A STUDY OF DIFFERENTIAL FERTILITY
JOHN VAN DYKE SAUNDERS
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
I RIO GRANDE
MAP OF BRAZIL
- 19 50
The materials contained in the following pages are the result of
a long-standing interest in Brazil. This interest was acquired as the
result of many years of residence in that country, and has been stinu-
lated by a concentration in Latin American and Brazilian studies in
my undergraduate and graduate academic career.
I should like to express my gratitude especially, to Dr. T. Lynn
Smith, the Mestre to his Brazilian students, who provided me with the
stimulus to pursue my graduate studies beyond the M.A. level, and who
through friendly guidance and wise counsel, contributed greatly not
only to what merits this study may have, but also to the essential
preparation necessary for its undertaking.
I also owe a debt of gratitude to Dr. Winston W. Ehrmann, and to
Dr. John M. Maclachlan, who at all times have shown a friendly interest
in my work and progress, and from whom I have had much to learn. To
Dr. Raymond E. Crist, Dr. A. Curtis Wilgus, and Dr. Donald E. Worcester,
I am grateful for the interest they have shown in me and for their
To my wife, Julia Vissotto Saunders, I wish to express my
appreciation of the many hours she spent aiding in the various pains-
taking and laborious tasks involved in the preparation of a manuscript
of this kind, and who always had a word of encouragement and good cheer.
Thanks are due as well to Mr. Joe Sardo, who made helpful sug-
gestions for the preparation of maps, and to Mrs. Agnes Philpot, who
undertook the task of preparing this manuscript in final typewritten
An expression of appreciation is due to the University of Florida,
that through its Graduate School provided me with a fellowship that
enabled me to devote full-time attention to the task of preparing this
dissertation; and to the Brazilian Institute of Geography and Statis-
tics (I. B. G. E.), that has been generous in providing me with many
of the essential materials necessary for this study. To ry many friends
on two continents who have lent their moral support to this task, I
am also thankful.
In conclusion, I would like to express the hope that I have been
half as successful in this undertaking as John Graunt, who in dedicating
his Natural and Political Observations Made Upon the Bills of Mortality
to the Right Honourable John Lord Roberts, Baron of Truro, Lord Privie
Seal, and "one of His Majestie's most Honourable Privie Council," said:
Now having (I know not by what accident) engaged my thoughts
upon the Bills of Mortality, and so far succeeded therein as to
have reduced several great confused Volumes into a few perspicuous
Tables, and abridged such Observations as naturally flowed from
then, into a few succinct Paragraphs, without any long series of
multiloquious Deductions, I have presumed to sacrifice these my
small, but first published Labours unto your Lordship...
TABLE OF C01YTEiTS
MAP OF BEAZIL, 1950. . . . ... .. 1
PREFACE . . . . . . . . . . . iii
LIST OF TABLES . . ... . . . .. vi
LIST OF FIGURES . . .. . . . .. . . ix
I. INTRODUCTION . . . . . 1
II. REVIEW OF THE LITERATURE . . . . .. 7
III. THE LEVEL OF THE RATE OF REPRODUCTION AND THE
GROWTH OF THE POPULATION . . . . . 31
IV. RESIDENTIAL DIFFERENTIAL IN FERTILITY . . 44
V. RACIAL DIFFERENTIALS TI FERTILITY .* * .* 79
VI. REGIONAL DIFFERENTIALS IN FERTILITY . . . 121
VII. NATIVITY, OCCUPATIONAL, RELIGIOUS, AND SOCIO-
ECONOMIC DIFFERENTIALS IN FERTILITY . . . 131
VIII. SUMARY AND CONCLUSIONS . . . . . 152
BIBLIOGRAPi ... . . . . . . . 154
LIST OF TABLES
1. Crude Birth Rates for Areas of the World Having
Complete or Nearly Co:r.plete Birth Registration,
1950 . . . . . . . . . . . .
2. Fertility Ratios for Major World Areas Exclusive
of Latin America, and for Brazil . . . .
3. Fertility Ratios for the Various Latin American
Countries . . . . . . . .
4. The Growth of the Brazilian Population, 1900-1950.
5. Percentage of the Brazilian Population Classified
as Urban, Suburban and Rural, by States, 1950. .
6. Fertility Ratios for the Mf.nicipios Containing the
Capital Cities, Brazil, 1950, by States. . . .
7. Sex Ratios Among Specified Age Groups in the tu-
nicipios of Bahia Which Had the Highest and Low-
est Fertility Ratios, 1940 . . . . .
3. Proportions of Females Ma.rried Among Specified
Age Groups in the IMnicipios of Bahia Which Had
the Highest and Lowest Fertility Ratios, 1940. .
9. Fertility Ratios for the Municipio Containing the
State Capital and the Remainder of the State,
1920, 1940 and 1950 . . . . . .
10. Infant Mortality in the Federal District, 1939-
1950 . . . . . . . . . . . .
11. Probability of Death and Life Expectancy for Ages
0, 1 and 5, in the Federal District, 1939-1941,
1950 . . . . . . . . . . . .
12. Infant Mortality in Five State Capitals, 1940-1950
13. The Population of Brazil Classified According to
Color, 1940 and 1950 . .. . . . . . .
14. The Population of Brazil Classified as Black, by
States, 1940 and 1950 . . . . . .
LIST OF TABLES continued
The Population of Brazil Classified as White, by
States, 1940 and 1950
. . . 0 * . *
Fertility Ratios for the Various Color and Resi-
dential Categories in the Population,
1950, by Regions . . . . . . . .
A Comparison of the Magnitude of the Fertility
Ratios for the Color and Residential Categories
of the States and Regions, With the Magnitude
of the Fertility Ratios for the Color and Resi-
dential Categories of the Nation as a Whole,
1950. Brazil = 100 . . . . . . .
Live Births per 100 Women 12 and Over, and
Live Births per 100 Women 12 and Over Report-
. . 92
. . 95
. 0 0
0 5 0
. 0 0
. 0 0
* 5 0
ing Live Girths, Compared With the Fertility Ratio,
by Regions and States, Brazil, 1940.
Brazil = 100 . . . . . . . . . *
Population of Brazil Classified According
to Color, 1950, by States . . . .
Fertility Ratios for the Various Color Cate-
gories in the Population, 1950, by States .
Fertility Ratios for the Various Color Cate-
gories in the Population, 1940, by States .
Fertility in Relation to Race, Rio de Janei-
ro, 1390 . . . . . . . .
Fertility Ratios by Residence and Race,
Brazil, 1950, by States . . . . .
Fertility Ratios by Residence and Race,
Brazil, 1940, by States . . . . . .
Percentages in the Racial Groups, Brazil,
1872, 1900, 1940 and 1950 . . . . .
Fertility Ratios by States, Brazil, 1950 . .
LIST OF TABLES continued
Number of Children Residing With Their Frailies,
According to the Nativity and Occupation of
the Family Head, Comzpared With the Number of
Children Residing With Their Families in the State
as a Whole, Sao Paulo, 1950 . . . . . .
. . . 0 .
. . 0 . .
. . 0 0
. 0 0 . .
Fertility Ratios for Religious Groups,
by States, Brazil, 1950 : . . .
Spiritualists in the Municipio Contain-
ing the Capital as a Proportion of the
Total Number of Spiritualists in the
State, and Fertility Ratios for
Spiritualists, by States, Brazil, 1950 .
The Color Composition of Socio-Economic
Classes in Recife, Brazil . . . .
Comparison of Size of Family in Superior
and Lower Classes for Groups in Brazil .
LIST OF FIGURES
1. Minimum Estimated Under-Registration of Births,
by States, Brazil, 1939-1941 . . . . .
2. The Level of Fertility by Residence and States,
1950 . . . . . . . . . . . .
3. The Distribution of the Fertility Ratio in Sao
Paulo, by Municipios, 1950 . . . . .
4. The Distribution of the Fertility Ratio in Bahia,
by Municipios, 1940 . . . . . . .
5. Origin of Migrants Lodged at the Hospedaria in
Sao Paulo, 1935-190 . . . . . . .
6. Distribution of Estimated Birth Rates in Santa
Catarina, 1940 . .. . . . . .
7. Racial Distribution of the Population of
Brazil, by States, 1940 . . . . . . .
8. The Level of Fertility by Race and States, Brazil,
1950 . . . . . . . . . . . .
The study of the different rates at which various segments of a
population reproduce, is one of the most significant aspects of deno-
graphic analysis. It greatly aids in the understanding of the mahe-up
of a people, and in the prediction of the "demographic" future.
Differentials in fertility appear to have existed for many centuries.
Only in relatively recent times, however, has it become possible to
study those differentials effectively, through the use of quantitative
Brazil occupies approximately one-half of the land mass of the
South American continent. It contains about one half of its population.
One third of the territory, and one third of the population of the
entire Latin American area belongs to Brazil. Latin America is rapidly
gaining in numbers and in world importance. Brazil has been setting
the pace. Therefore, analysis of its population, which represents
approximately one third of the total population of the Latin American
nations, is exceedingly important. To know and to understand the
differentials in the growth of the various segments of Brazil's popula-
tion,is to throw light on many of the social processes now in operation
in that nation, and is to get a glimpse of the composition and charac-
teristics of future generations. This study attempts to analyze, and
in so far as possible, to explain, the principal differentials in the
rate of reproduction of the Brazilian people.
The quantitative mass population data, contained in the 1940 and
1950 censuses of the Brazilian population constitute the major sources
of the material used for this study. Valuable supplementary material
employed is duly acknowledged in the text and in footnotes. Of special
value in many phases of this study has been the material contained in
the revised edition of Brazil: People and Institutions, by T. Ivnn Smith.
The measure of fertility most heavily relied on in the analyses
herein contained is the fertility ratio. Ideally, it is calculated by
relating the number of children under 5 to women in the childbearing
ages (15 to 44),1 and multiplying by a constant (100) to eliminate
awkward decimals, thus*
number of children 0-4
number of women 15-4 1
This measure of human reproduction offers a number of advantages
over the crude birth rate, or the number of live births during a one-
year period per 1,000 inhabitants. It does not depend on birth regis-
tration and it standardizes to a great extent for variations in sex
and age composition. On the other hand, the calculation of the ferti-
lity ratio for certain groups in the population, such as educational
and occupational groups, in which mother and child are likely to be in
lit is necessary in dealing with all Brazilian census data to
use either women 15 to 49 or 20 to 49 in computing the fertility ratio.
The number of children under 5 cannot always be obtained. Wherever
possible, children under 5 and women 15-49 have been used in calcu-
lating this ratio. Otherwise, children under 10 and women 20 to 49
are the basis of its calculation.
different categories, is not feasible.2
The crude birth rate is not used in the analysis of differential
fertility in Brazil, because of its extreme unreliability, caused by
widespread under-registration of vital statistics. The writer could
not express the situation with regard to the under-registration of
births in Brazil, better than the anonymous author of the following
Another difficulty /in determining the rate of reproduc-
tion of the racial groups in Brazil/--and :r.ore l entable
because it results fror. the inertia of man and not from the
nature of things--is caused by the failures of our registration
of births and deaths, that are not paralleled in any other
great nation of western civilization. Such is the seriousness
of these failures, that vital registration statistics do not
allow us to know, even with wide appro::inraticn, the total
number of births and deaths. These can only be estimated by
conjectures, with an anple margin of error. Under these condi-
tions, it would be an absurd prevention to obtain birth rates
or death rates for the different ethnic groups.3
The extent to which the registration of births fails in Brazil,
is graphically portrayed by Fig. 1, which presents minimum estimates of
under-registration of births in Brazil by states for the 1939 to 1941
period. In only one state of the federation is the percentage-of
omissions of birth registration estimated at less than 49. There is no
reason to believe that the situation has improved since 1939-1941, or
that the registration of deaths is very much more complete.
2For a full discussion of the I.Aerits and derrerits of the
fertility ratio, see T. Lynn Smith, Population Analysis, (lew York:
McGraw-Hill Book Company, 1943), pp. 193-193.
3I. B. G. E., Estudos de Estatistira Teorica e Aplicada,
Estatistica De-.ogrrfica No. 1L, Estudos Sobre a I.atalidade e a
Mortalidade no Brasil, (Rio de Janeiro: ServiCo Grafico do I. B.
G. E., 1952), 35.
Percent of Omissions
| t EZj Under 49.2 *,*.
49.2 61.1 1
; 97.2 over
Fig. 1. Minimum Estimated Under-Registration of Births, by States,
Brazil, 1939-1941. Prepared from data in I. B. G. E., Estudos de
Estatistica Teorica e Aplicada, Estatistica Demografica No. 4
Estimativas da Taxa de Natalidade para o Brasil, as Unidades da
Federacao e as Principais Capitais, (Rio de Janeiro: Servigo
Grfico do I. B. G. E., 194 ), p. 22.
A few comments that will illustrate some of the difficulties
encountered in dealing with the Brazilian census data are in order.
Five-year age groups were given for the first time in the 1950 census.
These are provided only for ages 0 to 20. In order to obtain the
number of children under 5 for the states and territories in 1940, it
is necessary to add the children 0 to 9, male and female, who are
literate, illiterate, and "undeclared". Since children 0 to 4 are
excluded from this tabulation, this sum yields the number of children
5 to 9, which may then be subtracted from the sum of males and fe-
males 0 to 9 in order to obtain the number of children 0 to 4. The
absence of cross-tabulations of racial and residential groups by age
and sex makes it impossible to calculate fertility ratios for such
groups as "urban white" and "rural Negro." Less satisfactory sub-
stitute devices must be employed.
The attempt to obtain a base map for a given state for a census
year is likely to be highly frustrating. Base maps for the states are
dated in late 1945. President Getulio Vargas placed a ban on the
creation of new counties or municipios during his first term in office
(1930-1945). Consequently, just as soon as he went out of power, the
nation erupted in a veritable rash of new municipios. In Sao Paulo
alone 99 new ones were created. Thus the 1945 base maps of the states
do not correspond to the division of the states either in 1940 or
1950. This makes it necessary as was the case for Fig. 4, to compare
the lanicipios in the census with those on the base map, and to make
tedious adjustments in order to match the one with the other.
In citing sources, in those instances where the tables have
been compiled and computed by the writer, this fact is indicated by
the phrase "compiled and computed from data in." Sources for
figures that have been prepared by the writer from data presented in
census or other volumes, are prefaced by the phrase "prepared from
data in." The sources for material that has been taken directly from
a volume, omit these prefatory statements.
The territory of Fernando de Noronha, consisting of a group of
small islands off the northeastern coast of Brazil, and which contained
but 581 persons in 1950, is obviously of no demographic significance,
and is omitted from the analyses. The word raunicipio, the approximate
equivalent of the United States county, is not italicized in the pages
that follow, because Of its frequent occurrence.
For the convenience of the reader who is not overly familiar
with the geographical distribution of the political subdivisions of
Brazil's territory, an outline map of Brazil has been placed following
the title page.
REVIEW OF THE LITERATURE
In the following pages the writer attempts a brief review
of some of the principal findings related to differentials in the
rate of reproduction in various parts of the world. The findings
of some pioneering studies are presented first, followed by those
that are pertinent to Europe, Asia, the United States, Latin America
and Brazil. What would be the monumental task of presenting' in detail
the findings of every study having to do with differential fertility
is not attempted. Rather, the writer hopes that the materials in
the following pages will serve as an introduction to the topic of
this dissertation, and will help bring the reader so to speak, up
to the "line of scrimmage."
One of the very earliest of scholars to present materials re-
lating to differential fertility was Giovanni Botero (1540-1617).
His work entitled Della region di stato, libri dieci. Con tre libri
delle causa della grandezza e magnificenza della cita, contains a
number of astute generalizations. Botero believed that fertility is
higher in rural than in urban areas, and that urban growth is prin-
cipally due to rural-urban migration.1
To John Graunt (1620-1674), however, goes a major part of the
1Pitrim A. Sorodin, Carle C. Zimmerman, and Charles J. Galpin,
A Systematic Sourcebook of Rural Sociology, (Minneapolis: The University
of MinnesotalFess, 1930), Vol. 1, rp. 7o-79
credit for laying the foundations of modern demographic theory and
methods. In his Natural and Political Observations Made Upon the Bills
of Mortality, he demonstrated the regularity of the occurrence of births
and deaths. In so doing he laid the foundations of modern demography,
and brought quantitative materials to bear upon the matter of differen-
tial fertility as well as of differential mortality.
The conclusions that he reached regarding differential fertility
are basically sound and are worthy of rather extensive quotation.
We come to shew, why although in the Country the
Christnings exceed the Burials, yet in London they do not.
The general Reason of this must be, that in London the
proportion of those subject to die unto those capable of
breeding is greater then in the Countrey (sic); That is,
let there be an hundred Persons in London, and as many in
the Country; we say, that if there be 60 of them Breeders
in London, there are more than 60 in the Country, or else
we must say, that London is more unhealthful, or that it
enclines men and women more to Barrenness, then the Country,
which by comparing the Burials and Christnings of Hackney,
Newington, and the other Country-Parishes, with the most
Smoaky and Stinking parts of the City, is scarce discernable
in any considerable degree.2
William Petty (1623-1635) in Several Essays in Political
Arithmetic, arrived at much the same conclusions as Graunt con-
cerning rural-urban differentials in fertility. In addition, he
presented materials for Dublin and additional materials for London.3
2John Graunt, Natural and Political Observations Made Upon the
Bills of Mortality, Walter F. Willcox, Ed., (Baltimore: The Johns
Hopkins Press, 1939), pp. 54-55.
3Sorokin, Zimmerman and Galpin, op. cit., pp. 99-100.
Gregory King (1614-1712) in his ;Natur-l Obsrc.r:atios arnd Conclu-
sions Upon the State and Conditions of Engla'd, Ipubish'.: in 1696, cor-
roborated the findings of Graunt and Petty. He further 'de-on-strated
that although a greater proportion is married in London tha in rural
areas, the fertility rer marriage in London vas lower. Arthur Young
in his The Far-er's Letterc to the People of England (176), presented
statistical data to show that urban fertility is hMghe.r t-an rural. :
Johann Peter Sussrilch (1707-1767), in Die gottliche Ordnung in
den Verranderungen des r-n.schlic..n Geschlechts, etc., reached "much the
same conclusions as his predecessors, but argued that the number of
births is about the same in u ,rban and rural families. The rural-urban
differential in fertility, he attributed to higher marriage rates in
Inasrz-ch as the first studies to be raade of differential fertility
took place in Europe, contemporary findings of demographic studies rela-
ted to differential fertility in that continent, are examired next.
A. M. Carr-Saunders, used materials furnished by the Registrar-
General for En.larj .~ Wales, giving births for 1911 classified
according to the occupations of the father, for the study of socio-
econor ,c 'ifferentials in fertility He ca.:e to the following
4Ibid., p. 101
5Ibid., p. 107.
Ibid., pp. 109-11.
All available evidence goes to support this conclusion,
namely, that the higher the social status, the lower is the
fertility, and that the initial difference is not removed by
subsequent differential mortality. This is apparently true
not only for England but also for other countries in which
the economic situation is similar and is a relatively recent
The same author also stated that in so far as is known, the upper
classes during the Middle Ages were the more fertile, whereas in the
later days of Greece and Rome, the position seemed to be reversed.
Additional support was given to the conclusion reached by
Carr-Saunders by Reuter, who presented data for Paris, Berlin, Vienna
and London, giving birth rates per 1,000 women in the reproductive ages
according to the quality of housing. Persons living in "Very Poor
Quarters" were found to be more fertile than those living in "Poor
Quarters," and these in turn more prolific than those housed in
"Comfortable Quarters," and so on through to "Very Rich Quarters."
The inhabitants of the latter type of housing were found to be the
least fertile. The consistent decline in fertility from the lowest to
the highest type of housing is evidenced in all four cities for which
data was given.9
Further confirmation of the above stated views came from Wright,
7The Population Problem, A Study in Human Evolution, (Oxford:
Clarendon Press, 1922), p. 317.
Ibid., p. 457
9Edward Byron Reuter, Population Problems, (Philadelphia:
J. B. Lippincott Company, 1923), p. 241.
with some qualifications however, that author stated that although
fertility is generally inversely associated with social class and
income, people in certain industries do not conf.Orm to this gradation.
"Textile workers for example, have as few children as the middle
classes _in England /; while miners have the largest number of all--
more than unskilled laborers." The different rate of reproduction
among the social classes was attributed at least in part by that
author, to postponement of marriage among the upper and middle classes,
because of the social pressures requiring greater earnings of the
person with middle or upper class status.
An unskilled or partly skilled workman earns as much
at twenty-one as he is likely to earn when he is forty...
but a lawyer is seldom in receipt of sufficient income
from his profession to maintain a wife and family until he
Other findings pertaining to differentials in the rate of
reproduction in the United Kingdom, are in the Registrar-General's
study (1911) and were reported by Lorimer and Osborn. That study
presented data on the fertility of marriage in relation to occupation,
making use of five broad social classes, with three intermediate ones,
a total of eight. These data reveal that
...even in marriages formed shortly after the mid-point
of the Nineteenth Century marital fertility was somewhat lower
for Class I (professional, business and clerical workers) and
for Class VI (textile workers) than for other professional
groups. But in the more recent marriages, these two classes
show deviations from the general average of all marriages that
10Harold Wright, Population, (New York: Harcourt, Brace and
Company, 1923), p. 114.
are twice as large as the deviations observed in the earliest
marriages included in this series.1
Lorimer and Osborn also supplemented with more recent data,
materials first presented by Heron, who correlated the number of legiti-
nate births per 100 married wo.-en with indices of social status for 1351
and 1901. The results indicated that "at least in the city of London
there was a well defined negative correlation between economic status and
fertility as early as 1851."12
Innes made use of new data contained in the Registrar-General's
Decennial Supplement for 1931 to analyze birth rate differentials by
occupation for 1921 and 1931 in England, and to compare results with
those obtained from the 1911 report. After adjusting the birth rates to
eliminate age differences, he found that whereas in 1921 an inverse status-
fertility association existed for the two highest classes (I and II), in
1931 the opposite was true.13
Contrary to most findings, Edin and Hutchinson discovered in
Stockholm a reversal of the usual inverse association between social class
and the rate of reproduction. The higher the socio-economic status, the
larger the family. These investigators obtained the number of live births
occurring in each family during the first ten years of marriage. They
11Frank Lorimer, and Frederick Csborn, Dynamics of Population,
(New York: The Macmillan Company, 1934), p. 79.
12Ibid., loc. cit.
13j. W. Innes, "Class Birth Rates in England and Wales, 1921-
1931," The Milbank Memorial Fund Quarterly, Vol. XIX, No. 1, (January,
classified the families according to income, educational status, and
age of wife at marriage. The number of live births was found to
increase as educational status increased, and as income increased.
Myrdal, however, reported that this situation is not typical
of Sweden as a whole:
--except for Stockholm and some other cities with a
very low level of average rerrcduction--in general a strong
negative correlation /exists_/ between income and social
status, on the one hand, and number of children on the other.
Thus, the majority of families with a great number of children
are concentrated in the lower income brackets.15
He also offered an explanation of the lower fertility of the lower
socio-economic classes in Stockholm:
Two things must be borne in mind...: first, the fact that
it is a question of differential cost--a child or another child
presses down the level of living which otherwise would be
possible in all income strata; and second, and most important,
that rational birth control is still spreading in society. The
practice is sinking down through the social structure, i.e.,
starting in the higher sttata and slowly making its way down
to the lower strata. The final outcome--a positive correlation
between fertility and economic status--is laVrady visible in
some Swedish cities, particularly Stockholm. o
This explanation implies that when complete control over births
is achieved by all the social strata, the number of children will be
14Karl A. Edin and Edward P. Hutchinson, Studies of Differen-
tial Fertility in S-eden, (London: P. S. King and Son, Ltd., 1935),
pp. 61-73. Cited in Warren S. Thompson, Population Problems, (New
York: :cGraw-Hill Book Company, Inc., 1953), pp. 191-94.
15Gurnnar Myrdal, Population, A Problem for Democracy, (Cambridge:
Harvard University Press, 194C), p. 47.
16Ibid., p. 197.
directly related to the parent's ability to provide for then.
Alva Myrdal holds to iuch the same view. Family planning ac-
cording to that author, first developed in upper-class groups. "It is
not the difference in fertility between different social groups that
needs to be explained, but the different time order and velocity in
the acceptance of birth control by these groups." Eventually, "as
the broad masses come into the swirl of changing conditions," they
too will become sophisticated and class differences in fertility will
That Sweden like other nations of the world possesses lower urban
than rural birth rates was demonstrated by Thomas. However, it is
interesting to note that the differential was reversed in the past:
The crude urban birth rate shows the following relation-
ship to the crude rural birth rate: lower from the 1320's
through the 1750's, higher from the 'sixties through the
'eighties, approximately equal during the last decade of the
19th and the first of the 20th century, and lower again in
the following decades. During the decade of the 1920's the
differential became more marhed than at any previous
decade, the ratio of urban to rural rates being only 73.
In France, a study which.used 1906 census materials revealed an
inverse association between socio-economic status and fertility. This
was found to be true whether occupation or income were taken as a measure
17Alva Myrdal, nation and Family, the Sw edish E rrimnt in
Democratic Family and Porlation Policy, (Iew York: Harper and
Brothers, 1941), p. 62.
1-'Dorothy Swaine Thomas, Social and Eccnor-c Aspects of
Swedish Po.plation Movements, 1750-1933, (Hew York: The Macmillan
Company, 1941), p. 23.
of socio-economic status.19 More recently, Landry came to the sa.e con-
clusion in a study of 20 arrcndissements of Paris, classified accord-
ing to the level of living of the inhabitants of each one. Landry
also reports a sizable rural-urban differential in fertility in France,
with the urban population reproducing at a slower rate than the rural.
The foreign-born were found to have slightly higher birth rates than
the native-born.20 A map presented by Huber, Bunle and Boverat, indicates
that fertility is highest in France's most rural provinces and lowest in
her urban areas.2
D. V. Glass in discussing data relating to differential fertility
in Germany (Wurttemberg) for the years 1934 and 1935, pointed out that
workers had higher birth rates than officials and the general salaried
class. Furthermore, he indicated that a considerable increase in ferti-
lity occurred from 1934 to the first quarter of 1935, and stated:
In so far as it is possible to compare the whole of 1934
with the first quarter of the next year, the rate of increase
is greatest among the salaried class and not among the workers
...Looking more closely at the general increase in births, we
find that the decrease in abortion has probably been a very
19Hannibal Gerald Duncan, Race and Population Problems, (NTe
York: Longmans Green and Company, 1929), p. 31-.
20Adolphe Landry, Traite de D?2rographie, (Paris: Payot,
1949), pp. 304-303.
21Michel Huber, Henri Bunle, and Fernand Boverat, La Population
de la France, Son Evolution et Son Perspectives, (Paris: Librairie
Hachette, ca. 1939), p. 134.
22D.V. Glass, The Struggle for Poiplation, (Oxford: Clarenden
Press, 1936), pp. 22-29.
Thompson in discussing data relating to diff rential fertility
in Germany, reported higher fer ility an.onr Catholic than among
Protestant women, and the usual inverse relationship between socio-
economic status and fertility. He also reported that families that
owned land had a larger num-ber of children than those that did not.
Lest this be thought to be a reflection of the rural-uroan differen-
tial in fertility, he stated:
Thus even in the professions cnly 9.7 percent of the
women were childless if the couple o-wned land, while 15.2
percent were childless if the couple did not own land.
Moreover, 53.5 percent of the owning couples had three or
more children, while only 39.7 percent of the noncw.ming
couples had three or :-or2 children. The differences were
proportionately mrch the same for other occupational
Relatively little attention has been paid to the analysis of
the population of the Asiatic nations. This is in no small measure
due to the scarcity of the necessary materials. Less attention still
has been given to differentials in the rate of reproduction of various
segments of the population of Asiatic countries. A few studies may,
however, be cited.
One of the most important of these is thit of Kingsley Davis,
who in his analysis of the Population of India and Pakistan, devotes
.considerable attention to the matter of differential fertility.24
2Thompson, o. cit., p. 139
24Kingsley Davis, The Population of India and Pakistan, Prince-
ton, New Jersey: rinccton Uriversift Press, 1951), rp. 7C-'2.
Davis finds that the familiar residential differential in fertility
holds good in India, between urban and rural areas, and between larger
and smaller cities. The more rural areas and the smaller cities have
higher fertility than the more urban areas and the larger cities. He
also found that fertility decreases as class status (measured by caste
affiliation), increases. Fertility was found to be greater arong
agriculturalists than an-.on trading and professional peoples. Within
the agricultural category, owners and tenants were found to have low
fertility, and farm laborers to have high fertility. Literacy was
also found to be inversely correlated with fertility. Muslims were
found to be more prolific than Hindus. Davis's main conclusions were
(1) A consideration of rural-urban differentials...
indicates that the cities have not yet, as in soie other
countries, inaugurated a general decline in the birth rate.
(2) An analysis of caste fertility on the basis of
racial status, occupation, and literacy shows that in the
Indian region, as in most other areas, fertility is inversely
correlated with social position. hBt the explanation does
not apparently lie primarily in the use of contraceptives
by the higher castes, as a ,'esterner might expect, but rather
in the fact that non-marriage, especially in the form of widow
celibacy is much higher in the upper classes...
(3) The religious differentials also demonstrate the
role of non-marriage in controlling the Indian birth rate.
The superiority of Muslim and Tribal fertility over the
Hindu is due in part to their greater toleration of widow
A study of education in relation to size of family made in
China by Griffing revealed on the other hand, a positive correlation
between educational status and fertility.
25Ibid., pp. 1-82.
The evidence...accumilated indicates that for the region
studied, education of parents results in increase rather t'-n
decrease in the nur.Iber of living children. The groups from
the highest cultural levels tend to leave the most survivors
and reproduce substantially more than their own number.20
These findings were corroborated by Lamson, who also found that in
China fertility increases as socio-economic status increases.27 Data
collected by the University of Nanking revealed a slight positive
association between fertility and the crop area of farrs.
A study of the Japanese population by Taeuber and Beal revealed
that in that nation rural-urban differentials in fertility are very
sharply defined. "Gross reproduction rates for prefectures in 1930-
1931 portray in accentuated form the urban-rural and agricultural-
industrial patterns of regional differentials..."29
The familiar rural-urban differential in the rate of reproduction
is also evidenced in the Near East. Jurkat reports that "...we find that
low birth rates--and in the Near East we consider as low those rates
26J. B. Griffing, "Education and Size of Family in China,"
The Journal of Heredity, Vol. XVII, No. 9, (September, 1926), p. 337.
27Herbert D. Lamscn, "Differential Reproduction in China,"
The Quarterly Review of Biology, Vol. X, No. 3, (September, 1935),
pp. 303-21. Cited in Frank W. TNotestein, "Class Differences in
Fertility," The Annals of the American Academy of Political and Social
Science, Vol.TCL~VTT. (November, 1936), p. 26.
23Notestein, op cit., loc., cit.
291rene B. Taeuber, and Edwin G. Beal, "The Dynamics of Popu-
lation in Japan," in Demographic Studies of Selected Areas of Rapid
Growth, Proceedings of the Round Table on Population Problems, ("ew
York: ilbank Memorial Fund, 1944), p. 20.
below 50-- ...mainly in the urban type of high-density areas."3
Kiser, writing about Egypt, reported on the other hand that "There
appears to be no substantial rural-urban differential in fertility and
mortality except in so far as the largest cities /Cairo and Alexan-
dria/are concerned."31 Kiser also stated that in Egypt religious
differentials in fertility are great, fertility being highest for
Moslems and lowest for Jews, with Christians occupying an intermediate
position. He gave the concentration of non-Moslem elements in Cairo
and Alexandria as a possible cause of low fertility in those areas.32
Population studies have been greatly developed in the United
States. This development has been aided by the availability of decen-
nial census reports, of which there is a total of 17.
One of the earliest studies to devote attention to differentials
in the rate of reproduction in the United States was made by Tucker,
who, it may be said in passing, made use in some of his analyses of
the index that has since come to be known as the fertility ratio.
30Ernest Jurkat, "Prospects for Population Growth in the Near
East," in Demographic Studies of Selected Areas of Rapid Growth,
Proceedings of the Round Table on Population Problems, (LeT York:
Milbank: Memorial Fund, 1944), p. 87.
31Clyde V. Kiser, "The Demographic Position of Egypt," in
Demographic Studies of Selected Areas of Rapid Growth, Proceedings
of the Round Table on Pcyulation Problems, (Hew York: Milbank
Memorial Fund, 1944), p. 111.
32Ibid., pp. 111-13.
Tucker showed in 147, that whites reproduced more slowly than slaves,
and stated in addition that "The census...offers persuasive evidence
that the natural increase of the free portion of the colored popula-
tion is less than that of the slaves."33
Raymond Pearl who used a largely biological approach to dif-
ferentials in fertility, noted that foreign-born women in America are
much more fertile than native born women, but that their descendants
soon "slough off this attribute of barbarity, as it is regarded by some."34
Thompson tentatively arrived at the same conclusions using 1920 census
data, and explained the major factor responsible for the differential
by saying that
The foreign-born have a relatively unrestricted birth
rate not because they are foreigners nor because they are of
certain nationalities, but because they are less urbanized,
even though living in cities, than most of our native popula-
tion. The foreign-born, both men and women, are dominated
to a great extent by the attitudes toward life brought with
them from their rural conrrunities in the "old country."35
In further analysis Thompson demonstrated that fertility is inversely
associated with size of urban center, and that the fertility of foreign-
born women as well as of native white women is affected by this factor.36
33George Tucker, Progress of the United States in Population
and Wealth in Fifty Years as "xhibited by the Decennial Census, (Iew
York: Press of Hunt's Merchants' Magazine, 143), pp. 96-97.
34Studies in Human Biology, (Baltimore: Williams and Wilkins
Company, 1924), pp. 206-10, 253-72.
35Warreh S. Thompson, Ratio of Children to Women, 1920,
Census Monograph XI, (Washington: United States Government Printing
Office, 1931), p. 37.
36Ibid., pp. 36-45, 96-107, 114-26.
He also presented fertility ratios for agricultural states, semi-
industrial states, and industrial states, for census years from 1'00
to 1920. These data clearly show that as early as 1I00 a differential
existed at least in the net reproduction rate if these groups of states,
with the agricultural states reproducing at the fastest pace, and the
industrial states at the slowest rate. The differential is found in
all the census years for which data was given.37
Brunner and Kolb made an interesting analysis of rural-urban
differentials in fertility, from the standpoint of distance from an
The data for Des Moines show that this Lfertility/ratio
is lowest in the City County and that it increases with each
succeeding tier out from the City County. There appears to be
a relationship between the birth rate and the distance from
the city of any measured unit of population such as the
people of a group of counties.
Furthermore...this relationship existed in the past.
It was true in both 1910 and 1920.33
Lorimer and Osborn analyzed the fertility of immigrant stocks,
and noted that French Canadians are the most prolific, being a
group of rural and orthodox Catholic origin, which suffered little
cultural change at first in drifting across the Canadian border into
New England towns.39 Another Catholic and rural immigrant stock, the
Irish, also were found to have high fertility, whereas the English
and Scotch were among the least prolific of the immigrant stocks.
37Ibid., p. 15.
3 Edmund de S. Brunner and J. H. Kolb, Rural Social Trends,
(New York: McGraw-Hill Book Company, 1933), p. 115.
39_p. cit., p. 50.
These authors also presented data from the 1910 Milbank Sample of
United States Census Data, that showed that fertility decreases in
urban areas as one goes from the unskilled manual wor-er occupational
category to the professional occupational category: and in rural areas,
from the farm laborer to the farm owner category.40 In a further
discussion of social class differentials, Lorimer and Osborn stated:
Social class differentials were found to be surprisingly
similar in different types of communities, in an analysis of
variations in fertility of marriage in 1900 among social
classes in one metropolitan area (Chicago), in a group of cities
ranging from approximately 50,000 population, in a group of villages
with few.r than 5,000 inhabitants, and in rural farm communi-
An inverse association between education and fertility and be-
tween occupational status and fertility, was found by Notestein. The
latter comparison held good even when the comparison was made between
similar educational groups. With regard to religious differentials,
Catholics were found to be more fertile than Protestants, and the dif-
ferential was discovered to be more accentuated in the upper than in
the lower classes. Both religious groups exhibited the characteristic
inverse association between status and fertility.2
Whelpton found the usual differentials in fertility associated
with nativity, occupation and religion, and in addition, further
OIbid., p. 67
41Ibid., pp. 83-84.
42Frank W. Notestein, "Class Differences in Fertility,"
The Annals of the American Acade.n of Political and Social Science,
Vol. CLDOVIII, (:Nover.ber, 1936), pp. 26-36.
analyzed differences in fertility as distance from a city increases.
The results show that generally, fertility increases as distance from
a city increases.h3
An intensive study of socio-economic differentials in fertility
was made by Kiser. This study reinforced previous conclusions re-
garding socio-economic differentials. Notestein in his study of
fertility in the East North Central states, found the usual inverse
association between fertility and size of cornunity, and between socio-
economic status (as measured by the value of the hone), and fertility.
He pointed, however, to some interesting exceptions:
In one case, that of the native white families in large
cities, the average nun-cr of children was smaller in families
with homes valued at less than $1,00O than in those whose
homes were worth from $1,C00 to $2,000...
At the other end of the scale the average numbers of
children were frequently larger in families with homes
valued at $10,000 or more than in those of the $5,C00 to
$10,OX) class... The I r.c or ;, direct associaticrn between
fertility and economic status in some of the upper economic
groups is becoming clear.-5
Notestein is inclined to accept provisionally as an explanation of
the phenomena that he noted, the beginning of a reversal of the more
43p. K. Whelpton, "Geographic and Economic Differentials in
Fertility," The Annals of the Amrican Academy of Political and Social
Science, Vol. CLC~VIII, (Novec.'.ber, 1936, pp. 37-55.
4Clyde V. Kiser, "V:riations in Birth Rates According to
Occupational Status, Family Income, and Educationsl Attainment,"
The Milbank Merorial Fund Quarterly, Vol. XVI, No. 1, (January, 1933),
45Frank W. Notestein, "Differential Fertility in the East North
Central States," The Milbankh .e-orial Fund Quarterly, Vol. XVI, No. 2,
(April 1933), p. 1i-
usual inverse association between fertility and socio-economic status
such as has been observed in Stochholm and a few other European cities.
Wilcox, analyzing data for 1930, discovered contrary to previous
findings, that native white wives were more fertile than the foreign-
born. However, by 1930, a large proportion of foreign-born wives were
in the advanced age groups. Standardizing the data for age, the figures
were reconciled with previous findings, and the foreign-born were shown
to be more fertile than the native-born.46 Willcox also noted that from
1890 to 1930, the fertility of foreign-born wives decreased much faster
than that of native-born wives.
In a study of data gathered by the National Health Survey in the
fall and winter of 1935-1936, Kiser discovered the usual inverse rela-
tionship between income and fertility, but concluded that the amount of
income was more closely related to fertility than occupational status.
The study of fertility rates in Kentucky made by Oyler, revealed
that high school attendance tends to reduce fertility, but more inter-
estingly, large urban centers were found to have a depressing effect
46Walter F. Willcox, Studies in American Demography, (Ithaca,
New York: Cornell University Press, 1940), pp. 274-75.
17Ibid., p. 281.
4Clyde V. Kiser, "Intra-Group Differences in Birth Rates of
Married Women," The Milbank Menorial Fund Quarterly, Vol. XIX, No. 2,
(April, 1941), pp. 147-17C. National Health Survey data are analyzed
in greater detail, in book form by Kiser in Group Differences in Urban
Fertility, A Study Derived From the ITational Health Survey, (Baltimore:
The Williams and Wilkins Company, 1942).
on the fertility of adjacent areas. The author also made the follow-
ing suggestive statement:
There is also an indirect relationship between highway
construction and decline in population fertility. The more
highway traffic passe through a county, the sooner the
birth rate declines. '
Beegle and Smith in their study of differential fertility, found
significant variations in the fertility of residential groups, with
the rural-farm population being the most prolific, the urban the least
prolific, while the rural-non-farm category occupied an intermediate
position. "This relationship prevails," the authors stated "among
both whites and Negroes, in both French and Anglo-Saxon Louisiana, and
in each of the five type-of-farming areas."50 An inverse relationship'
between the size of an incorporated center and fertility was also found.
Beegle and Smith noted as well, racial differentials in fertility, in
favor of the Negro, but suggested that the differential is largely due
to the greater concentration of Negroes than whites in rural areas.
Another differential noted is that between different ethno-religious
The population of French-Catholic south Louisiana is
multiplying far more rapidly than that of Anglo-Saxon,
1 Merton D. Oyler, Fertility Rates and Migraticn of Kentucky
Population, 1920 to 19O0, as Related to Ccr.---n: cation, Income and
Education, Reprint of Kentucky Agricultural Ex.periment Station Bul-
letin 469, (November, 1944), p. 26.
50J. Allan Beegle and T. Lynn Smith, "Differential Fertility
in Louisiana," Louisiana State University Agricultural EIperinen;
Station, Louisiana Bulletin No. 403, (June, 1946), p. 24. See also
T. Lynn Smith and Homer L: Hitt, The People of Louisiana, (Baton Rouge:
Louisiana State University Press, 1952), chap. XI.
Protestant north Louisiana. This difference prevails among all
residential groups of both races. Except for the rural-urban
differential it is the most significant difference observed.51
Smith, in Population Analysis devoted a full chapter to dif-
ferential fertility, which contains the following significant statement
in relation to regional differences in fertility:
Other than these reflections of the urbanity of the rural-
ity of the populations, there seems little to be said about
regional differences. In fact, the results of the analysis in
this and the two preceding sections may be sur.-ed up as follows:
Every refinement introduced into the analysis of the data con-
cerning the net reproduction rate in the United States tends to
emphasize the importance of the rural-urban differential in fer-
tility and to minimize racial and regional differences in the
rate of reproduction. When the residential differentials are
fully isolated, there is little if anything left to be attributed
to other factors.52
Demographic studies have progressed but slowly in Latin America.
As in Asia, this has been in part due to the lack of recent census
materials for many of the countries. Most analyses of the population
of the nations belonging to this area, have been made by scholars from
other parts of the world. It is to be hoped, however, that the in-
creasing availability of adequate and recent census tabulations for the
various Latin American nations will result in the appearance of up-to-
date studies making use of modern demographic analytical techniques.
One of the more significant studies of Latin American nations is
that made by Whetten of Mexico. In his discussion of marriage and
51Ibid., p. 5.
52T. Lynn Smith, Population Analysis, (New York: McGraw-Hill
Book Company, 1943), p. 218.
family institutions he brought data to bear on the matter of differen-
tial fertility. These data indicate that Mexico is no exception to the
general rule of lower fertility in urban than in rural areas.53
A study by Moore threw additional light on differential fertility
in Mexico. The study was based on interviews of 600 factory workers
in Puebla, in 1943-49. Although the sample is relatively snall, the
results of the inquiry are interesting and worthy of mention. Moore
found that fertility is significantly higher among couples who had
been united by religious rites (and are thereby considered by the author
to be "conservative'),than among couples who were united only by a civil
ceremony or by no ceremony at all ("liberal"). He also found, and this
is of particular interest, that fertility was directly associated with
income even when age was taken into consideration. Furthermore, the
income differentials were within an occupational class.5
Hatt presented some interesting conclusions resulting from a
study of fertility in Puerto Rico. He found an inverse association be-
tween fertility and socio-economic status, and between fertility and
urbanity, and concludes that 1920 represents the approximate date when
these "demographic effects usually attributed to industrial development
were first felt in Puerto Rico," inasmuch as these differentials did
53Nathan L. Whetton, Rural Mexico, (Chicago: The University
of Chicago Press, 1943), pp. 390-93.
5hWilbert E. Moore, "Attitudes of Mexican Factory Workers
Toward Fertility Control," in Approaches to Problems of High Fertility
in Agrarian Societies, Papers Presented at the 1951 Annual Conference
of the Milbank Memorial Fund, (iTew York: Milbank Memorial Fund, 1952),
not occur among those -arried before that date. Another conclusion
reached by that author is that fertility is negatively associated
with colored (de color) persons. However, he suggests that this may
be due to urban residence. He also found no relationship between
Catholicism and fertility as might be expected, and stated:
The second deviation from "common sense" judgments, the
fertility differential between consensual unions as opposed to
legal marriages, shows the letter to be generally and sub-
stantially the more fertile.55
In his analysis of data for Argentina, Taylor noted that
The higher birth rates and the higher infant death rates
are in the most rural provinces. One of these facts offers
the principal explanation of the other. But in addition to
this, in the rural and peripheral provinces are a far higher
percentage of the population with Indian heritage, a higher
percentage of the economically poor who live in poorer houses,
have less medical care, and are undoubtedly more ignorant of
proper child care.56
Mention of a rural-urban differential in fertility in Uruguay
was made by Solari who shows that in 1908 (the date of the last census)
fertility was much lower in urban than in rural areas. He concluded
that although more detailed study is impossible due to the lack of
recent census materials, every indication points to the conclusion
that the differential exists today.57
55Paul K. Hatt, Backgrounds of Hunan Fertility in Puerto Rico,
(Princeton, New Jersey: Princeton University Press, 1952), p. 333.
56Carl C. Taylor, Rural Life in Argentina, (Baton Rouge:
Louisiana State University Press, 194V), p. 83
57Aldo E. Solari, Sociologia Rural Nacional, (.Montevideo:
Biblioteca de Publicaciones Oficiales de la Facultad de Derecho y
Ciencias Sociales de la Universidad de Montevideo, 1953), p. 121.
Interesting material was presented by nelson, relating to racial
differentials in the rate of reproduction :-: Cuba. In that country,
although fertility ratios for the "colored group (which includes
Negroes, Chinese and r-ixed-bloods) are hig h- than the fertility ratios
for whites, the Negro and mixed-blood polplations have constituted
an ever decreasing proportion of the total --opulation. From 1931 to
1943 the decline in the relative importance of Tegroes was greater
than the same decline for mixed-bloods. 'iT apparent contradiction
of these facts is tentatively explained by a-aga.tion of the races.5
Studies of differential fertility in Brazil have been handi-
capped, as in most Latin American Countries by lack of data. The 1940
census, for example, for the first time since 1390 presented data re-
lating to the racial composition of the population. That census
(published in 1950), rnd the appearance of the principal results of
the 1950 census have made possible a number of studies of differentials
in the fertility of the Brazilian population.
These are to be found in two main sources: a series of publications
58Lowry Nelson, Rural Cuba, (Minneapolis: The University of
Minnesota Press, 1950), pp. 27-31.
of the Brazilian Institute of Geography and Statistics,59 and in the
two editions of Brazil: People and Institutions, by T. Lynn Smith.
The first of these sources unfortunately relies almost exclusively on
1940 census data which the present writer considers unreliable, and
which he discusses and analyzes in Chapter VII. In the second of these
sources, the author demonstrated that the usual rural-urban differen-
tial in fertility exists in Brazil, and that the INegro segment of the
population is reproducing more slowly than either the white or mixed-
blood segments, and that the population is undergoing a "whitening"
process.60 These conclusions are supported by the more intensive
analysis based on more complete data which is presented in the fol-
591. B. G. E., Estudos de Estatfstica Tedrica e Aplicada, Esta-
tistica Denografica: No. 4, Estimativas da Taxa de Natalidade para o
Brasil, as Unidades da Federago e as Principais Ca(19i-); o. 6,
A Prolificidade da Mulher, Segundo a Idade Inicial da Atividade Reprodu-
tora, no Brasil (1949); o. 3, Caracteristicas Demogr~ficas do Estado da
Bahia (1949); No. 10, Pesquisas Sobre a Natalidade no Brasil (1950);
No. 14, Estudos Sobre a Natalidade e a M.ortalidade no 3rasil (1952):
No. 15, Estudos Sobre a Natalidade em A!guwas Grandes Cidades do Brasil
(1953; No. 16, Pesquisas Sobre a Natalidade no Brasil, 2a. Sdrie (1953);
(Rio de Janeiro: Servigo Grafico do I. B. G. E.).
60T. Lynn Smith, Brazil: People and Institutions, Revised
Edition (Baton Rouge: Louisiana State University Press, 1954), pp.
THE LEVEL OF THE RATE OF REPRODUCTION AND THE
GROWTH OF THE POPULATION
Before proceeding to the analysis of differentials in the
fertility of Brazil's population, it is well to examine some inter-
national differentials in the rate of reproduction, and to corp.are
them with the rate of reproduction in Brazil. In this way, one can
obtain a better idea of the magnitude of that rate in Latin America's
Such a comparison is made possible if we examine the crude
birth rates, or the number of live births during a one-year period,
per 1,000 inhabitants, in various areas of the world. In Table 1,
in addition to the estimated birth rate for Brazil, are presented
birth rates for those areas of the world believed by the Statistical
Office of the United Nations to have complete or nearly complete
This tabulation offers confirmation of the impression that
the birth rate in Brazil is at a very high level. Only three of the
53 areas of the world for which reliable birth rates were available
in 1950, possessed a higher birth rate than that of Brazil (43.6).
These were Mauritius, Mexico, and Singapore. Of particular signi-
ficance is the fact that some of the world's most rural areas, that
can therefore be expected to reproduce at a very rapid rate are in-
cluded in the tabulation. In addition to Brazil, only six of the
53 areas of the world included in the tabulation had birth rates
CRUDE BIRTH RATES FOR AREAS OF THE WORLD HAVING
COMPITE OR NEARIM CO1CPLETE BIRTH REGISTRATION,
Area Crude birth rate
Union of South Africa
Non-indigenous population 33.9
South West Africa
United States 23.5
Bahama Islands 32.0
British Honduras 39.4
Leeward Islands 35.5
Windward Islands 37.3
Trinidad and Tobago 37.5
Virgin Islands 33.4
British Guiana 40.4
Federation of Malaya 42.0
TABLE I continued
CRUDE BIRTH RATES FOR AREAS OF THE WORLD HAVING
COMPLETE OR ITEARYI COMPLETE BIRTH REGISTRATION,
Area Crude birth rate
Bulgaria (1947) 24.0
Czechoslovakia (1949) 22.1
German Federal Republic 16.2
Greece (1949) 26.1
Hungary (1948) 19.1
United Kingdom 16.2
New Zealand 24.7
Fiji Islands 39.9
Pacific Islands (U.S.) 35.9
*Source: United Nations, Demographic Yearbook, 1953,(New York; Stata-
tistical Office of the United Nations, 1953), PP. 132-41.
(1) Estimate of the Brazilian Institute of Geography and
If the fertility ratio, or the number of children under 5 per
100 women aged 15 to 44, for the various nations and areas of the
world are compared with the one for Brazil, another, and more satis-
factory gauge of the level of the rate of reproduction in Brazil
is obtained. Such a comparison is made possible by Table 2, which
excludes the Latin American nations. Of the 33 areas of the world
for which these materials are presented, only the Moslem population
of Algeria, the population of the Canadian Yukon and Northwest
Territories, and the population of Formosa, are revealed to reproduce
more rapidly than the population of Brazil. It is significant that
among the areas included, are places such as Portuguese Guinea, India
and Japan, where the rate of reproduction is at a very high level.
Nevertheless, the rate of reproduction of these areas is far out-
distanced by that of Brazil.
Because of the marked under-registration of births in 17 of
the 20 Latin American republics, a comparison between the Brazilian
birth rate and the birth rates of the other Latin republics in the
hemisphere is largely meaningless. However, the fertility ratio,
which does not depend on the registration of vital statistics,
provides an adequate basis for comparison.
Unfortunately, it is not possible to calculate fertility ratios
for a fairly recent date for all of the Latin American Republics. In
Table 3 fertility ratios are presented for 18 of the 20 republics
belonging to that area, and for Puerto Rico, whenever possible for
FERTILIY RATIOS FOR MAJOR WORLD AREAS EXCL_ IVE
OF LATIT AMERICA, ATD FOR BRAZIL*
Fertility Nature of
Area ratio(1) Year data
Yukon and Northwest
TABLE 2 continued
FERTILITY RATIOS FOR MAJOR WORLD AREAS EXCLUSIVE
OF LATIN AMERICA, AM) FOR BRAZIL*
Fertility Nature of
Area ratio(1) Year data
Denmark 45.2 1950 Census
Austria 34.8 1951 Census
Belgium 38.0 1951 Estimate
Finland 54.4 1950 Census
France 46.1 1953 Estimate
German Federal Republic 30.1 1950 Census
Iceland 59.1 1950 Census
Norway 44.7 1950 Census
Portugal 44.9 1950 Census
Sweden 40.0 1950 Census
England and Wales 39.2 1951 Census
Northern Ireland 46.2 1951 Census
Scotland 41.7 1951 Census
Switzerland 39.1 1950 Census
Yugoslavia 49.5 1952 Estimate
New Zealand 55.8 1951 Census
Australia 50.8 1952 Estimate
*Source: Compiled and computed from data in United Nations, Demographic
Yearbook, 1953, (New York: Statistical Office of the United
Nations, 1953), pp. 114-130.
(1) Fertility ratio determined as follows: children 0-4 X 100
FERTILITY RATIOS FOR THE VARIOUS IATI~ A)IRICAN COUNTRIES*
Country Fertility ratio(1) Year
Brazil 70.5 1950
Argentina 46.8 1947
Chile 52.3 1940
Colombia 67.6 1938
Costa Rica 73.9 1950
Cuba 57.4 1943
Dominican Republic 59.8 1950
Ecuador 76.5 1950
El Salvador 67.4 1950
Guatemala 72.2 1950
Haiti 49.7 1950
Hondu ras 72.7 1950
Mexico 68.4 1950
Nicaragua 70.2 1950
Panama 75.4 1950
Paraguay 75.2 1950
Peru 71.5 1940
Puerto Rico 78.4 1950
Venezuela 76.7 1950
*Sources: Figures for Panama, Dominican Republic, Haiti, Mexico, Vene-
zuela, Peru, Guatemala, Cuba, Argentina, Chile and Colombia,
are from the MS of The People of Latin America by T. Lynn
Smith. Figures for Brazil, Honduras, Kicaragua, Costa Rica,
El Salvador, Puerto Rico and Ecuador, compiled and computed
from data in United Nations, Demografic Yearbook, 1953,
(New York: Statistical Office of the United Nations, 1953),
pp. 116-19. Figure for Paraguay com.put~ed from data in
Demografic Yearbook, 1954, p. 126.
(1) Fertility ratio determined as follows: children under 5
The rate of reproduction of the Brazilian population as measuredd
by the fertility ratio (70.5), is greater than that of Nicaragua,
El Salvador, Cuba, Haiti, Mexico, Argentina, Chile, and Coloz-bia.
The population of Honduras, Costa Rica, Panama, Puerto Rico, Ecuador,
the Dominican Republic, Guatemala, Peru, Paraguay and Venezuela, are
revealed to be reproducing at a more rapid rate than that of Brazil.
The great speed with which the Brazilian people is reproducing
itself, is also reflected in the terrific rate at which the population
is increasing. Although immigration has contributed somewhat to the
growth of the Brazilian population, the overwhelming factor respon-
sible for that growth is the high reproduction rate.
The increase of the population in absolute and relative num-
bers, from 1900 to 1950, and from 1920 to 1950, is given in Table 4.
From 1900 to 1950, the total population of Brazil increased from
17,433,434 to 51,944,397 inhabitants, an absolute increase of
43,505,963, and a relative increase of 193 percent.
Greatest increases are to be found in the states comprising the
Southern region, that have benefited the most from immigration. From
1900 to 1950, the population of Rio Grande do Sul increased by 262
percent, and that of Sao Paulo by 300 percent; Santa Catarina's
population registered and increase of 337 percent, and the people of
Parana increased by no less than 547 percent.
Outside of the Southern region, the greatest increases in
population have occurred in the two states comprising the West Cen-
tral region (Mato Grosso and Goias), which have been the goal of
THE GROWiH OF THE BRAZILIAN POPULATION, 1900-1950*
Region and State 1900 1920 1940 1950
R. G. do Norte
Rio de Janeiro
R. G. do Sul
TABIE 4 continued
THE GROWTH OF THE BRAZILIAN POPULATION, 1900-1950"
Region and State Absolute Percentage
R. G. do Norte
Rio de Janeiro
R. G. do Sul
TABLE 4 continued
THE GROWTH OF THE. BRAZILIUN POPUL0JION, 1900-1950*
Region and State Absolute Percentage
R. G. do Norte
Rio de Janeiro
R. G. do Sul
Compiled and computed from data in I. B. G. E., Conselho
Nacional de Estatistica, Servigo Nacional de RecenseT.ento,
VI Recenseamento Geral do Brasil, Censo Demografico (1 de
Julho de 1950), Estados Unidos do Brasil, Seleco dos Prin-
cipais Dados, (Rio de Janeiro: Servigo Grfio do I. B. G. E.
1951, 1952, 1953), passim.
(1) Territory. For 1950, the population of the territories
of Amap, Guapore and Rio Branco, are included in the popula-
tion of Para, Mato Grosso, and Anazonas, respectively.
S2) 1906. Results of the 1900 census were rejected.
3) 1906-1950. See (2) above.
considerable internal westward migration. Frcm 19C0 to 1950, the
population of Goias increased by 376 percent, and that of Mato Grosso
increased by 374 percent.
Greatest numerical gains from 1900 to 1950 were made by Sao
Paulo, Minas Gerais, Bahia, Rio Grande do Sul, and Pernarbuco,
listed in the decreasing order of the size of the increase.
Slowest to increase during the fifty-year period preceding
1950 were Alagoas (68 percent), Sergipe (81 percent), Amazonas (113
percent), Minas Gerais (115 percent), and Bahia (123 percent). The
great speed with which the Brazilian population is increasing is
emphasized by the fact that from 1900 to 1950 only two states of the
federation failed to double their numbers. During the same period,
eight of the nation's 22 political subdivisions increased by more than
250 percent. Six of the subdivisions of the nation's territory
(Ceara, the Federal District, Parana, Santa Catarina, Goias, and
Mato Grosso, more than doubled their numbers from 1920 to 1950.
Lowest rates of increase during the 30-year period preceding
1950 were recorded for the states of Alagoas with a percentage
increase of 12, Para (18 percent), Acre (24 percent), Minas Gerais
(31 percent), and Sergipe (35 percent).
On a regional basis, whether the 1900-1950 or the 1920-1950
period be considered, greatest rates of increase are to be found in
the Southern and West Central regions, and smallest population growth
occurred in the Northern region.
If the population of Brazil continues to grow even at a "low"
rate of 2.0 percent per year, it will surpass the 100 million mark
before the end of the century. Since considerable inroads are being
made against mortality, and the death rate is Imown to fall more
rapidly than the birth rate, there may well be 1CO million Brazilians
before the next fifty years have elapsed.
RESIDENTIAL DIFFERENTIALS IN FERTILITY
Perhaps no other differential in the fertility of human popu-
lations has been so oft studied, commented upon and analyzed as
the residential. Some of the first observations ever recorded on
fertility differentials are related to the subject of this chapter.
The different rates at which the urban and rural portions
of a population reproduce are without doubt of great importance.
In practically every human group of any size, men live to a greater
or lesser degree in comparatively small areas with a high popula-
tion density, which are characterized by an urban mode of life.
Conversely, rural areas are also to be found all over the globe.
Such questions, therefore, as the proportions of a given
population living in urban and rural areas, the rate of reproduc-
tion in them, and the rate at which a population is becoming
urbanized, are of paramount importance in attempting to predict the
demographic future of a population, along with all that is implied
in that expression.
As the result of studies conducted at different points in
time and in many parts of the world, one of the most universally
accepted of population theories has been the differential fertility
of rural and urban areas, with the latter possessing lower fertility
than the former. Although the level of fertility is by no means
the same for all urban areas even of comparable size, or for all
country areas even if possessing a comparable degree of rurality,
the lower urban level of fertility holds generally true when compared
with the surrounding rural population.
A number of factors may be postulated to account for the slower
pace at which the urban population reproduces. One that is often
cited is the economic burden that children place upon urban parents
whereas to the rural dwellers children are likely be be an economic
asset. This generalization is particularly applicable in highly
rural "under-developed" nations like Brazil, where a large portion
of agriculture is of the subsistence type and has not advanced beyond
the hoe or rudimentary plow system of agriculture. Methods of birth
control are at hand for the urban dweller to limit his family to the
desired size. Being better educated and more aware of the possibili-
ties of birth control and possessing greater economic resources than
his rural brother, the city man is more prone to adopt family
Another factor is the shortage of men in urban areas or a low
urban sex ratio, which is reflected in lower proportions of the city
population being married. The incentive to marry is smaller in
urban than in rural areas. In the city a bachelor may live comfort-
ably and enjoy many of the pleasures of life on a relatively
moderate income. The city with its many opportunities for diversion
and recreation of the most varied kinds, minimizes and supplants to
some extent the need for and function of a home. In addition, the
acquisition of a wife and the procreation of children places a much
greater financial burden on the city dweller than on the inhabitant
of country districts. Finally, the anonymity of the city makes it
possible for both men and women to engage in sexual intercourse out-
side of the bonds of the family with greater ease.
The Rurality of the Population
When appraising the degree of rurality (or urbanity) of the
Brazilian population, the Brazilian census definition of its three-
fold division of the population into urban, suburban, and rural must
be kept in mind.
The states of Brazil are subdivided into municipios which are
analogous to the United States county, and these in turn are divided
into districts. The census defines as urban the cores of the federal
capital, state capitals, seats of municipios and seats of districts.
The suburban area includes areas immediately adjacent to the urban
core. The federal decree of 1938 which established the criteria
for the delineation of urban, suburban and rural zones, specifies
The delineation of the suburban portion of the seats
of municipios or distritos, shall consist in the clear and
simple description of a line...L on a county map /embracing
an area that surrounds at a variable distance, the urban
section, an area into which the expansion of the urban zone
is already proceeding or to which due to its favorable topo-
graphic conditions this expansion is naturally destined.
The rural population consists of the inhabitants of those areas
that are not included in either the urban or suburban categories.
LT. Lynn Smith, Brazil: People and Institutions, Revised
Edition, (Baton Rouge; Louisiana State University Press, 1954),
In effect, the political function of a locality is the main criterion
for its classification as urban or rural.
This criterion is not altogether far-fetched, since with only
one minor exception the state capital is the largest urban center
in the state, and with probably very few exceptions the seat of the
municipio is always the largest urban center in that political sub-
division, and likewise for the seats of the districts. This prac-
tice, however, does lead to the inclusion in the urban category of
many localities that could hardly be considered urban from any other
standpoint.2 As a result, many hundreds of localities possessing
fewer than 1,000 inhabitants which would be classified as rural in
the United States are in Brazil considered to be urban. If the
United States census criteria of a minimum of 2,500 inhabitants
for urban places were employed, a still larger share of the
Brazilian population now considered urban, would be thrown into the
It becomes evident therefore, that the degree of rurality in
Brazil is actually greater than that indicated by census figures--
a fact that must be constantly kept in mind. In 1950, 25.0 percent
of the population was classified as urban, 11.2 percent as suburban
and 63.8 percent as rural. A more realistic estimate of the rurality
of the population can be had by considering the suburban population
as rural, thus placing 75 percent of the nation's inhabitants in
2Cf. Ibid., pp. 164-67
the rural category. Not only is Brazil highly rural from the quan-
titative standpoint, but it is also highly rural from a qualitative
point of view. It would be difficult to imagine places where a
greater degree of rurality prevails than in the many small hamlets and
villages scattered over Brazil's vast territory.
Examination of Table 5 reveals the high degree of rurality that
prevails throughout the national territory. If the Federal District
(Distrito Federal) a small area almost entirely occupied by the capi-
tal city of Rio de Janeiro, and the territory of Rio Branco possessing
an insignificantly small population are excluded, in only two states
do more than 25 percent of the population live in urban areas.
The three most rural states are to be found in the Northeast
region: Piauf with only 8.0 percent of its population in the urban
category; Maranhao, with 10.5 percent of its inhabitants so classi-
fied; and Ceara which is only 11.8 percent urban. The highest propor-
tions of urban population are to be found in Sao Paulo in the
Southern region (39.9 percent), and in Rio de Janeiro (40.5 percent)
and the Federal District (74.8 percent) in the Eastern region.
In making comparisons between the fertility ratio in the
various states of Brazil, the role played by infant mortality must
be kept in mind. Infant mortality in Brazil is high. It also varies
considerably from one part of Brazil to another. The extent of the
variation may be glimpsed from Table 12 Estimated infant mortality
rates for thirty S9o Paulo municipios, calculated by the Brazilian
PERCENTAGES OF THE BRAZILIAN POPU;0ATION CLASSIFIED
AS URBAN, SUBURBAN, AND RURAL, BY STATES, 1950"
Region and State Percent Urban Percent Suburban Percent Rural
Guapore(1) 20.0 17.4 62.6
Acre(l) 13.3 5.2 18.5
Amazonas 15.6 11.2 73.2
Rio Branco(l) 25.7 2.6 71.7
Para 16.6 18.1 65.4
Amapa(1) 17.8 19.3 62.9
Maranhao 10.5 6.9 82.7
Piauf 8.0 8.3 83.7
Ceara' 11.8 13.4 74.8
Rio Grande do Norte 17.7 8.5 73.8
Parafba 18.3 8.3 73.3
Pernambuco 14.7 19.7 65.6
Alagoas 13.7 12.5 73.8
Sergipe 21.4 10.5 68.2
Bahia 19.6 6.3 74.1
Minas Gerais 18.5 11.5 69.9
Espfrito Santo 15.8 6.8 77.4
Rio de Janeiro 40.5 7.0 52.5
Distrito Federal 74.8 22.0 3.1
Sao Paulo 39.9 12.7 47.4
Paranal 16.6 8.3 75.0
Santa Catarina 17.3 6.0 76.8
Rio Grande do Sul 24.4 9.8 65.9
Mato Grosso 32.4 10.7 65.9
Goias 14.9 5.3 79.8
Brazil 25.0 11.2 63.8
*Source: Compiled and computed from data in I. B. G. E., Conselho
Nacional de Estfstica, Servico Nacional de Recenceamento,
VI Recenceamento Geral do Brasil, Censo Demografico (1 de
Julho de 1950), Selegao dos Principais Dados, (Rio de
Janeiro: Servigo Grafico do I. B. G. E., 1951, 1952, 1953),
Institute of Geography and Statistics (I. B. G. E.), show a variation
of from 93.13 to 254.55 deaths of infants under the age of one per
thousand live births.3 Furthermore, wide variations no doubt occur
in the mortality of the 1-4 age group. As a result, the relation-
ship of the number of children under 5 in the population to the
number of live births is not constant. Nevertheless, the fertility
ratio remains a valuable index of what has been termed "effective
The fertility of persons living in urban, suburban, and rural
areas varies considerably in Brazil. In 1950, in rural areas there
were 77.8 children 0-4 per 100 women 15-49. In suburban areas this
figure was 53.9 and in urban areas only 43.2.
Residential differentials in the rate of reproduction are not
the same in magnitude or importance over the entire area of Brazil.
In an attempt to analyze these variations, a three factor map
of Brazil showing size of population, proportions urban, rural, and
suburban, and the variations in fertility in each of the residential
categories was prepared (Fig. 2), thus permitting a simultaneous
comparison of size of population, proportions in urban, suburban and
rural areas, and the rate of reproduction.
Analysis of this map shows that in all states with the exceptions
of Alagoas and Goias, the urban, suburban and rural differentials in
31. B. G. E., Estudos de E-tatistica Te6rica e Aplicada,
Estistica Demogrifica No. 14, Estudos Sobre a Natalidadee a Mortal
idade no Brasil, (Rio de Janeiro: Servigo Grifico do I. B. G. E.,
1952), p. 157.
NUMBER OF PERSONS r-
FERTILITY RATIO .
EI J432-531 ~ 732-831 :: ::
Fig. 2. The Level of Fertility by Residence and States, Brazil, 1950.
(Starting at 12:00 on the circles and reading clockwise, the segments
represent the urban, suburban and rural populations, respectively).
Prepared from data in I.B.G.E., Conselho Nacional de Estatfstica,
VI Recenseamento Geral do Brasil, Censo Demografico (1 de julho de 1950),
Seleabo dos Principais Dados, (Rio de Janeiro: Servigo Grafico do I.
B. G. E., 1951, 1952, 1953), passim.
fertility were sufficiently large in 1950, to place these residential
classes in an increasingly high fertility category as one moves from
the urban to the suburban to the rural category. In Alagoas and
Goias, although the suburban portion of the state has a higher rate
of reproduction than the urban, this difference was not sufficiently
large to place it in a higher category.
It is worthy of note that three of the most populous and most
urban states (Rio Grande do Sul, Sa'o Paulo and the Federal District),
possess the lowest rates of reproduction recorded for the urban
segment of the population, All of these are located in the southern-
most section of the country.
Only in Santa Catarina, Goias, and Mato Grosso, does urban fertil-
ity fall into the rather high range of from 53.2 to 63.1 children 0-4
per 100 women aged 15-49.
Highest rural fertility is to be found in Santa Catarina, in
the Southern region, Mato Grosso in the West Central region, in the
territory of Acre in the Northern region, and, in confirmation of
popular belief in Brazil, in the Northeastern state of Ceara.
The states possessing the lowest rate of reproduction in their
rural areas are Bahia and Maranhao, in which the fertility ratio
for that residential category falls in the 63.2 to 73.1 range, only
one category above the rates recorded for the urban portion of Santa
Catarina, Goias, and Mato Grosso.
The greatest differentials to be observed between the urban and
rural rate of reproduction occur in the states of Rio Grande do Sul,
Sao Paulo and Ceara.
Another indication of the influence of urbanity on fertility
may be gained by examining the fertility ratio of the municipio
containing the capital city, for each of the states (Table 6).
Considerable variations occur in the fertility ratio for these
areas, largely due to regional differentials in fertility (discussed
in a subsequent chapter), and because of the varying proportion of
the urban population in the municipios containing the state capitals.
Nevertheless, it may be noted that two of the lowest fertility ratios
obtained (37.6 and 36.8), are to be found in Sao Paulo and the Federal
District, Brazil's largest urban centers, and that all capital
municipios with a population of less than 140,000 possessed fertility
ratios of 43.5 or over.
It is also worthy of mention-that only one capital municipio
with more than 140,000 inhabitants had more than 50 children 0-4
per 100 women aged 15-49; whereas six of the eleven capital municipios
with populations of less than 140,000 exceeded that figure.
Residential Differentials in Sao Paulo
In order to further examine the influence of residence upon the
rate of reproduction, a map of Sao Paulo was prepared showing the
level of the fertility ratio in each of its 369 municipios (Fig. 3).
Due to the small size of the municipios the attempt was not made
to show the size of the population of these political subdivisions.
However, it may be noted that the greatest concentration of people
occurs in the municipio containing the state capital, and in the area
FERTILITY RATIOS FOR THE MUNICIPIOS CONTAINING
THE CAPITAL CITIES, BRAZIL, 1950, BY STATES*
Municipio Containing Number of children under
Capital City Population 5 per 100 women 15-49
Distrito Federal 2,377,451 36.8
Sao Paulo 2,193,096 37.6
Recife 524,682 46.6
Salvador 417,235 42.4
Porto Alegre 394,151 37.0
Belo Horizonte 352,724 45.3
Fortaleza 270,169 52.3
Belem 254,949 49.7
Niteroi 186,309 42.1
Curitiba 180,575 39.4
Manaus 139,620 57.4
Maceid 120,980 43.5
Sao Luis 119,785 44.1
Joao Pessoa 119,326 49.1
Natal 103,215 51.9
Teresina 90,723 58.6
Aracaju 78,364 46.9
Florianopolis 67,630 55.5
Cuiaba 56,204 64.1
Goiania 53,389 57.9
Vitoria 50,922 44.3
Compiled and computed from data in I. B. G. E., Conselho
Nacional de Estatistica, Servigo Nacional de Recenseamento,
VI Recenseamento Geral do Brasil, Censo Demografico
(1 de julho de 1950), Selegao dos Principais Dados, (Rio
de Janeiro: Servigo Grafico do I. B. G. E., 1951, 1952,
0 0 A 0
- o- a.
t: <. 0
.0 0 p
0 r a
>1 \ 0
immediately surrounding it, with lesser concentrations of population
stretching out from it in a northwesterly direction. The lowest
density of population occurs in the western part of the state.
A high degree of urbanity is indeed associated with low
fertility, and the greatest urban concentrations may be detected on
the map by observing the areas of lowest fertility which stretch from
the southeast coast (Santos) to the metropolis of Sao Paulo, and on
through the "old zone" of settlement.
The depressing influence on the rate of reproduction exerted
by the state's large urban concentrations may be observed in the
areas immediately surrounding the large cities, which are themselves
characterized by very low fertility. In but a very few instances
does a minicipio adjacent to an area of lowest fertility (under 42.0)
possess a fertility ratio sufficiently high to place it in the fourth
fertility category of 60.0 to 68.9, or above. Other municipios of
fairly low fertility (42.0 to 50.9)which occur at a distance from
the state's largest cities, themselves contain fairly good sized
Highest fertility rates in Sao Paulo are found to prevail in the
highly rural "zones of new penetration" that stretch along the
northern border of the state and extend westward and then southward
4For the distribution of population in Sio Paulo as well as for
Brazil as a whole, see T. Lynn Smith and Alexander Marchants, Eds.,
Brazil: Portrait of Half a Continent, (New York: The Dryden Press,
1951), p.o Ii.
to cover the western portion of the state's territory. In these very
rural municipios, many of them created after 1945, the rate of
reproduction reaches in many cases the astounding figure of 78.0 or
more children under 5 years of age per 100 women in the child-
bearing ages. Were infant and child mortality not such an important
factor in the decimation of the inhabitants in the 0-4 age group, this
rate would be even higher.
Residential Differentials in Bahia
In a further attempt to analyze residential differentials in the
rate of reproduction, a map was prepared of the state of Bahia,
showing the size of the population of the municipios and the level of
the rate of reproduction in each (Fig. 4).
Bahia was chosen because of its contrast with Sao Paulo.
Whereas Sao Paulo is one of the most highly urbanized and white states
of the union, Bahia is very rural and a great proportion of its inhabi-
tants fall into the black and mixed racial categories. Whereas Sao
Paulo has been a major goal of internal migration, Bahia has contribu-
ted heavily to the migratory stream.
A careful examination of Figure 4 reveals that one of the
lowest levels of fertility occurs in the capital municipio which
also possesses by far the largest urban concentration.
The depressing effect on the rate of reproduction exerted by this
large urban agglomeration can be observed in the municipios immediately
surrounding it, all of which possess low levels of fertility, although
in no case as low as that of the capital municipio (under 47.9).
J UNDER 47.9
479-558 = 719-798
11=559-638 M 799-OVER
Fig. 4. The Distribution of the Fertility Ratio iln Bahia by Municipios,
1940. Prepared from data in I. B. G. E., Recenseamento Geral
do Brasil (1 de setembro de 1940), Censo DemT-rnc fico, Porulagao
e Habitago, Serie Regional, Part XII, Vol. 1, (Rio de Janeiro:
Servico Grafico do I. B. G. E., 1950), passim.
Not surprisingly, the highest fertility ratios were recorded in
the interior and highly rural portions of the state. From the stand-
point of size of population, high rates of reproduction are to be
found especially in the municipios that lie to the southwest of the
Attention should be called to a group of twelve municipios all
lying outside the area of influence of the capital city, and
characterized by extremely low fertility. These areas are surrounded
by municipios of high fertility and constitute what might be termed
"low fertility islands." Inasmuch as these subdivisions of the state
territory are characterized by a high degree of rurality, and are
mostly located at a good distance from the state capital, the rural-
urban differential cannot be invoked as an explanation of this
Seven of these municipios lie in the central part of the state
and to the south of it, two lie in the northeastern section of the
state, and three are relatively close to the state capital. Two of
these municipios possess levels of reproduction that place them in
the lowest fertility category of under 47.9, a distinction they share
only with the mnnicipio harboring the state capital itself. The
other municipios fall into the next fertility category ranging from
47.9 to 55.9, still very low for the state of Bahia.
An explanation of this anomaly is to be found in the fact that
the municipios in question are mostly in areas of severe drought.
This has resulted in considerable migration from these areas princi-
pally to southern Brazil, and especially to Sao Paulo.
This migratory stream, composed largely of males in the early years
of adulthood characteristic of long distance migration, has resulted
in a shortage of men in the reproductive years of life, which in
turn has lowered the proportion of women in the fertile ages who are
married, thus lowering the rate of reproduction. An examination of
Figure 5, which gives the origin of migrants lodged in the migrant
receiving station (hospedaria) in the city of Sao Paulo, reveals that
the low fertility rural municipios in question are among those that
contributed most heavily to the flow of migrants from Bahia that
reached Sao Paulo. The only exceptions to this rule are the three
municipios located closer to the state capital. It is probable that
the depressing influence on fertility of the capital city combined
with migration to it rather than to the south is responsible for the
low rate of reproduction in these municipios.
The two low fertility municipios located in the northeastern
section of the state are Juazeiro (on the state line) and Bonfim just
to the south of Juazeiro. In addition to being an area that is losing
heavily by migration, Juazeiro is the terminal point of one lap of
the overland southward journey made by migrants from states farther
to the north. Bonfim is a railway and communications center. It is
likely that in these municipios, the pressures of a hostile environment
are aided in the impetus they give to migration by the wide dissemina-
tion of news of the promised land to the south.5
5For a substantial discussion of migration from Bahia and the
Northeast, see Smith, op. cit., pp. 272-81.
Fig 5 Origin of Migrants Lodged at the Hospedaria
in Sgo Paulo, 1935-1940. After T. Lynn Smith,
Brazil: People and Institutions, Revised Edition,
(Baton Rouge: Louisiana State University Press,
1951), p. 276.
Since the data presented in Fig. 5 for Bahia are for 1935-1940,
the significance of this map is heightened, inasmuch as it covers
the five year period preceding 1940, and thus relates to the period
when the children 0-4 used for the computation of the 1940 fertility
ratio for Bahia were coming into the world.
In order to gauge the effect of outward migration on the popula-
tions of the municipios in question, these were compared with those
municipios in the state possessing exceptionally high fertility--fall-
ing in the 71.9 to 79.8 range and into the 79.9 and over range.
These two groups of municipios of low and high fertility were compared
with regard to the sex ratio in the reproductive ages, and in rela-
tion to the proportion of married females in the fertile age groups.
Table 7 presents the results of the comparison of the sex ratios
in the reproductive age groups in each group of municipios. A low
sex ratio or a low proportion of men in the population aged 15-49 in
the low fertility municipios is evident. The total sex ratio for these
municipios in the age group in question is but 81.8 men per 100
women. In municipios of high fertility there are 97.8 men per 100
women for the 15-49 age group. The same differential in the sex
ratio is revealed in the 15-29 age range in the two groups of muni-
cipios. Of special significance however, is the acute shortage of men
in the low fertility municipios in the most reproductive ages--20-29.
In this age group there are only 76.3 men per 100 women in the low
fertility mnmicipios as compared with 93.9 men per 100 women in those
exhibiting high fertility.
SEX RATIOS AMONG SPECIFIED AGE GROUPS IN THE MU1tICIPICS
OF BAHIA 'WHICH HAD THE HIGHEST AND LOWEST
FERTILITY RATIOS, 1940h
Sex Ratio Sex Ratio Sex Ratio
Municipio 15-49 20-29 15-29
Amargosa 80.4 75.2 77.6
Areia 82.6 76.6 79.0
Anchieta 78.6 69.0 74.4
Andaria 97.8 88.5 90.5
Bom Jesus Dalapa 79.1 76.6 79.7
Bomfim 76.5 72.7 74.4
Esplanada 79.8 76.1 80.2
Guanambi 98.1 96.8 107.1
Juazeiro 76.6 71.0 75.1
Lengois 94.0 86.2 88.7
Palmeiras 93.5 93.8 93.2
Paramirim 67.5 59.3 64.1
- --- I
TABLE 7 continued
SEX RATIOS AMONG SPECIFIED AGE GROUPS IN THE MUNICIPIOS
OF BAHIA WHICH HAD THE HIGHEST AND LOWEST
FERTILITY RATIOS, 1940*
Sex Ratio Sex Ratio Sex Ratio
Municipio 15-49 20-29 15-29
Alcobara 107.3 106.6 103.4
Boa Nova 98.3 91.6 91.0
Camani 106.6 100.1 100.5
Casa Nova 89.5 87.1 88.8
Conc. do Coite 96.4 90.6 94.9
Conquista 95.1 91.4 91.4
Encruzilhada 101.3 98.4 95.5
Ire8e 95.6 90.4 92.0
Nilo Peaanha 101.8 91.8 94.8
Pilao Arcado 92.1 85.0 89.6
Povoes 96.8 96.2 93.2
Pombal 91.7 92.1 89.1
Porto Seguro 103.2 94.1 94.7
Sta. Cruz Cabralia 118.0 112.6 107.9
Santa Maria 79.7 73.4 77.8
Una 136.9 128.7 121.4
Total 97.8 93.9 93.2
*Source: Compiled and computed from data in I. B. G. E. Recenseamento
Geral do Brasil (1 de setembro de 1940), Censo Demografico,
Populago e Habitago, _Serie Regional, Part XII, Vol. 1,
(Rio de Janeiro: Servico Grafico do I. B. G. E., 1950),
The migration caused shortage of men in municipios exhibiting a
low rate of reproduction is also reflected in the proportion of women
in the childbearing ages who are married (Table 8 ). In low fer-
tility m~nicipios the proportion of women in the fertile ages who
are married is much lower than in the municipios of high fertility.
Here again the disproportion between the two groups of municipios
is especially large in the most fertile ages. Of those women who fall
in the 20-39 age group, only 51.5 percent are married in the low fer-
tility municipios, whereas in those municipios with a high rate of
reproduction, the proportion reaches 68.5 percent.
Residential Differentials in Santa Catarina
Estimated birth rates in Santa Catarina municipios in 1940 were
calculated by the Brazilian Institute of Geography and Statistics.
Although the rates obtained for the more rural portions of the state
can be regarded as no more than fair approximations, the rates for
the more urban areas are probably reasonably accurate due to better
registration of vital statistics.
The territorial distribution of the estimated birth rates for
Santa Catarina is presented in Figure 6. As was expected, lowest
birth rates (under 35.0) are to be found in the municipios harboring
the two principal cities of the state--Joinville and Blumenau. The
unmicipios adjacent to them also possess low birth rates. All but
two of the municipios falling into the rather low 35.1 to 40.0 birth
rate category are contiguous with the two highly urban municipios.
None of the municipios falling into the two highest estimated birth
PROPORTIONS OF FEMALES MARRIED AMONG SPECIFIED AGE
GROUPS IN THE MUNICIPIOS OF BAHIA WHICH HAD THE
HIGHEST AND LOWEST FERTILITY RATIOS, 1940*
Married females Married females
20-39 as a per- 20-39 as a per-
Municipio cent of all fe- cent of all fe-
males 15-49 males 20-39
Amargosa 27.7 47.6
Areia 28.0 47.5
Anchieta 31.8 51.4
Andaria 30.0 47.0
Bom Jesus Dalapa 34.2 55.1
Bomfim 31.5 53.1
Esplanada 30.8 51.9
Guanambf 31.1 54.0
Juazeiro 33.4 53.4
Lenqois 29.9 47.3
Palmeiras 32.4 52.5
Paramirim 33.8 55.4
Total 31.2 51.5
TABLE 8 continued
PROPORTIONS OF FEMALES MARRIED AMONG SPECIFIED AGE
GROUPS IN THE MUNICIPIOS OF BAHIA WHICH HAD THE
HIGHEST AND LOWEST FERTILITY RATIOS, 1940.x
Married females Married females
20-39 as a per- 20-39 as a per-
Municipio cent of all fe- cent of all fe-
males 15-49 males 20-39
Alcobaga 43.5 71.5
Boa Nova 43.5 71.5
Camaxni 29.0 47.4
Casa Nova 39.8 63.3
Cone. do Coite 41.3 68.4
Conquista 40.9 68.4
Encruzilhada 49.9 81.7
Irece 42.9 68.8
Nilo Peganha 32.4 52.1
Pilao Arcado 42.3 67.2
Pog~es 42.4 70.7
Pombal 39.1 65.2
Porto Seguro 46.0 75.8
Sta. Cruz Cabralia 42.8 64.4
Santa Maria 39.1 63.4
Una 46.8 72.6
Total 41.8 68.5
Compiled and computed from data in I. B. G. E., Recenseamento
Geral do Brasil (1 setembro de 1940), Censo Demografico,
Populacao e Habitacao, Serie Regional, Part XII, Vol. 1,
(Rio de Janeiro: Servigo Grafico do I. B. G. E., 1950),
I Under 55.o
B 35.1 40.0
/2 40.1 45.0
60.1 ov re
Fig. 6. Distribution of Estimated Birth Rates in Santa Catarina,
1940. After I. B. G. E., Estudos de Estatfstica Teorica e
Aplicada, Estatfstica Demografica No. 10, Pesquisas Sobre a
Natalidade no Brasil, (Rio de Janeiro: Servigo Gr&fico do I. B.
G. E., 1950), p. 135.
rate categories (45.1 to 50.0 and 5C.l and over), are adjacent either
to the municipios of lowest fertility (under 35.0) or to those muni-
cipios in the next category (35.1 to 40.0) that are adjacent to the
municipios of lowest fertility.
Trends in Rural-Urban Differentials
Unfortunately, among the many deficiencies of the Brazilian
censuses has been the failure to provide age and sex data for the
various residential categories. This important tabulation was pre-
sented for the first time in the sumnrary volumes of the 1950 census.
It becomes necessary therefore, when dealing with previous
censuses, or when wishing to make a comparison of data for the various
census years, to resort to a substitute device for gauging rural and
urban levels of fertility. Such a device is employed in Table 9
The capital city of each state is with one minor exception the
largest urban center of that state. The range in size from the
capital city to the next largest urban center is in most states very
considerable. As a result, an approximation of the rural-urban
dichotomy can be achieved by considering the unnicipio of the capital
city as urban, and the remainder of the state as rural.
It should be noted that in using this comparison, the rural por-
tion of the capital municipio is classed as urban and that the urban
portions of the remainder of the state are classed as rural. There-
fore, if "pure" urban and rural categories could be used in computing
the fertility ratio, the "urban" figure would be lowered and the
"rural" raised, thus resulting in greater differentials.
Analysis of Table 9 shows that Brazil is no exception to the
general rule of lower fertility in urban than in rural areas.
Whether the rural-urban comparison of the rate of reproduction be
made in 1920, 1940 or 1950, the conclusion is the same. Only in one
instance that of the territory of Guapore in 1950, is urban fertility
higher than rural fertility as measured by the capital municipio-
remainder of state dichotomy. The reader will also note that rural-
urban differentials for the territories of Amapa, and Guapore are
No significance should be attached to this however. These
territories had extremely small populations in 1950--Amapa possess-
ing 37,477 inhabitants, Guapore' 36,935 people and Rio Branco only
18,116. The capital municipio of these frontier territories con-
tains a majority of the population. 54.9 percent of the population
of Amapa' lives in the capital municipio, and like percentages for Guapore'
and Rio Branco are 73.8 and 95.2. Of the population of these capital
municipios only a small number and proportion are urban. The propor-
tion of city dwellers in the capital municipios of Amapa, Guapore'
and Rio Branco in 1950 was only 11.2 percent, 16.4 percent and 25.7 per-
cent, respectively. In the case of Rio Branco where the "reverse"
differential is observed, the total number of women 15-49 plus the
total number of children 0-4 present in the capital municipio add to
less than 300.
In all states and the one territory of the Northern region of
Brazil for which data is available for more than one census year,
FERTILITY RATIOS FOR THE MUNICIPIO CONTAINING TIH
STATE CAPITAL AND THE RUAINDER OF THE STATE,
1920, 1940 AND 1950"
Capital Municipio Remainder of State
Region and State Fertility Ratio(l) Fertility Ratio(l)
1920 1940 1950 1920 1940 1950
Acre(2) ---- 67.7 79.0 --- 92.4 93.9
Amapd(2) --- ---- 70.0 --- --- 78.2
Guapore(2) -- -- 72.0 -- ---- 760
Rio Branco(2) -- -- 84.4 ---- ---- 69.0
Amazonas 48.4 47.7 57.4 69.5 77.8 81.0
Par 41.3 43.9 49.7 59.5 66.6 73.5
Maranhgo 30.4 41.1 44.1 55.1 68.8 66.9
Piauf 50.6 53.5 58.6 63.5 75.0 77.5
Ceara 33.7 46.9 52.3 56.6 79.5 80.4
R. G. do Norte 34.9 41.1 51.9 62.2 70.0 76.4
Parafba 42.5 46.1 49.1 64.6 70.9 71.9
Pernambuco 34.1 37.3 46.6 59.5 65.3 69.5
Alagoas 29.1 34.3 43.5 59.0 66.8 72.6
TABLE 9 continued
FERTILITY RATIOS FOR THE MWUNICIPIO CCOTAITING THE
STATE CAPITAL AND THE REMAINDER OF THE STATE,
1920, 1940 AND 1950*
Capital Municipio Remainder of State
Region and State Fertility Ratio(l) Fertility Ratio(l)
1920 1940 1950 1920 1940 1950
Sergipe 31.1 36.7 46.9 53.0 66.7 73.1
Bahia 32.3 38.4 42.4 60.9 64.5 68.1
Minas Gerais 42.8 39.8 45.3 69.9 67.7 68.4
Espirito Santo 45.6 44.9 44.3 74.2 77.8 76.9
Rio de Janeiro 42.7 43.5 42.1 65. 68.2 69.5
Distrito Federal 40.9 36.5 36.8- -- -
Sao Paulo 46.0 34.8 37.6 70.0 66.2 65.1
Parana' 49.7 40.6 39.4 70.4 74.4 77.1
Santa Catarina 40.6 48.3 55.5 73.5 79.3 81.8
R. G. do Sul 39.3 34.1 37.0 70.3 68.5 69.8
Goids 54.5 66.7 57.9 63.1 70.7 72.7
Mato Grosso 53.1 58.6 64.1 71.0 74.0 78.4
1920 data from T. Lynn Smith, Brazil: People and institutions,
(Baton Rouge: Louisiana State University Press, 1947),
p. 238. 1940 data compiled and computed from data in I. B.
G. E., Recenseamento Geral do Brasil (1 de setembro de 1940),
Censo Demografico, Populaq.o e Habitagco Serie Regional,
(Rio de Janeiro: Servigo Grafico do I. B. G. E., 1950, 1952),
passim. 1950 data compiled and computed from data in I. B.
G. E., Conselho Nacional de Estatfstica, Servigo Nacional de
Recenseamento, VI Recenseamento Geral do Brasil, Censo Demo-
grafico (1 de julho de 1950), Selegao dos Principais Dados,
(Rio de Janeiro: Servigo Grafico do I. B. G. E., 1951, 1952,
(1) Fertility ratio determined as follows: children under 5
fertility has increased from one census year to another in both the
rural and urban portions of the population. The same conclusions
may be reached concerning the Northeastern region with the single
exception of a small decline between 1940 and 1950 in the rate of
reproduction of the rural portion of Maranhao.
Of the six states in the Eastern region, three (Sergipe, Bahia
and Minas Gerais) follow the lead set by those in the Northern and
Northeastern regions in that they show a consistent increase in
fertility in both the urban and rural categories, from 1920 to
1940 and from 1940 to 1950. The only exception is the slight de-
crease in the rural portion of Minas Gerais from 1920 to 1940, fol-
lowed by an increase from 1940 to 1950, bringing the 1950 fertility
ratio practically to the level of the 1920 one.
In Espfrito Santo a slight decrease is seen in the urban portion
from census year to census year, and a slight decrease is observed for
the rural areas from 1940 to 1950. The rate of reproduction in urban
areas remained practically unchanged in Rio de Janeiro from 1920 to
1940 and increased in the rural portion during that period. The
Federal District which is almost in its entirety urban showed a de-
crease in fertility from 1920 to 1940 with fertility in 1950 remaining
practically at the 1940 level.
In the rural portion of the four states constituting the Southern
region, the national trend of increasing fertility is observed in
Parana, Santa Catarina, and Rio Grande do Sul. Only in Sao Paulo is
the opposite tendency in evidence, the rural fertility ratio for that
state having declined from 70.0 in 1920 to 66.2 in 1940 to 65.1 in 1950.
In the urban areas of the states in the Southern region, only
Parana and Santa Catarina show a consistent trend. In Parana this
amounted to a very sharp decrease in fertility from 1920 to 1940,
followed by a slight decrease from 1940 to 1950. In Santa Catarina
on the other hand, fertility consistently increased from 1920 to
1950. Sao Paulo and Rio Grande do Sul both showed the same type of
variation--in their urban rate of reproduction--a decrease from 1920
to 1940, followed by an increase from 1940 to 1950, which however
was insufficient to bring them back to the 1920 level.
In the West Central region, composed only of the states of Goias
and Mato Grosso, a consistent increase in the rate of reproduction
is also observed in the rural section from 1920 to 1950. In Mato
Grosso the same tendency is evidenced in the urban section. In
Goias a slight decrease occurred between 1940 and 1950 in the urban
category, preceded by an increase from 1920 to 1940. This decrease
was not sufficient however to bring the 1950 figure below that for 1940.
It is difficult if not impossible to determine the fluctuations
in the birth rate for Brazil except for a few areas of fairly adequate
registration. It is probable however, that the birth rate has remained
more or less constant, and that the fluctuations observed in the fer-
tility ratio are to be explained in large measure by two factors: a
decline in infant mortality and continued rapid urbanization.
Infant mortality in Brazil is undeniably high. Since the fertility
ratio is based on children 0-4, a reduction in the deaths of children
under one year of age, will cause an increase in the fertility ratio
if the birth rate remains the same. Similarly, a reduction of
mortality in the 1-4 age group will have the same effect. When a
high level of mortality in the tender ages prevails, caused largely
by infectious and communicable diseases, considerable reductions in
the deaths of infants and children may be accomplished by the intro-
duction of a few simple methods of sanitation and child care.
Counteracting the inflating effect on the fertility ratio of
reduced mortality in the first years of life, are the pressures of
an urban mode of life that are known to have a depressing effect on
the birth rate. Thus it is possible that increases in the fertility
in states such as Ceara and Sergipe, where urbanization has been slow,
may be due to a reduction of the mortality of children under five
and especially in the 0-1 age group. In states such as Sao Paulo and
in the Federal District, where fertility has decreased or has shown
little change, it has probably done so in spite of reduced infant
mortality, as a result of rapid urbanization.
Although infant mortality rates are available for only a few
areas in Brazil because of the general under-registration of births
and deaths, those for the Federal District and some of the state
capitals where more or less adequate registration of vital statistics
exists, may be indicative of the general trend.
Mortara gives fairly satisfactory estimates of the number of
deaths of infants under one year old per thousand live births in the
Federal District for the years 1939-1950 (Table 10). This table
clearly indicates that infant mortality steadily decreased in the
Federal District during the 1939-1950 period. In view of this, the
rapid urbanization that the Federal District has experienced,
increasing in urbanity from 46.2 percent in 1940 to 74.8 percent in
1950, must be adjudged the major cause of the lack of change in the
rate of reproduction in that area from 1940 to 1950.
INFANT MORTALITY IN THE FEDERAL DISTRICT, 1930 TO 1950*
Year Infant Mortality
*Source: I.B. G. E., Estudos de Estatfstica Tedrica e Aplicada, Esta-
tistica Demogrifica No. 14, Estudos Sobre a Natalidade a
Mortalidade no Brasil, (Rio de Janeiro: Servico Grifico do
I. B. G. E., 1952), p. 16.
Similar conclusions can be reached by an examination of
Table 1 which gives the probability of death-per thousand and life
expectancy for age 0, 1 and 5 for the 1939-1941 period and for 1950
in the Federal District.
PROBABILITY OF DEATH AD LIFE EXPECTANCY FOR AGES
0, 1 AND 5, IN THE FEDERAL DISTRICT, 1939-1941 AiD 1950*
Death Per Thousand Life Expectancy
Age 1939-1941 1950 1931-1941 1950
0 159.24 98.02 42.31 52.51
1 66.52 36.75 48.23 57.16
5 6.09 3.05 51.32 56.75
*Source: I. B. G. E., Estudos de Estatfstica Tedrica e Aplicada, Esta-
tistica Demografica No. 13, Pesquisas Sobre o Desenvolvimento
da Populagao do Brasil, (Rio de Janeiro: Servigo Grafico do
I. B. G. E., 1951), p. 38.
Not only has the probability of death for the under 1 age group
been substantially reduced and life expectancy increased, but the same
is also true of those persons one year old and five years old, indicating
a reduction in mortality at all age levels under 5.
That reduced infant mortality has not been restricted to the
Federal District can be seen from Table 12.
INFANT MORTALITY IN FIVE STATE CAPITALS, 1940 AND 1950*
Infant Mortality Rate
Years next Years next
to 1940 to 1950
Sio Paulo 135.21 86.05
Recife 267.03 243.56
Salvador (Bahia) 201.40 162.57
Porto Alegre 187.46 107.26
Belem 159.67 110.96
*Source: I. B. G. E., Estudos de Estatfstica Teorica e Aplicada, Esta-
tistica Demografica No. 16, Pesquisas Sobre a Natalidade no
Brasil, (Rio de Janeiro: Conselho Nacional de Estatistica,
I. B. G. E., 1953), p. 84.
Although the figures given are estimates most of them are probably
RACIAL DIFFEREINTIALS IN FERTILITY
The racial composition of Brazil's population and the differen-
tials among the rates of reproduction of the various racial groups,
are of great interest. Brazil is truly a "melting pot" into which
have been introduced representatives of all the major racial stocks
First among these were the caucasoids, in the form of Portuguese
officials, soldiers and settlers who found the country already in-
habited by members of the mongoloid race, represented by the original
Indian inhabitants. The Portuguese were soon followed by Negroids,
slaves imported from Africa for the purpose of working the mines and
plantations of the expanding economy. In more recent times, large
contingents of Portuguese, Italians, Spaniards, Germans, Poles and
other caucasian stocks, and a large number of Japanese have been
introduced into the country; the latter together with their descen-
dants numbered in 1950 more than 300,000.
Largely due to the abolitionist zeal of a Minister of Finance (Rui
Barbosa) who in 1891, shortly after the abolition of slavery, ordered
all records pertaining to the slave traffic destroyed,l much less is
known about the racial composition of the Brazilian population of a
century ago than would otherwise be the case. Among our principal
1Arthur Ramos, The Negro in Brazil, (Washington: The Associated
Publishers, Inc., 1939), p. 5
sources for this information are, in addition to such official records
as survived, the estimates and accounts of those indefatigable globe-
trotters who for various reasons left their Victorian comforts of
home to take up residence for shorter or longer periods in what to
them must have seemed, at least on departure, the barbaric and primi-
tive shores of Brazil.
One of these, Thomas Ewbank, writing in 1856, cites a newspaper
article appearing in the Diario of December 11, 1847, written by a
"Senhor de Sousa, a native statistician." According to this authority,
the population of Brazil was at that time, 7,360,000, a figure which
excluded "untamed Indians," estimated at 800,000. Of those included
in the total, 2,160,000 or 29.3 percent were classified as white,
3,120,000 or 42.4 percent as Negro slaves; 1,100,000 or 15.0 percent
as free colored (meaning mixed-bloods); 800,000 or 10.9 percent as
domesticated Indians; and finally, 180,000 or 2.4 percent as free
blacks. It should be stressed that this estimate as well as the many
others pertaining to the population before 1872, the date of the first
census, are at best educated guesses.
The figures cited by Ewbank, even if containing large errors, are
eloquent testimony of the extent to which racial miscegenation occurred
in Brazil, practically all of it between Portuguese and Negroes or
Portuguese and Indians, since little crossing between Negroes and
2Thomas Ewbank, Life in Brazil, or a Journal of a Visit to the Land
of the Cocoa and the Palm, (New York: Harper Brothers, 1056), pp. 431,
Indians is believed to have existed. Consequently, the Portuguese
contributed to the racial composition of the population far out of
proportion to their numbers.
The Portuguese were themselves the products of prolonged and
extensive miscegenation. Having been overrun by various nations
and tribes, and finally going through several centuries of Moorish
occupation, Portugal acquired as a result of the many contacts with
alien cultures, a "social plasticity" not possessed by any other
European nation. The Portuguese colonist was always ready to com-
promise and brought with him no fixed prejudices. Because of the
superior status which the Moors enjoyed in Portugal for centuries,
the Portuguese considered them, including even more Negroid types as
worthy of respect, prestige and even marriage.
These attitudes of the Portuguese combined with the scarcity of
white female settlers in Brazil, served to produce a period of great
miscegenation. The attempt by a nation of 1,200,000 people to
colonize effectively an area as vast as that of Brazil, was a stag-
gering task. Indeed in 1550, half a century after discovery, only
3,000 colonists, a large majority of them men, resided in Brazil.3
Under these conditions it became a matter of necessity and even
policy to permit if not encourage miscegenation with the local popu-
lation. That miscegenation did occur on a large scale first with the
3Fernando de Azevedo, Brazilian Culture, (New York: The Macmillan
Company, 1950), p. 31.
Indian and then with the Negro is unmistakable.
It was during the first century of colonization especially, that
the mixture occurred between the Portuguese and the Indian:
The European landed sliding on nude Indians; even priests
of the Company had to go shore with care so as not to become
immersed in flesh... The women were the first to give themselves
to the white men, the more ardent ones rubbing themselves against
the legs of those they supposed to be gods.
Truly the Portuguese have never been surpassed or equaled in their
miscibility. On arrival the Portuguese "mixed himself deliciously with
women of color...multiplying himself in mestizo children," so that a
handful of aggressive Portuguese were able to gain possession of vast
areas in competition with other colonial powers. Another factor in
this miscegenation was the Portuguese ideal of beauty, acquired from
prolonged contact with the Moors. This ideal, contrary to that in
existence in other parts of Europe, was that of the enchanted Mooress,
a delicious type of dark women with black eyes, wrapped in sexual
mysticism--always wearing red, always combing her hair or bathing
in rivers or in the waters of haunted fountains.5
It is easy then to understand why the Portuguese felt no reluctance
in entering into sexual relations with the Indian woman who strongly
resembled the physical type idealized by them. Indeed, even if they
had not a taste for exotic women, there would have been no other choice.
At the time when the Indian population along the coast had either
Gilberto Freyre, Casa Grande e Senzala, (Rio de Janeiro: Jose
Olympio, 1946), Vol. 1, p. 209.
5Ibid., p. 93
been driven inland or been almost completely absorbed, a sufficient
number of Negro women were imported as slaves for the process of
racial mixture to continue on an extensive basis.
The latifundium was an important factor in the fostering of
racial miscegenation. Conditions on the great plantations of
colonial Brazil were highly conducive to the fusion of the races.
Oliveira Vianna states that
In reality during the Colonial epoch the latifundium
was the breeding field par excellence. In it the whites--the
owners, their relatives, their agregados--exercised a dominating
role. They were the sires, the great impregnators of the
Indian women, the fiery stallions of the Negro females. Some
of them, even among the most noble, left only "filhos naturaes
e pardos" (illegitimate children and mulattoes) according
to the testimony of the Conde da Cunha.... The hale bloods,
are, then, a historical product of the latifundia.
The female Negro slave came to play an important role in Brazil-
ian family life. A Negro boy was the first playmate of the slave-
owner's children and a mulatto girl more often than not initiated the
son of the master into sexual experience. The superior position of the
slave owner, aided by a widely prevalent double sex standard, facili-
tated and encouraged the fusion of the races and a proliferation of
progeny who were half white, and who in turn fused with others of
varying combinations of color until according to the source quoted by
Evbank there were in 1847 more than one million mixed bloods in the
country. Rugendas writes that
F. J. Oliveira Vianna, Populaqdes Meridionaes do Brasil, (Sao Paulo:
1933), pp. 73-79, quoted in T. Lynn Smith, Brazil: People and Insti-
tutions, Revised Edition, (Baton Rouge: Louisiana State University
Press, 1954), p. 149.
marriages between white men and colored women are very
common in the middle and lower classes, and there is nothing
shocking about them some occur even in the higher classes.
They only are formalized when a white woman from a rich and
weil-known family marries a man of very dark color, but such
unions are reason for surprise rather than censure.7
Perhaps the situation with regard to the racial composition of
the Brazilian population has been best summarized by Oliveira Vianna,
who, after a lengthy discussion of the great variations present
among the Brazilian Indians: the Negro slaves and the white stocks
that contributed to the racial make-up of the nation, states:
These three fundamental races so different and diverse,
fuse in our territory in varying proportions, according to
the greater or lesser degree of the concentration of each at
this or that point of habitable territory. In some regions,
Indian elements dominate and preponderate in their misce-
genetion with the whites; in others, Negroes are the principal
basis of the fusion; in still others, the percentage of white
elements is higher, and the emergent ethnic type reflects in
its characterization this particular condition of its genesis.
Because of this diversity in the geographic distribution of
the three formative races,regional ethnic types do not re-
veal the same unity of morphological characteristics, or the
same identity of temperament and mentality; which results in
the impossibility from the point of view of anthropology and
ethnology, of binding them together in a single national type.3
The Meaning of "Color" and Census Color Categories
In discussing the racial categories in Brazil, it is necessary
to keep in mind the peculiarly Brazilian concept of color. This con-
cept is based largely on social status In cases in which the indivi-
dual is of mixed blood, and is neither obviously white nor obviously
7Joao Mauricio Rugendas, Viagem Pitoresca Atraves do Brasil,
(Sao Paulo: Livraria Martins Editora. 19417, p. 95.
0F. J. Oliveira Vianna, Evolugao do Povo Brasileiro, (Sao Paulo:
Monteiro Lobato e Cia., ca. 1922), pp. 121-22.
black, his classification into either the white,black, or mulatto
category is largely dependent upon his position in the social
pyramid, always with preference given to the lighter color.
That this situation is not of recent origin is evidenced by
Koster, who writing in 1817 commented upon the malleability of the
definition of "white", pointing out that a mulatto will enter a
religious order or the magistrature so long as his documents say he
is "white", although he obviously is not, and relates an incident
which clearly illustrates the point.
In conversing on one occasion with a man of colour who
was in my service, I asked if a certain Capitam-mor was not
a mulatto man; he answered, "he was, but is not now." I
begged him to explain, when he added, "Can a Capitam-mor be
a mulatto man?"9
The situation regarding the changing definition of a man's
color according to his social class, is well summarized in the popular
saying that "a rich Negro is white, and a poor white Negro." Dark
mulattoes may be found in the membership of exclusive upper-class
clubs and associations in contradiction of the discriminatory
attitudes assumed by the club. The explanation usually given is that
these men are not regarded as being colored. To them have been ascribed
such qualities as honesty, wealth, good manners, professional competence,
and so forth, that raises them out of the mulatto category and places
9Henry Koster, Travels in Brazil, (Philadelphia: M. Carey and
Sons, 1817), Vol. II, pp. 175-76.
them on the same level as white people.10
As a result, the racial categories presented in the Brazilian
census do not, as is also the case in the United States census,
correspond to an anthropological classification of the population
into racial groups. No doubt, many who are classed as white more
properly belong in the mixed category, and many who are classified
as mixed-bloods, would be better considered as black. Since to be
considered black is by far the least desirable of the three choices,
the number that anthropologicallyy" belong in this color group is
probably the most distorted.
An examination of census results and procedures with regard to
color in 1940 and 1950 may throw some light on this question.
In 1940, the census enumerator was given the task of classifying
the persons interviewed as to racial affiliation. It is probable that
there was a tendency on the part of the enumerator to place some
persons who might consider themselves to be white in the mixed-blood
category, and to place others who might consider themselves to be of
mixed-blood in the black category. In the 1950 census, "the declara-
tion f/of colors was left to the discretion of the person being
enumerated, lending in this manner, greater veracity to the results
of the census."
10Emilio Willems, "Race Attitudes in Brazil," The American Journal
of Sociology, Vol. LIV, No. 5, (March, 1949), pp. 402 ff.
1TV-HRecenseamento Geral do Brasil, 1950, Estado de Sao Paulo,
Censo Demografico, (Rio de Janeiro: I. B. G. E., Conselho Nacional de
Estatistica, 1954), p. xviii.
This procedure which the National Census Commission thought would
result in greater veracity of the results of the census, could be
expected to result in the self-enumeration of many persons as being
mixed, who would probably be judged black by the census enumerator,
and in the self-enumeration of others as white who would probably have
been placed in the mixed category by the census taker.
Rios, speaking of the racial categories as presented in the 1940
Frequently among us, the individual enumerated as black or
mixed by the overcautious census taker, is not so classified by
his friends or associates /for, it might be added by himself _7.
Sometimes the position he holds or the s atus that he occupies
confer upon him an artificial whitening.2
The Brazilian Institute of Geography and Statistics, in analyzing
the results of the 1940 census states:
With regard to whites one can affirm with certainty that
the number enumerated considerably exceeds that which would
result from a classification based on objective criteria.
The number of blacks enumerated would on the contrary be
considerably inferior to reality if the declaration were to
be made by those being enumerated, but one must remember that
...in some cases numerous mixed-bloods were included among
those classed as black.
The number of mixed-bloods enumerated probably is below
that which would result from an objective classification, and
the number of mixed-bloods classified as whites is probably
larger than the possible excess in favor of the mixed bloods,
in the exchange of classification with the blacks.13
Thus it would seem that when the enumerator classifies the
enumerated as to color, the tendency is to place the enumerated in a
12Jose Arthur Rios, '"Mdanqas na Distribuigao Racial no Estado do
Rio," Sociologia, Vol. XII, No. 2, (maio de 1950), pp. 131-32.
13Quoted in Ibid., p. 143
darker racial group than that in which he would place himself.
Some confirmation of these suppositions is to be found in the
data presented in Table 13. Of particular interest is what other-
vise would be the startling decrease of blacks in the population from
1940 to 1950. Although as will be seen below, the black segment of the
THE POPULATION OF BRAZIL CLASSIFIED ACCORDING
TO COLOR, 1940 and 1950*
Color 1940 1950 1940 1950
White 26,171,778 32,027,661 63.47 61.66
Black 6,035,869 5,692,657 14.64 10.96
Yellow 242,320 329,032 0.59 0.63
Mixed 8,744,365 13,786,742 21.20 26.54
Undeclared 41,933 103,255 0.10 0.21
*Source: I. B. G. E., Conselho Nacional de Estatfstica, VI Recensea-
mento Geral do Brasil, Censo Demografico (I de Julho de 1950),
Estados Unidos do Brasil, Selegao dos Principals Dados, (Rio
de Janeiro: Servico Grdfico do I. B. G. E., 1953), P. 1.
population reproduces at a slower pace than the other racial groups,
it seems unlikely that they are failing to replace their numbers even
considering the loss to their racial category that occurs when inter-
racial mating produces mixed rather than black offspring. It would
seem likely therefore, that the self-enumeration of color used in the
1950 census tended as expected to reduce sharply the number of people
included in the black category.
This tendency can be clearly seen in Table 14. The very large
THE POHIATION OF BRAZIL CLASSIFIED AS BLACK, EB
STATES, 1940 AND 1950"
Region and State Percent Black 1940 Percent Black 1950
Acre(2) 14.2 5.2
Amazonas 7.2 3.4
Para' 9.5 5.4
Maranhio 27.6 15.8
Piauf 31.9 12.9
Ceara 23.3 10.4
Rio Grande do Norte 13.4 9.5
Parafba 13.7 13.0
Pernambuco 15.5 9.3
Alagoas 13.8 7.4
Sergipe 18.7 14.2
Bahia 20.1 19.2
Minas Gerais 19.3 14.6
Espfrito Santo 17.1 11.9
Rio de Janeiro 21.3 17.7
Distrito Federal 11.3 12.3
THE POPULATION OF BRAZIL CLASSIFIED AS BLACK, BY
STATES, 1940 AND 1950
Region and State Percent Black 1940 Percent Black 1950
Sao Paulo 7.3 8.0
Parana' 4.9 4.3
Santa Catarina 5.2 3.6
Rio Grande do Sul 6.6 5.2
Mato Grosso 8.5 9.7
Goias 16.9 10.1
Brazil 14.6 11.0
*Sources: 1940 figures compiled and computed from data in I. B. G. E.,
Recensearento Geral do Brasil (1 de setembro de 1940),
Censo DemogrEfico, Populaco e Habitaggo, Serie Regional,
(Rio de Janeiro: Servigo Grifico do I. B. G. E., 1950, 1952),
passim. 1950 figures compiled and computed from data in I.
B. G. E., Conselho Nacional de Estatlstica, Servigo Nacional
de Recenseamento, VI Recenseamento Geral do Brasil, Censo
Demografico (1 de Julho de 1950), Selegao dos Principais
Dados, (Rio de Janeiro: Servigo Grafico do I. B. G. E., 1951,
1952, 1953), passim.
(1) For 1950 the populations of Amazonas and Rio Branco;
Para and Amapa; and Mato Grosso and Guapore, have been com-
bined to assure comparability with 1940 figures.
reduction in the proportion of blacks frorn 1940 to 1950, which in a
number of states amounts to more than one half percentage-vise, could
only be the result of a self-classification into a lighter racial
group, in most cases probably the mixed category. It is possible
that some of the state census supervisors for 1940 in the Northern
and Northeastern states, where the discrepancies are the greatest,
believing that the population should be darker than the census data
revealed it to be, took it upon themselves to "adjust" the figures
to coincide with their notions. At any rate the states showing the
sharpest decreases in the proportion of the black population from 1940
to 1950 are located in these regions.
Inasmuch as the reproductive powers of both whites and blacks
contribute to the mixed category, it might have been expected to in-
crease, as it did in the ten-year period from 1940 to 1950. Part of
this increase would seem to be the result, however, of the inclusion
of persons formerly classified as black.
Contrary to expectation, the proportion in the white category
actually showed a slight decrease from 1940 to 1950 (Table 15). This
expectation is engendered by the process of the "whitening" of the
population discussed below, by the ever increasing proportion of
whites recorded in censuses previous to 1950, and by the expected tend-
ency of self-enumeration to place a larger proportion of the population
in the white category.
Decreases in the proportion of white people from 1940 to 1950 are
especially sharp in the Northern and Northeastern states, where the