderheilkunde, 148, 810-815.
Keller, A., Vogtmann, Ch., Keller, E., Hermanussen, M., and Kiess, W., 2000c, Differences of short-
term growth in relation to varying doses of dexamethasone administration in premature infants.
Hormone Research 53 (suppl 2), 29
Klein, K.O., Munson, P.J., Bacher, J.D., Cutler, G.B., and Baron, J., 1994, Linear growth in the rabbit is
continuous, not saltatory. Endocrinology, 134, 1317-1320.
Kobayashi, M., and Togo, M., 1993, Twice-daily measurements of stature and body weight in two
children and one adult. American Journal of Human Biology, 5, 193-201.
Lampl, M., and Johnson, M.L., 2000, Distribution methods and analysis of nonlinear longitudinal data.
Methods in Enzymology, 321,182-195.
Lampl, M., Ashizawa, K., Kawabata, M., and Johnson, M.L., 1998, An example of variation and
pattern in saltation and stasis growth dynamics. Annals of Human Biology, 25, 203-219.
Lampl, M., Veldhuis, J.D., and Johnson, M.L., 1992, Saltation and stasis: a model of human growth.
Science, 258, 801-803.
Ledford, A.W., and Cole, T.J., 1998, Mathematical models of growth in stature throughout childhood.
MaIling-Hansen, R., 1886, Periods in the growth of children and in the heat of the sun. Copenhagen,
Vilhelm Tryde; cited by: Michaelsen KM, 1994, Short-term measurements of linear growth using
knemometry. Journal of Pediatric Endocrinology, 7, 147-154.
Marshall, W.A., 1971, Evaluation of growth rate in height over periods of less than one year. Archives
of Disease in Childhood, 46, 414-420.
Michaelsen, K.F., Skov, L., Badsberg, J.H., and Jorgensen, M., 1991, Short-term measurement of linear
growth in preterm infants: validation of a hand-held knemometer. Pediatric Research, 30, 464-468.
Milani, S., 2000, Kinetic models for normal and impaired growth. Annals of Human Biology, 27,1-18.
Mortensen, H.B., Main, K., Michaelsen, K.F., Kastrup, K.W., Jorgensen, J.T., and Skakkebaek, N.E.,
1991, Predicting and monitoring of growth in children with short stature during the first year of
growth treatment. Acta Paediatrica Scandinavica, 80, 1150-1157.
Nicholl, R.M., and Gamsu, H.R., 1999, Changes in growth and metabolism in very low birth weight
infants fed with fortified breast milk. Acta Paediatrica, 88, 1056-1O6\.
Rol de Lama, M.A., Perez-Romero, A., Hermanussen, M., Ariznavarreta, R.C., and Tresguerres,
J.A.F., 1998, Sexual dimorphism in growth as measured by microknemometry: Different responses
to GH deficiency and exogenous GH administration. Neuroendocrinology, 68, 210-219.
332 CHAPTER 25
Rol de Lama, M.A., Perez-Romero, A., Tresguerres, 1.A.F., Hermanussen, M., and Ariznavarreta,
R.C., 2000, Recombinant human growth hormone, rhGH, enhances tibial growth in peripubertal
female rats but not in males. European lournal of Endocrinology, 142,517-523.
Ross, 1.L., Long, L.M., Skerda, M., Cassorla, F.G., Loriaux, D.L., and Cutler, G.B., 1986, Growth
response relationship between growth hormone dose and short term growth in patients with Turner's
syndrome. lournal of Clinical Endocrinology and Metabolism, 63, 1028-1030.
Seidel, Chr., Schafer, F., Walther, U., and Schlirer, K., 1991, The application of knemometry in renal
disease: preliminary observations. Pediatric Nephrology, 5, 467-471.
Shrivastava, A., Lyon, A., and Mcintosh, N., 2000, The effect of dexamethasone on growth. Mineral
balance and bone mineralisation in preterm infants with chronic lung disease. European 10urnal of
Tanner, 1.M., 1978, Foetus into man. 1st ed. London, Open Books Publishing Ltd, pp 7-8.
Togo, M., and Togo, T., 1982, Time-series analysis of stature and body weight in five siblings. Annals
of Human Biology, 9, 425-440.
Valk, I.M., 1971, Accurate measurement of the length of the ulna and its application in growth
measurements. Growth, 35, 297-310.
Valk, I.M., Langhout Chabloz, A.M.E., and van Gilst, W., 1983b, Intradaily variation of the human
lower leg length and short term growth: a longitudinal study in fourteen children. Growth, 47, 397-
402.
Valk, I.M., Langhout Chabloz, A.M.E., Smals, A.G.H., Kloppenborg, P.W.C., Cassorla, F.G., and
Schutte, E.A.S.T., 1983a, Accurate measurement of the lower leg length and the ulnar length and
its application in short term growth measurements. Growth, 47, 53-66.
Van den Bosch, 1.S.G., Smals, A.G.H., Pieters, G.F.F., Valk, I.M., and Kloppenborg, P.W.c., 1981, The
effect of low dose oestrogens on short term growth and concomitant biochemical phenomena in
girls with tall stature. Acta Endocrinologica, 98, 156.
Visser, MJ., van Aalderen W.M., Elliott B.M., Odink RJ., and Brand P.L., 1998, Short-term growth in
asthmatic children using fluticasone propionate. Chest, 113,584-586.
Wales, 1.K.H., and Gibson, A.T., 1994, Short-term growth - rhythms, chaos or noise? Archives of
Disease in Childhood 71,84-89.
Wales, 1.K.H., and Milner, R.D.G., 1987a, Knemometry in the assessment of linear growth. Archives of
Wales, J.K.H., and Milner, R.D.G., 1987b, Knemometry as a predictor of response to somatrem in
Turner's syndrome. Acta Paediatrica Scandinavica Supp!., 337, 37-39.
Wales, 1.K.H., and Milner, R.D.G., 1989, Changes in leg length and height during treatment with
somatropin. Archives of Disease in Childhood, 64,1708-1712.
Wit, 1.M., van Kalsbeek, EJ,. von Wijk-Hoek I.M., and Leppink, G.1., 1987, Assessment of the
usefulness of weekly knemometric measurements in growth studies. Acta Paediatrica
Scandinavica, 76, 974-980.
Wolthers, O.D., 1996, Long-, intermediate- and short-term growth studies in asthmatic children treated
with inhaled glucocorticosteroids. European Respiratory lournal, 9, 821-827.
Wolthers, O.D., and Pedersen, S., 1990, Short term linear growth in asthmatic children during treatment
with prednisolone. British Medical Journal, 301, 145-148.
Wolthers, O.D., Hansen, M., luul, A., Nielsen, H.K., and Pedersen, S., 1997, Knemometry, urine
cortisol excretion, and measures of the insulin-like growth factor axis and collagen turnover in
children treated with inhaled glucocorticosteroids. Pediatric Research 41, 44-50.
CHAPTER 26
RELATIONS BETWEEN HEAD GROWTH AND

PSYCHOMOTOR DEVELOPMENT IN LOW-BIRTH-
WEIGHT CHILDREN
W. FURMAGA-JABLONSKAt, H. CHRZASTEK-SPRUCHt, M.
KOZLOWSKAt, A. ORZECHOWSKI:j:
tDepartment of Developmental Paediatrics, Medical University School of Lublin, Poland,

:f:Department of Biophysics, Medical University School of Lublin, Poland
1. INTRODUCTION
Child development is a multi-directional process taking place simultaneously at the

physical and psychical level. These processes are coexistent permanently and leading
to a child's psycho-physical maturity. Especially during the first two years of life
body size seems to strongly correlate with psychomotor activity (Brandt 1986,
Livshits et at. 1992, Lindhal 1987, Ounstedt et at. 1984, Piekkala et at. 1988).
Mutual relationships between the physical and psychomotoric development in the
early childhood are extremely interesting as they allow to make a prognosis of the
level of psychomotoric development, based on the routine anthropometric tests, and
thus they allow for early stimulation of children being at hazard of psychomotoric
deficiencies in the further development. The most frequently quoted characteristics of
physical development, which may further influence psychomotoric development of a
child, are birth parameters (Furmaga-Jablonska et at. 1998, Livshits et at. 1992,
1993, Piekkala et at. 1988, Simon et at. 1993, Sheth et at. 1995). Livshits et at.
(1993) confirmed that a child's physical characteristics and maturity, which are
defined by the gestational age, body weight and height as well as the head
circumference at birth, may be important for predicting its future psychomotoric
development. Those authors underline that the head circumference may be of
statistical significance for predicting psychomotor development (Livshits et al.
1992).
The authors try to establish the relations between physical growth and
psychomotor development in children born with low-birth-weight (LBW) on the
basis of longitudinal observations. Correlations between head growth and
psychomotor developmental quotients in three groups of low-birth-weight children
were studied. Existing relationships allow to predict a child's future psychomotor
development on the basis of longitudinal observations of head growth dynamics.
333
334 CHAPTER 26

Relationships between head circumference, head circumference velocity and
psychomotor development in 283 low-birth-weight (LBW) children, born in Lublin
(Poland), were longitudinally analysed. The subjects were divided according to
gestational age and centile position of birth weight and thus three groups were
created: group I - 181 appropriate-for-gestational age (AGA) preterms, group II - 70
small-for-date (S-f-D) newborns, group III - 32 small-for-gestational age (SGA)
preterms.
Circumferences of the head (in cm) and head circumference velocities (in
cm/month) were determined using standard anthropometric measurements performed
in each child at birth and in 1st, 2nd, 3rd, 6th, 9th, 12th, 18th and 24th month of the
chronological age. On paediatric examinations neither central nervous system
pathology nor signs of rachitis were found in any of the children.
An assessment of psychomotor development was done between the third and
twenty fourth month of life of each child during subsequent follow-ups at the
Growth Clinic. The level of general developmental quotient of each child (DQ) was
estimated by means of Brunet-Lezine Inventory (1951) in consecutive months of
observation. The correct DQ values range from 90 to 110. Statistical analysis
(analysis of variance (ANOVA), Pearson's correlation coefficients) was performed
after assessment of normality of the traits, using the computer program, Statistics
95 (Moczko et al. 1998).
3. RESULTS
The examined three groups of LBW children differed as for their gestational age. The
mean gestational age was 33.5 weeks for AGA preterms (group I), 35.8 weeks for S-
f-D newborns (group II), and 35.0 weeks for SGA preterms (group III). The
differences found for gestational age between S-f-D infants and AGA preterms as well
as for S-f-D infants and SGA preterms were highly statistically significant (ANOVA
and Duncan test, p < 0.001).
The biggest head circumference at birth and in the first two months of life was
revealed by S-f-D infants. After birth, mean differences between AGA preterms
(group I) and S-f-D infants (group II) amounted 1.5 cm, and between S-f-D infants
(group II) and SGA preterms (group III) - 1 cm. The differences were highly
significant for group I and II (p < 0.001) and for group II and III (p < 0.05),
respectively. Statistical significance was still observed in the first and second month
after birth, however, in later months of observation the differences for head
circumference between the groups were decreasing and were not statistically
significant (Figure 1).
In the subsequent months of the follow-up, a change of the rate of head
circumference velocity in individual groups of LBW children occurred. Between 3
months and 2 years (chronological age), the largest velocity of head circumference
was observed in AGA preterms (group I) while S-f-D infants (group II) tended to
have the lowest velocity. Between the 6th and 9th month, statistically significant
HEAD GROW11I AND PSYCHOMOTOR DEVELOPMENT 335
differences occurred for mean values of head circumference velocity between groups I
and II as well as between I and III (p < 0.05) (Table 1).
Table 1. Head circumference velocity (cmlmonth) in three groups of LBW children
AGA preterms S-f-D SGA preterms

months (group!) (group II) (groups III)
follow-up mean SD mean SD mean SD
1/0 3.25 1.13 3.33 1.83 3.45 0.89
2/1 2.19 0.68 1.90 0.34 2.25 0.58
3/2 1.72 0.51 1.61 0.42 1.68 0.29
6/3 1.32 0.27 1.20 0.33 1.28 0.20
9/6 0.75 *t 0.23 0.60 * 0.22 0.61t 0.14
12/9 0.46 0.15 0.44 0.21 0.49 0.12
18/12 0.24 0.11 0.23 0.07 0.20 0.06
24/18 0.19 0.10 0.15 0.09 0.19 0.09
t* p < 0.05
Development Quotient (DQ) in the three groups of LBW children differed from
one another in subsequent months of the investigation (Table 2). On the first
psychological examination, i.e. at 3 months, the lowest developmental level was
presented by AGA preterms, obtaining DQ = 84, which is below the 'normal' range.
Slightly higher mean score, DQ = 89, but still below the cut-off value of 90, was
found in the SGA preterms. Only S-f-D infants had, at 3 months, a DQ = 96, which
is within the 'normal' range. AGA preterms reached the developmental standard level
of psychomotor development of DQ = 90 by the age of 6 months month, while
mean DQ values in S-f-D newborns (DQ = 98) and in SGA preterms (DQ = 96)
exceeded this value. Differences between general developmental quotients in the 3rd
and 6th month of the longitudinal study for group I - AGA preterms and II -S-f-D
infants were statistically significant (in the 3rd month: p < 0.001, in the 6th month
p < 0.05).
Table 2. Mean values of developmental quotient (DQ) in the three groups of LBW children
months of AGA preterms S-f-D SGA preterms

follow-up (group!) (group II) (groups III)
3 84 ** 96 ** 89
6 90 * 98 * 96
9 93 98 98
12 95 101 95
18 97 103 100
24 103 105 104
* p < 0.05; ** p < 0.01
336 CHAPTER 26
The prevailing highest mean, general developmental quotients in S-f-D infants

and the lowest in AGA preterms were observed in consecutive months of the follow-
up. Thus, the group of S-f-D infants, characterised by the highest gestational age
(38.8 weeks), reached the highest mean developmental quotients compared with the
other groups. The lowest DQ values in the 2 year period of observation were
presented by AGA preterms whose mean gestational age (35.0 weeks) was the lowest
compared with S-f-D infants and SGA preterms.
52
48
E
u 44
~.
~ .: .:
~ 40
L*~iLI ~ L~p,,"'m, ll
§
~
'u
1"l
r "', ~ ~
36
<l)
::r: ~ --0-- ~
S-f-D infants
32 ~ ~ ...... j.........
mmmmrmmmmrmm : -------(r- SGA preterms .. ji.
28 I ;
-3 o 3 6 9 12 15 18 21 24
Months offollow-up
Figure 1. Mean head circumference the three groups of LBW children (* significant
diffe rences)
Differences between developmental quotients for individual groups were

decreasing and starting from the 9th month of the chronological age of the children
under study were no longer statistically significant. With the increasing age of the
examined children, a tendency to level the quotients of psychomotor development
between different groups was observed. Although S-f-D infants (group II) revealed in
the observed period the highest mean developmental quotient, compared with the
other groups, the progress of psychomotor development in S-f-D infants was the
lowest, which is evidenced by a merely 9 point difference between mean
developmental quotients obtained in the initial (3rd month) and final (24 month)
stage of the longitudinal studies. Between the 6th and 9th month of life, no advance
in the psychomotor development was observed in S-f-D infants, and the examined
infants obtained the same developmental quotients (DQ = 98) in consecutive months.
In this very period, the lowest head circumference velocity was also observed in S-f-
D infants compared with AGA preterms and SGA preterms, what was confirmed by
analysis of variance (p < 0.05) (Table 1).
HEAD GROWTH AND PSYCHOMOTOR DEVELOPMENT 337
The greatest acceleration of the psychomotor development was found in AGA

preterms, which is shown by mean developmental quotients obtained in subsequent
months of the longitudinal study and the difference between the first examination in
the 3rd month of life and the last one in the 24th month for children from these
groups amounted 19 points. Considerable progress in psychomotor development was
also characteristic of SGA preterms obtained in the 24th month of life the mean
development quotient by 15 points higher than in the initial examination in the 3rd
month.
Table 3. Correlation coefficients between head circumference (HC) and general

developmental coefficient (DQ) in the three groups of LBW (all correlations shown in this
Table are significant: p < 0.05)
AGA preterms (group I)

months
follow-up 3 6 9 12 18 24
0 0.32 0.57 0.35 0.32 0.58
1 0.29
2 0.41 0.41 0.36 0.45 0.51 0.51
3 0.41 0.35 0.36 0.40
6 0.33 0.39 0.44 0.52 0.44
9 0.29 0.42 0.29 0.37
12 0.38 0.34 0.38
18 0.35
24 0.28
S-f-D infants (group II)
months
follow-up 3 6 9 12 18 24
1 0.52 0.59 0.56
2 0.44 0.48 0.48
3 0.45 0.50
6 0.49
9 0.41
SGA preterms (group III)
months
follow-up 3 6 9 12 18 24
0 0.48
1 0.79
A similar tendency of changes in the three LBW groups was also observed in
anthropometric examinations. AGA preterms, despite the smallest mean head
circumferences at birth, were characterised by the highest growth rate of this trait
during the 24 month observation, while the smallest rate was found in S-f-D infants
338 CHAPTER 26
(Figure 1, Table 1). The above observations suggest correlations between physical
growth and psychomotor development which made the authors calculate Pearson's
correlation coefficient between head circumferences in individual months, head
circumference velocity expressed in cm/month and mean developmental quotients
obtained in the consecutive months of follow- up.
The correlations between head circumferences and psychomotor development for
AGA preterms are shown in Table 3. The highest correlations were found between
head circumference at birth and the general developmental quotient of AGA preterms
in the 6th (r = 0.57, p < 0.05) and in the 18th (r = 0.58, p < 0.05) month of their
lives. In AGA preterms, general developmental quotients in the 24th month of life
most strongly correlated with head circumference in the 2nd month of the
chronological age (r = 0.5, p < 0.05). Slightly lower correlation coefficients (r =
0.28 - 0.44), but still statistically significant (p < 0.05) were observed in AGA
preterms between the majority of physical and psychomotor parameters.
In the S-f-D infants, no significant correlations were observed between
psychomotor development in the 6th and 12th month of life and parameters of
physical growth (Table 3). However, general developmental quotient of S-f-D
children in the 3rd, 18th and 24th month of the chronological age were directly
proportional to head circumference in the first month of life (r = 0.52 - 0.59, p <
0.05). Correlations in S-f-D children occur for head circumference measurements
obtained in the 2nd month and DQ in the 9th, 18th and 24th month (r = 0.44 - 0.48,
p < 0.05). Psychomotor development in the 9th month also strongly correlates with
head circumferences of S-f-D children in the 2nd, 3rd and 6th month (r = 0.44 - 0.50,
p < 0.05).
The least significant correlations between head circumferences and psychomotor
development were found in group III of SGA preterms (Table 3). Statistically
significant correlations occurred in SGA preterms only between head circumference
in newborns and in the 1st month, and DQ in the 3rd and 6th month respectively (r =
0.48, r = 0.79, p < 0.05).
The velocity rate of head circumference did not correlate with DQ in the AGA
preterm group as well as the S-f-D infants. However in SGA preterms, we observed
highly significant correlation between velocity of the head circumference in the first
month and DQ in the 6th month of chronological age (r = 0.94, p < 0.05).
4. DISCUSSION
Several investigations of the literature have confirmed a close relationships between
the increase of head circumference and the brain development in normally developed
children during the first two years of their life (Simon et at. 1993, Brandt 1978,
Mauser 1984, Dobbing 1974). Based on the literature, we can assume that the
biggest development of brain starts in the mid-pregnancy and reaches the extreme
values in the 2-3rd year of life (Brandt 1978, Mauser 1984, Dobbing 1970, Dobbing
and Sands 1978, Winnick and Rosso 1969). This process is parallel to the biggest
increase of head circumference. In those periods of life the increase of brain mass and
the head circumference are connected with the development and fast progress of
psychomotoric functions in the youngest children. According to Henneberg et at.
(1984) the intelligence quotient is proportional to the efficiency at which the brain
processes and stores the information, and the bigger number of nervous elements is
HEAD GROWTH AND PSYCHOMOTOR DEVELOPMENT 339
placed in it, the bigger efficiency it has. The above considerations concern AGA full-
term children who present a regular biological development at the birth and whose
further physical growth and psychomotoric development is realised according to the
definite standards (Kurniewicz-Witczakowa 1980). In the available references no
attempts were made to define the relationships between the physical and
psychomotoric development in the group of children born with low-birth-weight.
However the investigations carried out in the world confirmed that the physical
growth and psychomotoric development between the LBW children and the AGA full
term-children differs substantially (Chrzastek-Spruch et al. 1998, Gajewska et al.
1995, Kaliszewska-Drozdowska 1989).
Earlier observations indicated that the existing differences in the dynamics of head
growth in the three groups of children, born with a low-birth-weight, influence the
different advancement of this feature in the second year of life (Furmaga-Jablonska et
al. 1997, Furmaga-Jablonska et al. 1999). Therefore, we made an attempt to define
possible relationships between the head circumference and psychomotoric
development in AGA preterms, S-f-D infants and SGA preterms, in view of
predicting the psychomotoric development in children on the basis of longitudinal
observations of head circumference development.
When we consider the relationships between physical and psychomotoric
development, the biggest relationships were confirmed between those two processes
in the AGA preterms. Statistically significant correlations were confirmed between
the head circumference at birth and the psychomotoric development in the subsequent
months of investigations starting from the 3rd till the 18th month of life. The
psychomotoric development (defined by general developmental quotient DQ) of AGA
preterms in the second year of life is also directly proportional to their head
circumferences in the 2nd, 3rd, 6th, 9th and 12th month of observation; but the
correlation coefficient decreases with the age of the child. Fewer correlations were
confirmed between the head circumference and psychomotoric development in the
group of S-f-D infants in the subsequent months of life. However the measurements
of head circumference performed in the 1st and 2nd month of chronological age are
diagnostically significant for the level of psychomotoric development in this group
in the second year of life.
In SGA preterms only measurements of head circumference at birth and in the
first month of life are statistically significant for prediction of their psychomotor
development in the 3rd and 6th month. And thus the newborns with double
pathology, i.e. pre-maturity and intrauterine growth retardation (IUGR) are
characterised by dissonance of further physical growth and psychical development.
However only in SGA-preterms a very direct correlation was confirmed between the
head circumference in the first month of life and the developmental quotient levels
(DQ) in the 6th month of life.
Considering the relationships between head circumference and psychomotoric
development in early childhood which are indicated in this paper, it seems that the
measurements of head circumference during developmental period may be a potential
screening test for evaluating the psychomotoric disorders in the group of low-birth-
weight children.
340 CHAPTER 26
5. REFERENCES
Brandt, I., 1978, Growth dynamics of low-birth-weight infants with emphasis on the perinatal period. In
Human growth, edited by F. Falkner and J.M. Tanner (New York: Plenum Press 2) p. 415.
Brandt, I., 1986, Patterns of early neurological development. In Human growth, edited by F. Falkner
and J.M. Tanner (New York: Plenum Press 2) p. 469.
Brunet, 0., Lezine, 1.,1951, Le developpement psychologique de la premiere enfance. Paris.
Chrzastek-Spruch, H., Jablonska, R, Furmaga-Jablonska, W., Kozlowska, M., Billewicz-Kraczkowska,
A., Hauspie, R, Orzechowski, A., 1998, Rozw6j psychomotoryczny wczesniak6w - badania
dlugofalowe. In Skad idziemy, kim jestesmy, dokad zmierzamy, edited by D. KornafeI (Wroc1aw:
Polskie Towarzystwo Antropologiczne) p. 33.
Dobbing, J., 1970, Undernutrition and the developing brain. American Journal of Diseases of Children,
120,411-415.
Dobbing, J., 1974, The later development of the brain and its vulnerability, In Scientific Foundation of
Paediatrics, edited by J.A. Davies and J. Dobbing, (London: Heinemann Medical Books) p. 565.
Dobbing, J., Sands, J., 1978, Head circumference, biparietal diameter and brain growth in fetal and
postnatal life. Early Human Development, 2, 81-87.
Furmaga-Jablonska, W., Chrzastek-Spruch, H., Kozlowska, M., 1999, Longitudinal analysis of head
growth in children born with low birth weight. Perspectives in Human Biology, edited by M.
Henneberg, Centre for Human Biology 4, 23-32.
Furmaga-Jablonska, W., Chrzastek-Spruch, H., Kozlowska, M., Jablonska, R, Orzechowski, A., 1998,
Rozw6j psychomotoryczny wczesniak6w w drugim roku zycia. In Czlowiek wczoraj, dzis, jutro,
edited by A. Gasiorowski, 1. Gurba and W. Kozak-Zychman, (Lublin: Wydawnictwo UMCS)
p.181.
Furmaga-lablonska, W., Kozlowska, M., Kulik-Rechberger, B., 1997, Dynamika rozwoju obwodu
glowy w trzech grupach dzieci urodzonych z mala masa ciala. Pediatria Polska, LXXII, 12, 1127-
1136.
Gajewska, E., Fuchs, B., Paluszynska, D., 1995, Noworodek 0 malej masie ciala, problem nadal
aktualny. (Warszawa: Centrum Zdrowia Dziecka)
Henneberg, M., Budnik, A., Pezacka, M., Puch, A., 1984, Wielkosc glowy a inteligencja:
wsp61zaleznosci wewnatrzgatunkowe u czlowieka, Przeglad Antropologiczny, 50, 299-313.
Kaliszewska-Drozdowska, M.D., 1989, Odrebnosci rozwojowe roznych grup niemowlat - propozycje
metodyczne oceny ich rozowju fizycznego., Pediatria Polska, 64,479-481.
Kurniewicz-Witczakowa, R, 1980, Rozw6j somatyczny, In Diagnostyka rozwoju dzieci i mlodziezy,
edited by 1., Kopczynska-Sikorska (Warszawa: PZWL) p.l20.
Lindhal, E., 1987, Motor performance of neonatal risk and non risk children at early school age, Acta
Paediatrica Scandinavica, 76, 809-817.
Livshits, G., Cohen, Z., Ortemski. I., 1992, Interrelations between early child development, gestational
age and birth weight, International Journal of Anthropology 2, 41-51.
Livshits, G., Cohen, Z., Ortemski, I., 1993, Relationship between physical growth and motor
development in infancy and early childhood: Multivariate analysis. American 10urnal of Human
Biology, 5, 481-489.
Mauser, 1.1., 1984, Growth in the high - risk infant, Clinics in Perinatology, 2,19-40.
Moczko, J.A., Breborowicz, G.H., and Tadeusiewicz, R, 1998, Statystyka w badaniach medycznych
edited by Springer PWN, Warszawa.
Ounstedt, M., Moar, V.A., Scott, A., 1984, Association between size and development at four years
among children who were small-for dates and large-for dates at birth, Early Human Development,
9,259-268.
Piekkala, P., Kero, P., Sillanpaa, M., 1988, The developmental profile and outcome of 325 unselected
preterm infants up to two years of age., Neuropediatrics. 19, 33-40.
Sheth, RD., Mullett, M.D., Bodensteiner, J.B., 1995, Longitudinal head growth in developmentally
normal preterm infants. Archives of Pediatrics and Adolescent Medicine 149, 1358-1361.
Simon, N.P., Brady, N.R, Stafford, RL., 1993, Catch-up head growth and motor performance in very-
low-birthweight infants. Clinical Pediatrics, 7, 405-411.
Winnick, M., Rosso, P., 1969, The effect of severe early malnutrition on cellular growth of human
brain, Pediatric Research, 3, 181-184.
CHAPTER 27
GROWTH IN LENGTH AND WEIGHT OF

THALASSEMIC CHILDREN IN WEST BENGAL, INDIA:
A MIXED-LONGITUDINAL STUDY
J. BANERJEEt, s. SAHAt, I. DASGUPTAt, D.K. BHATTACHARYYAt,

P. DASGUPTA,-r
t Thalassemia Unit, Vivekananda Institute of Medical Sciences, Calcutta, India,

IApplied Statistics Unit, Indian Statistical Institute, Calcutta, India, 'liAnthropology
and Human Genetics Unit, Indian Statistical Institute, Calcutta, India
1. INTRODUCTION
In India thalassemia trait varies between 1 and 17 percent with an average of 3.3
percent in normal population and Hb-E disease is exclusively present in the
population of West Bengal and Assam (Chatterjee 1965, Talukder and Sharma
1994). It co-exists with beta-thalassemia in West Bengal producing serious public
health problems. Growth retardation occurs in homozygous beta-thalassemia at
about the age of 8-1 0 years and many affected children attain a very short final height
(Weatherall and Clegg 1981). Various auxological studies indicate that stunted
growth and delayed puberty is frequent in transfusion-dependent thalassemics
(Borgna-Pignatti et al. 1985, Perignon et al. 1993, Yeslipek et al. 1993; Saka et al.
1995). In India, where most of the thalassemic children are under-transfused and do
not get adequate chelation therapy (Verma et al. 1992), the age of onset of growth
retardation may not be the same as that of the developed countries. In the present
communication an attempt has been made to report growth and growth increments
in length/height and weight of thalassemic Bengali children aged 1-6 years in
relation to the non-thalassemic control.

Children referred to a blood transfusion Centre in Calcutta and the thalassemia clinic
of the Vivekananda Institute of Medical Sciences, Calcutta, were the subjects of the
present study. A cohort of 126 children (81 males and 45 females) with homozygous
beta-thalassemia and E-beta thalassemia, and having different amounts of blood
transfusion (between 1 and 72 units) and 71 non-thalassemic controls (41 males and
30 females), between 1 and 6 years of age, constituted the study sample (Table 1).
341
342 CHAPTER 27
Forty-five percent of the patients participating in the study came from various
districts of West Bengal whereas 42% were residents of the city of Calcutta. The
patients were predominantly low caste Hindus (75%). Twenty-three percent parents
of the affected families were recorded as illiterate while 15% possess higher
education. Fifty-seven percent of the patients were found to represent the families
from service holders. However with respect to the per capita annual income level
(Rs. 9,000) economic status of the patients appeared to be low with average family
size of 5.2. Seventy-one percent of the children below six years of age showed
evidences of malnutrition in varying degrees whereas the rest (29%) appeared to be
normal. Mortality rate among the thalassemics were 12.12%.
Normal children were examined by the expert paediatricians and they were non-
thalassemics without having any genetic disorder. Among them about 39% stayed
in the city of Calcutta while the rest were from various districts of the state of West
Bengal. Eighty-four percent of the them comprised the higher castes like Brahmin,
Vaidya and Kayastha while the rest were affiliated to low or scheduled castes. Five
percent of the mothers were illiterate while 58% had education up to the school level
and 37% had education up to the college level. With respect to the occupation of
father only 16% were engaged in Agriculture, 49% were service holders, 35% were
involved in small scale business. The average family size was 3.5 with the per
capita annual income ofRs. 12,510.
Table 1. Frequency of measurements in the mixed-longitudinal study by sex and

variable.
n Nbr. of occasions
2 3 4 5 6
thalassemic height (boys) 81 34 17 10 10 1 9
thalassemic height (girls) 45 9 15 8 5 2 6
thalassemic weight (boys) 81 34 17 10 10 1 9
thalassemic weight (girls) 45 9 15 8 5 2 6
non-thalassemic height (boys) 41 3 5 12 4 3 14
non-thalassemic height (girls) 30 1 2 2 4 5 16
non-thalassemic weight (boys) 41 3 5 12 4 3 14
non-thalassemic weight (girls) 30 1 2 2 4 5 16
A mixed-longitudinal method of growth study was adopted. From the first visit
onward each child of non-thalassemics and thalassemics category was followed up
till six years. However, there were several drop outs which had been replaced by the
new entrants into the study. The frequency of measurements made on the
thalassemic children and the controls, by sex and variables are shown in Table 1.
The pure longitudinal element of the data will be treated in a subsequent analysis.
The children visited the blood transfusion units as and whenever their
haemoglobin fell to a level when they physically felt very ill. During such visits
they were measured for lengthlheight and weight but only the measurements taken
on their birth dates were used in the present communication. Thus the ages refer to
exact ages.
GROWTH OF THALASSEMIC CHILDREN 343
All children were measured for lengthiheight and weight at forenoon following
the recommendation of Gam and Shamir (1958) and the measurements were taken
following Martin and Saller (1957). Supine length of the infants was taken up to the
nearest centimetre till 2 years of age with the help of an infantometer specially
designed for this purpose. Measurements were taken by one of us (JB). Body weight
was measured with a portable weighing machine and height was measured with an
anthropometric rod.
For clinical diagnosis of the thalassemics (i) estimation of haemoglobin, PCV,
MCV, reticulocyte count, (ii) agarose gel electrophoresis at pH 8.6 for detection of
haemoglobin variants, (iii) foetal haemoglobin by alkali denaturation and (iv) HbA2
by gel elution technique (Dacie and Lewis 1990) were performed. During the follow-
up period, pre-transfusion haemoglobin level was estimated for each child.
Data for length/height and weight of the thalassemics and the controls were
analysed by the method of Tanner (1951) applied for the mixed-longitudinal design.
Age specific means and variances for lengthiheight and weight of the thalassemics
were compared with the controls. Mean and variances of the increments have been
calculated on the same data by Tanner's (1951) method. Statistical significance of
the mean difference between the thalassemics and the non-thalassemics in all ages in
attained size as well as increments were tested by the unpaired Student's t-test. A p-
value of less than .01 have been considered to be significant.
3. RESULTS
Age and sex specific means and variances of lengths/heights and weights of the
thalassemic and non-thalassemic children are presented by age and sex in Tables 2
and 3. The means refer to the M h' calculated according to the formula of Patterson
(1950), an estimator of central tendency extracting most of the information contained
in mixed-longitudinal growth data (Tanner 1951).
In both traits and for both sexes the thalassemics showed lower mean values than
the non-thalassemics controls. For height, in both sexes, the mean difference
increased from one year and continues till 5 years after which it declined. For weight,
the increasing trend continued in all ages for boys excepted for the girls only at 6
years. As a whole, both thalassemic and non-thalassemic boys showed larger means
than the girls.
The variances in height for both sexes of the thalassemics were relatively larger
during the earlier ages which declines in mid ages and finally increased in the
subsequent years. On the contrary, variation in weights over different ages for both
sexes was more or less consistent.
The non-thalassemics, however, did not exhibit any consistent trend in variances
with respect to height, but in weight they exhibited the same consistency as in the
thalassemics. Statistically significant differences were observed between the means of
the thalassemics and non-thalassemics at all ages in both traits (p<.0 1).
One-year increments were calculated as follows: the increment between 1 and 2
years is assigned to the age 1.5 years, i.e. the centre of the interval (see Tables 4 and
5). The mean increments of lengthiheight of the thalassemics, as expected, were less
than the non-thalassemics, except at 5.5 years among the girls where a reversal was
noticed. The maximum difference in the mean increments between the thalassemics
344 CHAPTER 27
and the non-thalassemics was at age 1.5 in boys. However, among the girls the
maximum difference was at 4.5 years. Within the thalassemics group, in both sexes,
the maximum decline of the increments in height was observed after 3.5 years. For
weight, the difference in mean increments between the thalassemics and non-
thalassemics was maximum at 1.5 years (in both sexes). Further, in weight within
the thalassemics group the maximum decline in increment was noticed after 2.5
years of age.
Table 2. Means and variances of length/height (em) of the thalassemics and non-
thalassemics (normal controls) by age and sex. All differences were statistically
significant (t-Student p<.OJ)
thalassemics non-thalassemics
Age, years n mean variance n mean variance
Boys
1 23 71.59 1.85 20 73.50 0.36
2 22 77.98 1.79 21 82.83 0.64
3 37 83.56 0.96 29 90.25 1.11
4 37 90.14 0.76 30 97.54 1.06
5 40 95.66 0.76 33 105.08 0.54
6 38 101.47 0.81 31 110.45 0.25
Girls
1 13 68.50 3.71 16 73.00 0.68
2 21 76.27 1. 78 21 82.00 0.25
3 25 81.16 0.97 25 88.50 0.37
4 22 86.82 0.96 27 95.52 0.92
5 27 92.31 1.24 29 103.50 0.77
6 21 97.84 2.22 30 108.52 0.74
Table 3. Means and variances of weight (kg) of the thalassemics and non-thalassemics
(normal controls) by age and sex. All differences were statistically significant (t-
Student p<.OJ)
Boys
1 24 7.44 0.04 20 8.50 0.03
2 23 8.86 0.04 21 10.97 0.04
3 37 10.84 0.04 29 13.58 0.08
4 37 11.84 0.06 30 14.91 0.06
5 40 13.11 0.06 33 16.64 0.07
6 38 14.46 0.11 31 18.81 0.07
Girls
1 13 7.03 0.09 16 8.01 0.06
2 21 8.51 0.09 21 10.52 0.04
3 25 10.56 0.05 25 12.75 0.04
4 22 11.34 0.07 27 14.00 0.04
5 27 12.86 0.10 29 15.75 0.06
6 21 14.21 0.19 30 17.00 0.07
GROWTH OF THALASSEMIC CHllDREN 345
Table 4. Means and variances of increments in iengthiheight (ern/year) of the

thalassemics and non-thalassemics (normal controls) by age and sex. All differences
were statistically significant (t-Student p<.OJ) except for girls at the age of 5.5 years
Boys
1.5 22 5.65 0.53 21 9.09 0.32
2.5 37 5.05 0.45 29 7.54 0.35
3.5 37 5.71 0.54 30 7.93 1.30
4.5 40 4.80 0.42 33 7.79 0.42
5.5 38 4.79 0.38 31 6.11 0.18
Girls
1.5 21 6.41 1.01 21 8.72 0.16
2.5 25 4.45 0.40 25 6.15 0.08
3.5 22 5.32 0.79 27 6.70 0.25
4.5 27 3.97 0.30 29 9.31 0.29
5.5 21 4.49 0.18 30 3.98 0.14
Table 5. Means and variances of increments in weight (kg/year) of the thalassemics

and non-thalassemics (normal controls) by age and sex. All differences were
statistically significant (t-Student p<.OJ)
Boys
1.5 23 1.36 0.02 21 2.67 0.02
2.5 37 1.47 0.04 29 2.61 0.04
3.5 37 0.70 0.02 30 1.58 0.02
4.5 40 0.97 0.03 33 1.74 0.03
5.5 38 1.09 0.03 31 2.27 0.02
Girls
1.5 21 1.36 0.05 21 2.51 0.01
2.5 25 1.80 0.06 25 2.18 0.01
3.5 22 0.66 0.06 27 1.34 0.004
4.5 27 1.10 0.03 29 1.69 0.01
5.5 21 0.94 0.03 30 1.24 0.01
Statistically significant difference (p<.Ol) in mean increments between the

thalassemics and the non-thalassemics was also found (except for the girls height at
age 5.5 years).
346 CHAPTER 27
4. DISCUSSION
In the present study the thalassemic group of children have displayed early onset of
reduced size and delayed growth (from age 1 year) and behaved like untreated
patients most probably due to inadequacy and irregularities in taking blood
transfusion. It has been suggested by several workers that retardation of various
parameters at an early age are more likely to be secondary to chronic hypoxia,
followed by inadequate blood transfusion therapy (George et al. 1997). Some
workers (Fuchs et al. 1996, Benso et al. 1990) have remarked that while hyper-
transfusion regimen can modify the pattern of growth disorders, retardation in growth
and development persists even in a transfused children.
120,
Boys
§ 110
~ 100
'iii
<E 90
~
'&
c:
~ 80 . . - Thalassemia
>- Non-Thalassemia
•• _. & •••
'0 ~ICMR
0 70 ~NCHS
CD
60 0
25 ~
Boys
I
20
OJ
~
1: 15
Ol
'iii
==
>- 10 . . - Thalassemia
'0
0 Non-Thalassemia
- - - -&. - •
CD
5 &--ICMR
)( NCHS
00
Figure 1, Mean length (top) and weight (bottom) of thalassemic and non-thalassemic
boys, compared with the 50th centile of the Indian (ICMR 1972) and the NCHS
reference (Hamill et at. 1979)
Our results are in variance with the study by Johnston et at. (1966), and
Piomelli et ai, (1974) who found that growth retardation of the thalassemic children
from four years onwards only. In a study by Kattamis and Kattamis (1995), it was
reported that untreated patients with thalassemia major (type 1 and 2) showed
GROWfH OF THALASSEMIC CHILDREN 347
growth retardation during the first decade of life and were below the 3rd centile for
weight and height in comparison to the non-thalassemic children. In a recent Indian
study, carried out in New Delhi (George et al. 1997), the authors did not fmd any
significant difference between the tha1assemics and the non-thalassemics children in
average height. till 9 years. The study, however, has not provided any estimate of
growth increments.
120 I
Girls
E 110
0
Cl 100
1:
'iii
.c
:cc;, 90
c
..!!? 80 ....-- Thalassemia
>. - - - -€>- - - - Non-Thalassemia
'0
~ICMR
co 70
0
)( NCHS
60 0
2 3 4 5 6 7
Age, years
25, Girls
20
Cl
.:.!
1: 15
Cl
'iii
~
>.
10
'0 ....-- Thalassemia
0
co - - - - €>- - - - Non-Thalassemia
5 ~ICMR
)( NCHS
00 7
Figure 2, Mean length (top) and weight (bottom) of thalassemic and non-thalassemic
girls, compared with the 50th centile of the Indian (ICMR 1972) and the NCHS
reference (Hamill et at. 1979)
Age-specific mean heights and weights of the thalassemic children of both sexes
show lower scores than the 50th centiles of the ICMR (Indian Council of Medical
Research 1972) and the NCHS (Hamill et al. 1979) standard (Figures 1 and 2).
Moreover, it has been observed that mean body weight of the thalassemic children
corresponds to the 25th percentile of the ICMR standard at all ages. But for height
the corresponding ICMR percentile was less than 10. Thus it clearly indicates that
the children of the present study were shorter with respect to their weight. Overall,
the thalassemic children fell below the 5th percentile of the NCHS standard (Hamill
et al. 1979) and showed stunted growth from the age of one year.
348 CHAPTER 27
A number of studies have now established that stunted heights of the thalassemic
children are due to disproportionate growth between the upper and lower segments,
sitting height or trunk height being more affected than leg length (Rodda 1994,
Fuchs et al. 1997). Unfortunately, the present study does not provide any
opportunity to examine growth velocity of the two principal body segments.
The causative factors which have been found to be responsible for growth
retardation of the thalassemics, may be summarised as follows:
(i) Abnormal Hypothalamus Pituitary Gonadal axis function (Costin et al. 1979).
(ii) Decrease in the production oflGF -1 (Herington et al. 1981).
(iii) The toxic effects of Desferrioxamine (de Virgilis et al. 1988) (though not
applicable in the present study).
(iv) Cellular hypoxia, due to anaemia (Fuchs et al. 1996).
(v) Deficiencies in the micronutrients (Arcasoy and Cavdar 1975).
(vi) Undernutrition, due to inadequate nutrient intake (Fuchs et al. 1997).
(vii) Chronic liver disease (de Sanctis et al. 1991).
However the results from some of these investigations are inconsistent or
contradictory (Fuchs et al. 1997).
The present study however was not aiming to investigate the role of these factors
for the causation of the growth retardation in this heterogeneous thalassemic children
with small sample size. Heterogeneity might have been caused due to different
clinical histories of the patients. Constantoulakis et al. (1975) have commented that
low haemoglobin level and severity of the disease may hinder normal growth of the
thalassemic patients. Nevertheless, it is one of the first mixed-longitudinal study on
the thalassemic children of India which has generated some basic growth data.
Future studies with larger sample size should be carried out to disentangle the
relative role of these factors resulting in growth retardation of the Indian thalassemic
children.
Acknowledgements. The authors are grateful to the authorities of the Vivekananda

Institute of Medical Sciences Department of Biotechnology, for kindly supporting
the project. R. Saha of the Anthropology and Human Genetics Unit, Indian
Statistical Institute deserves special thanks for his help in data processing.
5. REFERENCES
Arcasoy, A., and Cavdar, A.C., 1975, Changes of trace minerals (serum iron, zinc, copper and
magnesium) in thalassemia. Acta Haematologica, 53, 341-345.
Benso, L., Pastorin, L., Gambotto, S., Petri, A., Gianino, P., Signorile, F., and Gabutti, V., 1990,
Auxological Study of the onset of puberty and of the age at menarche in B-thalassemic patients.
Acta Medica Auxologica, 22, 51-56.
Borgna-Pignatti, C.D., Stefano, P., Zonta, L., Vullo, C., de Sanctis, V., Melevendi, C., Naselli, A.,
Masera, G., Tarzoli, S., and Gabutti, V., 1985, Growth and sexual maturation in thalassemia major.
Chatterjee, J.B., 1965, Some aspects of haemoglobin E and its genetic interaction with thalassemia.
Indian Journal of Medical Research, 53, 377-398.
Constantoulakis, M., Panagopoulos, G., and Augoustaki, 0., 1975, Stature and longitudinal growth in
thalassemia major. Clinical Pediatrics, 14, 353-368.
Costin, G., Kogut, M.D., Hyman, C.B., and Ortega, J.A., 1979, Endocrine abnormalities in thalassemia
major. American Journal of Disease in Childhood, 133,497-502.
GROWTH OF THALASSEMIC CHILDREN 349
Dacie, J.V., and Lewis, S.M., 1990, Investigation of the abnormal haemoglobins. In Practical
Haematology (England: Churchill, Livingstone).
de Sanctis, V., Atti, G., Banin, P., Orzincolo, C., Cavallini, A.R, Patti, D., and Vullo, C., 1991, Growth in
thalassaemia major. Acta Medica Auxologica, 23, 29-36.
de Virgilis, S., Congia, M., Frau, F., Argiolu, F., Diana, G., Cucca, F., Varsi, A., Sanna, G., Podda, G.,
Fodde, M., Pirastu, G.F. and Cao, A., 1988, Desferrioximine-induced growth retardation in patients
with thalassaemia major. Journal of Paediatrics, 113,661-669.
Fuchs, GJ., Tienboon, P., Khaled, M.A., Nimsakul, S., Linpisarn, S., Faruque, A.S.G., Yutrabootr, Y.,
Dewier, M., and Sus kind, RM., 1997, Nutritional support and growth in thalassaemia major.
Archives of Disease in childhood, 76, 509-512.
Fuchs, GJ., Tienboon, P., Linpisarn, S., Nimsakul, S., Leelapat, P., Tovanabutra, S., Tubtong, V.,
Dewier, M., and Suskind, R.M., 1996, Nutritional factors and thalassemia major. Archives of
Gam, S.M., and Shamir, S., 1958, Methods of Research in Human Growth (Springfield, Illinois: e.e.
Thomas).
George, A., Bhaduri, A., Sen, S., and Choudhry, V.P., 1997, Physical growth parameters in thalassemic
children. Indian Journal of Pediatrics, 64, 861-871.
Hamill, V.V., Drizd, T.A., Johnson, C.L., Reed, RB., Roche, A.F., and Moore, W.M., 1979, Physical
Growth: National Centre for Health Statistics Percentiles. The American Journal Of Clinical
Nutrition, 32, 607-629.
Herington, A.C., Werther, G.A., Matthews, RN., and Burger, H.G., 1981, Studies on the possible
mechanism for deficiency of non-suppressible insulin-like activity in thalassemia major. Journal of
Clinical Endocrinology and Metabolism, 52, 393-398.
Indian Council of Medical Research, 1972, Growth and Physical development of Indian infants and
children. Technical report series no. 18 (New Delhi: ICMR).
Johnston, F.E., Hertzog, K.P., and Malina, RM., 1966, Longitudinal growth in thalassemia major.
American Journal of Disease in childhood, 122,396-403.
Kattamis, e.A., and Kattamis, A.C., 1995, Management of thalassemias: Growth and development,
honnone substitution, vitamin supplementation and vaccination. Seminars in Haematology, 32, 269-
279.
Martin, R, and Saller, K., 1957, Lehrbuch der Anthropologie. (Germany: Jena, Gustav- Fischer).
Patterson, H.D., 1950, Sampling on successive occasions with partial replacement of units. Journal of
the Royal Statistical Society, 12,241-255.
Perignon, F., Braunar, R, Souberbielle, J.e., de-Montalembert, M., and Girot, R., 1993, Growth and
endocrine function in major thalassemia. Archives of French Paediatrics, 50, 657-663.
Piomelli, S., Karpatkin, M.H., and Arzanian, M., 1974, Hypertransfusion regimen in patients with
Cooley's anemia. Annals of the New York Academy of Sciences, 232,186-192.
Rodda, C.P., 1994, Body proportions in patients with homozygous beta - thalassemia. Acta Pediatrica
(Supplement), 406, 107-108.
Saka, N., Sukur, M., Bundak, R., Anak, S., Neyzi, 0., and Sedikoglu, G., 1995, Growth and puberty in
thalassemia major. Journal of Pediatric Endocrinology and Metabolism, 8, 181-186.
Talukder, G., and Sharma, A., 1994, Haemoglobinopathies in India. Journal of Human Ecology, 3, 125-
130.
Tanner, lM., 1951, Some notes on the reporting of growth data. Human Biology, 23, 93-159.
Venna, I.C., Choudhry, V.P., and Jain, P.K., 1992, Prevention of thalassemia: A necessity in India.
Indian Journal of Paediatrics, 59, 649-654.
Weatherall, DJ., and Clegg, lB., 1981, The Thalassemia Syndromes (Boston: Blackwell Scientific
Publications) 3rd edition.
Yeslipek, M.A., Bircan, I., Oygur, N., Ertug, H., Vegin, 0., and Guven, A.G., 1993, Growth and sexual
maturation in children with thalassemia major. Haematologica, 78, 30-33.
ApPENDICES
Life sketch of Sudhir Ranjan Das
Sudhir Ranjan Das received his initial M.Sc. degree in Pure Physics from Calcutta
University in 1933. He worked as a Research Fellow in Biophysics at the Bose
Institute from 1942 to 1948. He developed special interest in radiology and was
appointed Radiologist-Anthropologist in the Anthropological Survey of India in
1948. It was a time when the Survey was in search of young talented individuals to
build a strong research department.
With this background and training, he specialised in human genetics and human
physical growth. He developed the radiology section of the Survey, and commenced
a longitudinal growth study among the Brahmin, the Kayastha and Vaidya
communities in the suburbs of Calcutta (Sarsuna-Barisha Growth Study). A
radiological study of the skeletal maturation formed the part of the project. Besides
being involved in this long-term project, Das explored a number of areas like
serology, somatology, PTC taste factor, sickle-cell traits and colour blindness,
distribution of LDH variants and placental alkaline phosphatase, enzyme, serum
proteins and Bart's haemoglobin survey. His works were not limited to Bengal or
Calcutta suburbs, rather he worked on populations in Orissa, Nagpur and Kerala as
well.
He retired from the Survey in 1968. After his retirement from the Survey he
joined the Indian Statistical Institute, Calcutta, as Professor and Head of the
Anthropometry and Human Genetics Unit (now Anthropology and Human Genetics
Unit) and continued his research activities until he left the Institute. Das knew four
languages, namely, Bengali, Hindi, Sanskrit and English.
Das had published extensively and at times it is difficult to prefer one above the
other. Nonetheless some of them are of special significance, namely, Inheritance of
PTC taste character in man: an analysis of 126 Rarhi Brahmin families of West
Bengal in Annals of Human Genetics (1958), Blood groups (ABO, M-N, Rh), ABH
secretion, sickle-cell, PTC taste and Colour Blindness in the Mahar of Nagpur, in
the Journal, Royal Anthropological Institute (1961), A spectrophotometric skin
colour survey among four Indian caste and Tribes, in Zeitschrift flir Morphologie
und Anthropologie (1963), Caste variations of lactate dehydrogenase in the
Dhangars of Maharashtra, India, in the Japanese Journal of Human Genetics
(1973), Blood groups, Serum proteins, haemoglobin, and some serum and red-cell
enzymes among the Kaoras of 24-Parganas in West Bengal, in Human Heredity,
(1974), and A Radiological Study of Skeletal Maturity in Hand-Wrist-Area of
Bengali Boys and Girls, in Anthropological Survey ofIndia, Calcutta (1976).
The major contribution of late Prof. Das are also given herewith chronologically:
Born on 11 October 1909 in Khulna District, now in Bangladesh.

2 Activities during 1935-39: Obtained M.Sc. in Physics, Calcutta University,
1933 and joined the Research X-ray Laboratory there in 1935. On his own
initiative, started investigations into the structure of S-molecular or atomic
aggregates in the various solid forms, the liquid (114 0 C - 160 0 C), the plastic
sulphur and the super-cooled (to room temperature) liquid sulphur.
Besides many new phenomena, a new crystalline variety of Sulphur (S-omega)
was discovered and its hexagonal crystalline structure, its stability, and
353
354
conversion to the rhombic, and other properties were found out. The results
appeared mainly in Indian Journal of Physics, Calcutta and two papers in
Nature, London, 1937/39. J.R. Partington, London, Newton & Friend, from
Cambridge University, quoted the findings elaborately in the Advanced
Chemistry books. A paper was read and published in the Proceedings of the
International Union of Chemistry on sulphur and selenium in Munster
(Germany) 1956.
The entire work was done with new types of suitable X-ray diffraction cameras,
designed by him.
During 1940-47 he worked in the Bose Institute (Acharya J.C. Bose), Calcutta,
with a break of 18 months in Kabul University as a lecturer of Physics. In the
Bose Institute he took up:
(i) the construction of a powerful source of ultra-sonic generator for the Defence;
(ii) the study of protein structure by the X-ray diffraction method;
(iii) the induction of mutations in some plants by X-ray irradiation of seeds;
(iv) the construction of a powerful artificial source of neutron beam suitable for
use in medicine.
3 He worked in the Anthropological Survey, Government of India, from 24
October 1948 to 31 March 1968. The following lines drew his attention:
(a) X-ray study of hand-wrist skeletal (by epiphyseal union) and physical growth in
about 500 Bengali Boys and Girls - from birth to 21 years - in mixed
longitudinal samples (Sarsuna-Barisha Growth Study). Twenty-two
anthropometric characters of each child, measured on birth-days or half birth-
days, supplied the basic growth data.
The growth and maturity field study covered a period of 14 years - 4 April
1952 to 31 March 1966 - and all measurements were take by a single observer
(S.R.D.). The skeletal maturity was assessed on the Greulich and Pyle
Standards U.S.A. 1958. A report published by Anthropological Survey of
India, Government ofindia, Calcutta, 1970.
Professor J.M. Tanner (London) kindly contributed to the analysis of the
voluminous growth metric data in 1969-70. The data have since been only
partially analysed, by Prof. J.M. Tanner, Prof. R.C. Hauspie (Brussels), Prof.
S.R. Das, Dr. M.A. Preece and Dr. M.J. Baines (London). Several papers have
already been published in Annals of Human Biology and elsewhere. More
papers are getting ready for press.
Scholars from all over the world may have access to these materials on
application to the Director of the Anthropological Survey of India, Government
ofindia, Calcutta, India
(b) A paper in three parts appeared in Metron, Roma (1953-55) on "A Mathematical
analysis of the phenomena of human twins, triplets, and higher plural births, I,
II, III". This paper answered definitely several long-standing questions about
these phenomena, removing certain wrong concepts, and throwing new light on
them.
(c) S.R. Das introduced a correct methodology, developing an instrument,
dactylometer, introducing a scientific positioning of the hand-wrist-forearm on a
board and measuring the actual positions of the finger tips in this standard
position.
355
He studied a few castes and tribal populations for distribution of the manual
Digital Formulas in them.
The relative lengths and positions of the hand bones were investigated in
relation to the digital formula of the hand by X-ray siagrams in some 200 adult
men and women. A paper appeared in Zeitschrift fUr Morphologie und
Anthropologie.
(d) Phenylthiocarbamide (PTC) taste inheritance:
The single-locus di-allelic inheritance with complete dominance of the Taster
(T) gene seemed not to fit in with carefully collected PTC taste sibship or
family data (Harris & Kalmus, Annals of Human Genetics, London 1956-58).
Similar data obtained from some 125 Bengali families also indicated this
discrepancy. The hypothesis was modified by supposing that the T-gene was
not completely dominant in the heterozygotes (Tt) and as a result some of the
Tt individuals were non-tasters like (tt).
Papers appeared in Annals of Human Genetics, London 1956 & 1958 and
another in Proceedings of the Indian Anthropological Society, 1960, which
demonstrated satisfactory agreement of the incomplete dominance hypothesis
with his family data.
(e) Gene distribution in various Indian castes and tribes - the following gene
system were considered:
(i) Blood groups and types (AI, A2, B, 0, MN, Ss, Rh-C-D-E, Kell, Duffy,
Diego);
(ii) Secretion of ABH in saliva;
(iii) Sickle-cell;
(iv) red-green colour-blindness;
(v) PTC taste genes.
S.R. Das worked in the Indian Statistical Institute, Calcutta from 1968 (April)
to June 1973, when he retired from anthropology officially. Here he worked in
collaboration with Prof. B.N. Mukherjee of the Institute and Professor R.L.
Kirk of the Australian National University. In the Institute, he carried out the
following research activities:
He studied distribution of the genetic variants of haptoglobin and transferrin,
abnormal haemoglobins, the various red cell enzyme variants - the method
adopted was that of starch-gel Electrophoresis. In course of the studies two new
genetic variant ofLDH namely, Cal-l and Cal-2 were discovered. Hb-Barts was
detected in a sample of about 800 cord blood samples (1 %). Enzymes in human
placenta were investigated in the Calcutta hospitals and the results appeared in
Japanese Journal of Human Genetics. A number of papers appeared in Human
Heredity (Basel), in Human Genetik (Germany), etc ...
He passed away on 22nd August, 1998, at the age of 89 years.

List of Publications of Sudhir Ranjan Das
before 1941
Das, S.R., 1936, A study of the structures of the various allotropes of Sulphur by
the X-ray Powder Photograph method. Science and Culture, vol. 1, June, 784-
785.
Das, S.R., and Ray, K, 1936, Structure of the allotropic Modifications of Sulphur.
Science and Culture, vol. II, August, 108-109.
Das, S.R., and Ray, K, 1937, The allotrope of Sulphur (alpha) study by the X-ray
diffraction method. Science and Culture, vol. II, June, 650-652.
Das, S.R., 1937, X-ray analysis of the structure of jute fibres. Science and Culture,
vol. II, June, 652-653.
Das, S.R., and Ghose, K, 1938, The structure of Sulphur particles in colloidal
suspension in water. Science and Culture, vol. IV, August, 132-133.
Das, S.R., 1938, A study of sulphur allotropes by the X-ray diffraction method. Part
1. Indian Journal of Physics, 12, 163-181.
Das, S.R., and Ghosh, K., 1939, A study of Sulphur allotropes by the X-ray
diffraction method (Part II). White Sulphur, Black Sulphur and Colloidial
Sulphur Suspensions in Water. Indian Journal of Physics, 13,91-105.
Das.S.R., and Dasgupta, K, 1939a, Conversion of Vitreous and monoclinic(alpha)
Selenium to the Hexagonal modification. Nature, 3613, 165.
Das,S.R., and Dasgupta, K., 1939b, X-Ray Diffraction by Supercooled Liquid
Sulphur. Nature, 143, 332.
Ray, B.B., Das, S.R., and Bagchi, N.1939. The K-absorption edges of cobalt in
cobalt metal and its compounds. Science and Culture. vol. IV, 528-529.
Ray, B.B., Das, S.R., and Bagchi, N., 1940, Secondary K-absorption edges of
Cobalt salts in solid and liquid solutions. Indian Journal of Physics, 14,37-54.
Roy, B.B., Dasgupta, K., Bose, H., and Das, S.R., 1940, The Flouroscence of
organic compounds by X-rays. Science and Culture, 5, 496-497.
1941 - 1950
Dasgupta, K., and Das, S.R., 1941, X-ray study of selenium in the liquid and
colloidial state. Indian Journal of Physics, 15,401-409.
Dasgupta, K, Das, S.R., and Ray, B.B., 1941, A study of allotropes of Selenium
by the X-ray diffraction method. Indian Journal of Physics, 15,389-399.
1951 - 1960
Das, S.R., 1952, The Human Manual digital formula and a standard hand axis - a
new method and an instrument foe their correct determination. Bulletin of the
Department of Anthropology Calcutta, 1, 105-114.
Das, S.R., 1954a, An instrument of direct determination of manual digital formula -
The dactylometer. Bulletin of the Department of Anthropology, Calcutta, 3, 152-
154.
357
358
Das, S.R., 1954b, A study of manual digital formula among the Bengalee and three
south Indian tribes. Bulletin of the Department of Anthropology, Calcutta, 3, 73-
85.
Das, S.R., and Ghose, L., 1954, A genetic Survey among the Paniyans. A South
Indian aboriginal tribe. ABO, MN, blood groups, secretor factor and taste
ability. Bulletin of the Department of the Anthropology, Calcutta, 3, 65-70.
Das, S.R., 1955, A somatological study of the Paniyans of Wyanaad. Bulletin of
the Department of Anthropology, Calcutta, 4, 51-61.
Das, S.R., and Mukherjee, D.P., 1955, A Roentgenometric Study of hand bones of
the Bengalee. Bulletin of the Department of Anthropology, Calcutta, 4, 2, 11-30.
Das, S.R., 1956a, A contribution to the heredity of the PTC taste characters based
on a study of 845 sib pairs. Annals of Eugenics, 20, 334-343.
Das, S.R., 1956b, A mathematical analysis of the phenomena of Human twins in
higher plural births (Refmed method). Part III. Metron, 18,63.
Das, S.R., 1958, Inheritance of the PTC taste character in man, An analysis of 126
Rarhi Brahmins families of West Bengal. Annals of Human Genetics, 22, 202-
212.
Das, S.R., and Sastry, D.B., 1960, Simplified maximum likelihood estimation of
the A-B-O gene frequencies: Results for a few Indian people surveyed. Bulletin of
the Anthropological Survey ofIndia, 9, 2, 1-8.
1961 - 1970
Das, S.R., Kumar, N., Bhattacharjee, P.N. and Satry, D.B, 1961, Blood Groups
(ABO, MN & Rh ), ABH secretion, sickle cell, PTC taste and colour blindness
in the Mahar of Nagpur. Journal of the Royal Anthropological Institute, London,
91, 345-355.
Mukherjee, D.P., and Das, S.R., 1961, Effect of Geological Aging on the particle
size of the bone mineral. Current Science, 30, 138-139.
Das, S.R., and Mukherjee, D.P., 1962, The anatomical basis of the human digital
formulas. Zeitschrift flir Morphologie und Anthropologie, 52, 163-175.
Das, S.R., Bhattacharjee, P.N., Sastry, D.B., and Mukherjee, D.P., 1962, Blood
groups (ABO, MN, Rh) ABH secretion, sickle cell trait and colour blindness in
the Bado Gadaba and Pareng Paroja of Koraput District of Orissa. Bulletin of the
Department of Anthropology, 11,324, 145-157.
Das, S.R., and Bhattacharjee, P.N., 1963, Blood groups (AI, A2, B) ABH
secretion, sickle cell, PTC taste and colour blindness in the Rajbanshi of the
Midnapur District, West Bengal. Bulletin of the Department of Anthropology,
12, (1 and 2), 1-6.
Das, S.R., and Mukherjee, D.P., 1963, A spectrophotometric skin colour survey
among four Indian castes and tribes. Zeitschrift fUr Morphologie und
Anthropologie, 54, 190-200.
Das, S.R., Mukherjee, D.P., and Bhattacharjee, P.N., 1963, PTC taste threshold
distribution in the Bado Gadaba and the Bareng Paroja of Koraput District. In
Orrissa. Acta Genetica et Satistica Medica, Basel, 13,369-377.
Das, S.R., and Mukherjee, D.P., 1964, Phenylthiocarbamide taste sensitivity
survey among the Pareng Gadaba, the OHaro Gadaba and Konda Paroja of
Korapur District, Orissa. Acta Genetica et Statistica Medica, Basel, 14, 168-176.
359
Das, S.R., 1966, Genetics of PTC taste character in Man - incomplete dominance
and the penetrance of the taste gene. In: Human Adaptability to environments
and physical fitness. M.S. Malhotra (ed). Defence Institute of Physiological and
Allied Sciences, Madras, Research and Developmental Organisation. Ministry of
Defence, Government ofIndia, 250-258.
Das, S.R., Sastry, D.B., and Mukherjee, D.P., 1966, Blood Groups (AI, A2, BO,
MN, Rh) and ABH secretion in the Pareng Ollaro Gadaba and the Konda Paroja
of Koraput District in Orissa. Acta Genetica et Statistica Medica, Basel, 16, 179-
183.
Das, S.R., 1966, Application of phenylthiocarbamide taste character in the study of
racial variation. (data on World taste gene distribution). Journal of the Indian
Anthropological Society, Calcutta, 1,63-80.
Das, S.R., Mukherjee, D.P., and Sastry, D.B., 1967, Sickle cell trait in Koraput
district and other parts of India. Acta Genetica et Statistica Medica, 17, 62-73.
Das, S.R., Mukherjee, D.P., and Bhattacharjee., P.N., 1967, Survey of the Blood
Groups and PTC taste among the Rajbanshi caste of West Bengal. (ABO, MNS,
Rh, Duffy and Diego). Acta Genetica, Basel, 17,433-445.
Das, S.R., Mukherjee, D.P., and Sastry, D.B., 1968, A somatological Survey of
the five tribes in the Koraput district, Orissa. Bulletin of the Department of
Anthropology, Calcutta, 17,4, 400-422.
Das, S.R., Mukherjee, B.N., and Das, S.K., 1970, The distribution of some
enzyme systems among Bengalis. The Indian Journal of Medical Research, 58,
866-875.
Das, S.R., Mukherjee, B.N., Das, S.K., Blake, N.M., and Kirk, R.L., 1970,
Placental alkaline Phosphatase types in Calcutta, India. Japanese Journal of
Human Genetics, 3,155-158.
Das, S.R., Mukherjee, B.N., Das, S.K., Ananthakrishnan, R., Blake, N.M., and
Kirk, R.L., 1970, LDH variants in India. Humangenetik, 9, 107-109.
1971 - 1980
Das, S.R., Mukherjee, B.N., and Das, S.K., 1971, Genetic variants of Lactate
Dehydrogenese in India. Journal of the Indian Anthropological Society, 6, 2,
135-144.
Das, S.R., Mukherjee, B.N., and Das, S.K., 1972, Caste and age variations of the
incidence ofLDH variants in the Bengali Hindus. Humangenetik, 14, 151-154.
Das, S.R., Mukherjee, B.N., Das, S.K., and Malhotra, K.C., 1972, Four types of
genetic variants ofLDH found in India. Human Heredity, 22, 264-270.
Das, S.R., 1973, Tests for incomplete penetrance of the dominant allele in the
heterozygote in the unifactorial autosomal diallelic inheritance. Journal of the
Indian Anthropological Society, 8, 1-12.
Roy M., and Das, S.R., 1973, Blood Group Genetic Survey among the Poundra
Kshatriya Caste of 24 Parganas in West Bengal. Journal of the Indian
Anthropological Society, 8, 1, 83-86.
Das, S.R., Malhotra, K.C., Mukherjee, B.N., and Das, S.K., 1973, Genetic
variants of Lactate dehydrogenase in the Dhangars of Maharashtra, India. Japanese
Journal of Human Genetics, 18,305-308.
360
Das, S.R., Das, S.K, and Dawn, C.S., 1973, Hemoglobin Barts in Bengali Hindu
caste studies in Calcutta. Human Heredity, 23, 381-385.
Das, S.R., Mukherjee, B.N., Das, S.K, Roy, M., and Chhatui, S.S., 1974, Blood
groups, serum proteins, haemoglobins and some serum and red cell enzymes
among Kaoras of24 Parganas in West Bengal (India). Human Heredity, 24, 24-
31.
Das, S.R., Mukherjp,e, B.N., and Das, S.K, 1974, Caste variation in the
distribution of Placental alkaline Phosphatase genes among the Hindus of West
Bengal. Annals of Human Biology, 1,65-71.
Das, S.R., Malhotra, KC., Mukherjee, B.N., and Das, S.K, 1974, LDH and MDH
variants in five castes around Delhi, India. Japanese Journal of Human Genetics,
18,401-404.
Mukherjee, B.N., and Das, S.R., 1974, Distribution of some Polymorphic Enzyme
Group Systems in India. In: Human Population Genetics in India.(edited by
Sanghvi,L.D, Balakrishnan,V., Bhatia, H.M., Sukumaran, P.K, and Undevia,
J.N.) Proceedings of the first Conference of the Indian Society of Human genetics,
held in Bombay February, 14- 16, page, 21-49.0rient Longman Ltd.
Chakraborty, R., Das, S.R., Roy, M., Mukherjee, B.N., and Das, S.K, 1975, The
effect of parity on Placental Weight and birth weight.Interaction with Placental
alkaline phosphatase genotype. Annals of Human Biology, 2, 227-234.
Das, S.R., Roy, M., Paul, A., and Chakraborty, R., 1975, A growth study of
Indian Infants. Non relationship with Placental alkaline phosphatase genotypes.
Chakraborty, R., Roy, M., and Das, S.R., 1975, Proportion of low birth weight
infants in an Indian population and its relationship with maternal age and parity.
Human Heredity, 25,73-74.
Mukherjee, B.N., Das, S.R., 1975, Distribution of some polymorphic enzyme
group systems in India.Proceedings of the International Conference of Indian
Society of Human Genetics Volume, 1.
Das, S.R., Mukherjee, D.P., Bose, L., and Das, A., 1976, A radiological study of
skeletal maturity in hand and wrist area of Bengali Boys and Girls. Memoir no.
38. Anthrpological Survey oflndia, Calcutta.
Das, S.R., 1976, Evaluating Rh chromosome frequencies. Analysis of Lepcha data.
Anthropologist, 23, 1-5.
Mukherjee, B.N., Das, S.K, Malhotra, K.C., and Das, S.R., 1977, Haptoglobin
and acid phosphatase gene distribution in Dhangars of Maharashtra, India.
Journal of Genetics, 63, 39-45.
Hauspie, R.C., Das, S.R., Preece, M.A., and Tanner, J.M., 1980, A longitudinal
study of the growth in height of boys and girls of West Bengal (India) aged six
months to 20 years. Annals of Human Biology, 7, 429-441.
1981 - 1990
Hauspie, R.C., Das,S.R., Preece, M.A, and Tanner, J.M, 1982, Degree of
resemblance of the patterns of growth among siblings in families of West Bengal
(India). Annals of Human Biology, 9,171-174.
Hauspie, R., Das, S.R., Preece, M.A., and Tanner, J.M., 1984, Interrelationships
between various aspects of the growth pattern in weight, and adult sex
361
dimorphism in weight of Bengali (Indian) children. In: Human Growth and

Development, edited by J. Borms, R. Hauspie, A. Sand, C. Susanne, M.
Hebbelinck (New York: Plenum Press), 271-276.
Das, S.R., 1985, Mixed-Longitudinal growth data for 22 measures - The Sarsuna-
Barisha series, India, volume 1 boys and volume 2 girls. Anthropological
Survey of India, Calcutta.
Hauspie, R.C., Das, S.R., Preece, M.A., Tanner, J.M., and Susanne, C., 1985,
Decomposition of sexual dimorphism in adult size of height, sitting height,
shoulder width and hip width in a British and West Bengal Sample. In: Human
Sexual Dimorphism, edited by J. Ghesquiere, R.D. Martin and F. Newcombe,
Taylor and Francis, 207-215.
After 1991
Hauspie, R.C. and Das, S.R., 1995, Short term variations in growth rate in height,
sitting height and biacromial diameter in Bengali Children. In: Essays on
Auxology presented to James Mourilyan Tanner by his former colleagues and
fellows. Edited by R. Hauspie, F. Falkner and G.Lindgren. Castlemead
Publications., 260-268.
Dasgupta, P., and Das, S.R., 1997, A cross-sectional Growth study of trunk and
limb segments of the Bengali Boys of Calcutta. Annals of Human Biology, 24,
363-369.
INDEX
Aborigines, 162 Genetic factors, 123, 127-128,232
Adipose tissue, 91 Genetics, 109,205-206,208-209,212-213,216
Adolescent growth, 92-93, 237-238, 241, 247- Growth and thalassemia, 341
248 Growth charts, 226
Africa (secular trend in -), 161-162, 165 Growth faltering, 251-252, 262-263
Age at growth completion, 21, 26-30 Growth model, 17-23,26,30
Age at peak velocity, 17-18, 21, 27 Growth monitoring, 45-36
American-Japanese parentege, 281 Growth references, 45
Animal short-term growth, 324 Growth retardation, 251-252, 262, 264-265
Appropriate-for-gestational age (AGA), 334 Growth spurt, 238, 243, 245
Australian Aborigines, 109 Growth standards, 46
Growth velocity, 251, 261, 263-265
Basque country, 33
Bengali children, 223 Heritability, 118-119
Bioimpedance analysis, 71 Hormones and short-term growth, 328
Biological parameters, 17-18, 21-22, 26-27, 30, Hungarian National Growth Study, 301
230
Birth weight, 133 India, 223, 237, 341
Blacks (secular trend in -), 161-162 Infectious diseases, 264-265
Body composition, 67-69 Interaction (among genes), 80-81, 85, 87
Body mass index, 67-69,143,171
Body proportions, 142,205-211,213-214,216, Knemometry,321-322
218,314
Body segments (growth of -),234,241,313 Linkage (tree-), 80, 85
Breast feeding, 347-48, 52 Linkage analysis, 80, 82-87
Budapest children, 301 LMS-method, 33-35, 38-40, 225
Logical inductive ability, 186
Calcutta, 91 Longitudinal growth studies, 147, 223
Centile lines, 33-35, 39-41, 225 Longitudinal growth, 251-253, 259, 261, 263
Chaotic growth, 324 Low-birth-weight (LBW), 133, 333-334
Child growth, 252, 263 Low-income countries, 252, 262-264
Cognitive thinking, 179, 181 Lumping and splitting, 80, 84, 87-88
Count-Gompertz model, 17-18,20-23,26,30
Major genes, 80, 86-87
Dental arches, 111-112, 114 Malnutrition, 123, 174-175
Dentofacial morphology, 109 Mastication, 109, 115, 120
Developmental coefficient in neonates, 337 Maya, 210, 212-214, 216
Dietary environment, 127 Mean-constant curve, 17, 23, 25
Dietary habits, 124-125, 127 Menarche, 18,21,23,26-30, 139, 142
Dissimilarities in growth patterns, 234 Mental ability, 179, 181
Distance curve, 18, 226 Meta-analysis, 80, 84
Dual energy X -ray absorptiometry, 68-69 Mexico, 55
Dynamics of growth, 223 Mid-growth spurt, 19,21-22,30
Migration, 214
Early growth studies, 6 Miliary data, 5
Early measuring instruments, 5 Mini-growth spurts, 323
Ethnic differences, 269-270, 274 Mini-knemometry, 322
European studies, 11 Mixed parentage, 281, 288
Multivariate methods, 84
Familial growth data, 232
Family case studies, 123 Nature and nurture, 9
Fat distribution, 91-100 Neonatal growth problems, 333
Feeding recommendations, 45-48, 51-52 Neonates (short-term growth in -), 328
Fels Longitudinal Study, 147 Nepal, 251
Fijian children, 269 Newborn growth, 333
363
364
Non-linear growth model, 22-23, 30, 225 Splitting (Lumping and -),84,87-88
North American studies, 10 Study design, 82-83, 148
Nutritional status, 129, 169-170 Stunting, 251-252, 256, 258-259, 262-265
Obesity, 141-142 Teeth, 111-115, 118
Overweight, 141-142 Thalassemia, 341
The Ancient World, 3
Paediatrics, 67-69 The eighteenth century, 5
Patrilineal families, 126 The Middle Ages, 4
Patrilocal families, 126 The nineteenth century, 7
Peak-height velocity, 233, 240-241, 243, 245 The Renaissance, 4
Peri-urban, 251 Tooth emergence, Ill, 113-1l4
Periodicity in growth, 323 Tree linkage, 80, 85
Physique, 305 Twin studies, 123
Planning a new study, 152
Poland (secular trend in -), 162-165 Undernutrition, 55, 64, 130,240-241,247,249
Population growth studies, 12
Prague children, 314 Velocity curve, 18, 20-22, 226, 238, 245, 226
Preece-Baines model I, 34-35, 40-41, 225 Venezuela, 129
Principal Component Analysis, 91, 93, 95 Verbal ability, 186
Proportions, 142
Psychomotor development, 333 Wasting, 252, 256, 258-259, 262
Public health, 140-141 Weight-for-age, 132
Weight-for-height, 170, 258
Racial classification, 8 Whites (secular trend in -),161-162
Reference curves, 226
Rural area (growth in -), 237-238, 240, 245-247
Saltatory growth, 323

Samoan children, 269
Sarsuna-Barisha Growth Study, 223
Scholastic performance, 169-170, 175
Seasonality, 252, 264
Secular decrease in height, 139
Secular trend (factors), 159
Secular trend, 124, 129, 137-143, 147
Sex dimorphism in growth, 230-231, 252, 263-
264,269,314
Sexual maturation, 142, 179, 183
Short-term growth in animals, 324
Short-term growth, 234, 321
Short-term growth (mathematics of-), 325
Similarities in growth patterns, 234, 314
Similarities test, 190
Skeletal maturity (population differences in -),
286
Skeletal maturity, 55-57, 63-64" Ill, 115-116,
281,284
Skelic index, 269

Skinfolds, 70-71, 91-98,102,173
Small-for-date (SID), 334
Socio-economic factors, 91,130,160-165,174,
179, 181, 183,211-213
Somatotype, 301
South Asia, 251, 262, 264

Perspectives in Human Growth

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Perspectives in Human Growth

Uploaded by

Copyright:

Available Formats

PERSPECTIVES IN HUMAN GROWTH,

DEVELOPMENT AND MATURATION

A tribute to Professor Sudhir Ranjan Das

Springer-Science+Business Media, B.Y.

ISBN 978-90-481-5820-1 ISBN 978-94-015-9801-9 (eBook)

Printed on acid-free paper

All Rights Reserved

Softcover reprint of the hardcover 1st edition 2001

List of contributors .................................................................................. .ix

Foreword .................................................................................. xiii

Editors note ........................................................................................... xv

Methodological Aspects of Growth Studies ................................................. 1

The Measurement of Human Growth: A Historical Review

2 Individual Physical Growth Models and Biological Parameters of

3 Smoothing Centile Curves of Height of Basque Boys and Girls by

4 A New International Growth Reference for Young Children

5 Fels and Tanner-Whitehouse Skeletal Ages of School Children 7-13

6 Assessment of Childhood and Adolescents Body Composition: A

Genetic and Environmental Factors ......................................................... 77

7 Genetics of Complex Traits with Particular Attention to Fat

8 Subcutaeneous Adipose Tissue Distribution in 7- to 16-Year Old

9 Dentofacial Morphology, Growth and Genetics: A Study of

10 Riddles in Human Development Patterns: a Few Family Case Study

11 Nutrition in Venezuela at the End of the Millennium

12 Thoughts on Secular Trends in Growth and Development

13 Secular Trends and Longterm Serial Growth Studies

14 Secular Trends in Body Height - Indicator of General Improvement

15 Patterns of Growth and Scholastic Performance - A Case Study of

16 Mental Ability and Cognitive Thinking in Relation to Sex,

Population Differences in Growth ...................................................... ... 203

17 How Genetic Are Human Body Proportions?

18 The Sarsuna-Barisha Longitudinal Growth Study: Dynamics of

19 Linear Components of Growth Among Rural Indian Children

20 Growth Stunting among Children, Aged Birth to 5 Years, in Peri-Urban

22 Skeletal Maturity in Children of Mixed American and Japanese

Biological Aspects of Growth ................................................................ 299

23 Somatotypes of Budapest Children

24 Differential Rate of Growth of the Human Body Parts

26 Relations between Head Growth and Psychomotor Development in

27 Growth in Length and Weight of Thalassemic Children in West

Appendices ......................................................................................... 351

Life sketch of Sudhir Ranjan Das ...................................................... 353

List of publications of Sudhir Ranjan Das 357

Index ................................................................................................. 363

Balakrishna N., National Institute of Nutrition, Indian Council of Medical Research,

Kanazawa E., Department of Anatomy, Nihon University School of Dentistry at

Satake T., Department of Anatomy, Nihon University School of Dentistry at

Weare delighted to introduce this Volume on Auxology as a mark of homage to

Our heartfelt thanks go also to P. De Liefde, the Publishing Director, Peter

P. Dasgupta and R.C. Hauspie

THE MEASUREMENT OF HUMAN GROWTH: A

1. THE ANCIENT WORLD

There has long been an interest in anthropometry - the science of human

2. THE MIDDLE AGES

4. EARLY MEASURING INSTRUMENTS

5. THE EIGHTEENTH CENTURY

5.1 Military data

5.2 Early growth studies

6. THE NINETEENTH CENTURY

6.1 Statistics and the quantification of society

(Encyclopedia Britannica 1926). Photographs were unreliable - people could change

6.2 Racial classification

6.3 Standard sizes and mass production

6.4 New techniques and instruments