M.A. (CAPE TOWN), PH.D. (EDIN.)
INSTITUTE OF RACE RELATIONS
P.O.BOX 97, JOHANNESBURG
PRINTED IN BASUTOLAND, SOUTH AFRICA
MORIJA PRINTING WORKS
WHOSE ENCOURAGEMENT IN TIMES OF-
STRESS AND DIFFICULTY MAINTAINED
IN ME THAT FAITH IN THE HUMAN
SPIRIT WITHOUT WHICH SCIENTIFIC
ENDEAVOUR WITHERS AND DIES
This book was started in 1939. The intention of writing it arose
from an attempt to review Dr. L. Fick's monograph, The Edu-
cability of the South African Native, No. 3 of the Research Series
published by the S.A. Council for Educational and Social Research.
A critical examination of this monograph raised so many issues that
I found myself compelled either to be dogmatic, countering assertion
with assertion, or else to deal systematically with each problem and
to examine the fundamental assumptions of intelligence-testing and
its applications to racial comparisons.
It appeared worth while to pursue the latter course. The review
grew into a monograph; the monograph developed into a book.
The outbreak of the war compelled a diversion of energies to
other and more immediately urgent tasks, and comparatively little
time could be devoted to the completion of the manuscript.
When I eventually enlisted for full-time military service, one
chapter remained to be written, and the earlier ones needed revision.
This was accomplished by April 1941.
Much time elapsed before the necessary giants for publication
were obtained and the MS. was ready for printing. I was compelled
to leave the preparation for the press entirely to others. I am greatly
indebted to the late Professor R. F. Alfred Hoernl6, of the University
of the Witwatersrand, for undertaking this task. His advocacy assured
that the necessary grants for publication were received. Though
already burdened with many other duties, both civil and military,
he saw the MS. through its final stages and read the proofs. These,
however, were not his only contributions to the publication of this
book. My close association with Professor Hoernld as colleague and
friend created a heightened awareness of the importance for the
future of South Africa, of a dispassionate and scientific study of the
problem of race differences. My decision to write this book owes
much to this. On many points I sought his advice and obtained the
benefit of his mature experience as a writer of many books, pamphlets,
and articles pleading for the exercise both of reason and of humanity
in the complex racial affairs of South Africa.
In a way, therefore, this book is the first fruit of the scientific
and spiritual heritage which Professor Hoernli has left to South
Africa. I trust that it is not unworthy of being offered as such.
As a result of these numerous and inevitable delays in the
publication of the book, much of the statistical material is no longer
up to date. Wages and prices have risen; population figures have
altered; housing conditions are no longer quite the same in detail;
more recent and more accurate data on nutrition have become avail-
able ; and policies are being changed. None the less, it is reasonably
certain that, even if all these details had been brought up to date, a
substantially similar picture of African life in the Union would have
emerged. The general effects of environmental circumstances on
the growth of African intelligence would have remained the same.
Even if drastic changes had occurred, or were to occur, it would
take many years before their effect could become apparent.
Finally, the principles relating to interracial intelligence studies,
which have been fully stated in this book, remain in themselves
unaffected by changes in the circumstances to which they are
applied. They are as valid now as at the time of writing,.and the
necessity for their publication is just as great. Their validity can
only be altered by contrary scientific evidence. As long as these
principles stand, and as long as environmental circumstances remain
substantially the same, such conclusions as are arrived at in this
book must, therefore, themselves remain unchanged.
I wish to express my thanks to Mr. A. Lynn Saffery and Mr.
Julian Rollnick for their assistance in providing material for the
statistical data; to Miss Miriam Janisch and Mr. G. Ballenden for
permission to use Johannesburg Municipal Census figures; and to
some of my former students, notably Miss E. Morrison, for the ana-
lysis of these figures. My indebtedness is particularly great to my
colleague, Mrs. A. Botha, for her readiness to assist with the drawing
of graphs and charts and the checking of references.
Acknowledgement is made of grants in aid of publication
received from the Publications Committee of the University of the
Witwatersrand aid from the South African Council for Educational
and Social Research. Neither body is to be understood as approving,
by virtue of its grant, of any of the statements made or views ex-
pressed in this publication.
I also acknowledge my thanks to the South African Institute of
Race Relations, not only for taking the ultimate financial responsibi-
lity for publication, but also for the typing and re-typing of the MS.
.Air Force Station, Lyttelton, 1943
I THE NATURE OF INTELLIGENCE 1
II THE MEASUREMENT OF INTELLIGENCE 12
The pioneering work of Alfred Binet. The growth of
intelligence. The Intelligence Quotient. The constancy
of the Intelligence Quotient. What do intelligence-tests
measure ? The significance of test-intelligence. The
control group. Interracial intelligence studies in South
III THE SAMPLING PROBLEM 23
How would a representative sample of the African popula-
tion be constituted ?
IV THE "INFLUENCE OF GENERAL CULTURAL MILIEU ON
Attitude towards test-situation. Cultural unfamiliarity of
intelligence-test contents and testing routines.
V SUITABILITY OF EXISTING INTELLIGENCE-TESTS FOR THE
MEASUREMENT OF AFRICAN INTELLIGENCE 45
Group-tests. Scales of the Binet type. Performance-tests.
Suitability of existing intelligence-tests. Can the control-
group device be used ?
VI HOME-ENVIRONMENT 64
VII HOME-ENVIRONMENT (CONTINUED) 76
VIII THE CONTROL OF HOME-ENVIRONMENTAL INFLUENCES
IN THE AFRICAN EXPERIMENTAL GROUP 82
IX THE HOME-ENVIRONMENT OF URBAN AFRICANS 88
The general setting. Material conditions of urban African
homes. Family life in the urban African home. The mental
atmosphere of the home. Can a European control group
for home-environment be found ? The degree of inferiority
of the African urban home-environment.
X THE INFLUENCE OF THE SCHOOL ON INTELLIGENCE 115
The general effect of schooling on test-intelligence.
XI THE SCHOOL EDUCATION OF THE UNION' AFRICAN
The evidence of educational retardation provided by
achievement tests. The qualifications and background of
African teachers. Can the educational factor be controlled ?
XII THE EFFECT OF MALNUTRITION ON THE GROWTH OF
Malnutrition and its effect on the growth and maintenance
of the nervous system. The effects of malnutrition on
XIII THE NUTRITIONAL CONDITION OF THE UNION AFRICAN
Introduction. The tribal African. The African farm
labourer. The urban African population. Can a Euro-
pean nutritional control group be formed ? The possible
effect of illness on the test-intelligence of Africans.
XIV TEMPERAMENT AND INTELLIGENCE 182
Can temperament affect test-intelligence ? The nature of
temperament. The Heymans temperament theory. The
relationship between the Heymans factors and test-intelli-
gence. The effect of temperament on the test-intelligence
XV THE CONTROL-GROUP DEVICE: SOME CONCLUSIONS 191
XVI CRITIQUE OF "THE EDUCABILITY OF THE SOUTH AFRICAN
THE NATURE OF INTELLIGENCE
The concept of intelligence has undergone many changes in the
course of psychological history. At first it was only used as a counter-
part toinstinct. An organism was said to behave intelligently when,
in the course of action initiated by some instinct, it could profitably
experience and regulate its conduct accordingly.
When individual differences between men ceased to be looked
upon as irritating imperfections, obscuring the absolute qualities of
"the mind", and became themselves objects of psychological study,
this usage was gradually abandoned. The term "intelligence" was
thereafter applied almost exclusively to indicate differences in that
capacity which enabled individuals to cope more or less successfully
with the problems of life. Psychological analysis of this new concept
p oved an extremely difficult task, and, despite many years of specu-
lation and research, no finality has yet been reached.
By some, intelligence was identified with some unitary mental
function, such as "the ability to learn", "the power of good combina-
tion", "the ability to act effectively under given conditions", "the
.ability to think in abstract terms", etc. Others, among them Alfred
Binet, the father of intelligence-testing, thought of intelligence as
determined by a number of faculties, notably invention, comprehen-
sion, direction, and censorship. Disagreement was rife, not only
regarding the number of faculties man possessed, but also as to their
relative importance for intelligent behaviour. Could it be said, for
example, that there was a "faculty of memory" in view of.the apparent
independence of memory for faces, for names, for numbers, and for
tunes ? And if, nevertheless, there was such a faculty, did differences
in respect of it lead to differences in intelligence, as popularly under-
stood, or were the latter chiefly the result of differences in the faculty
of comprehension, or ,of reasoning, or of judgement ? These and
similarquestionswere constantly raised, but, in theabsence ofobjective
methods of analysis, it was impossible to provide valid answers.
A third school of thought preferred to look upon intelligence as
the resultant of a large number of abilities, such as imagination,
practical judgement, visual perception, auditory memory, visual
memory, visual discrimination, ideation, and a host of others.
Intelligence, on this view, lacked any unity of its own, becoming a
kind of "general level" or "average", the magnitude of which was
determined by the magnitudes of the constituent elements.
Meanwhile, despite this theoretical confusion, intelligence-test-
ing got into its stride. Somehow these tests succeeded in providing
quantitative measures of whatever it was that the man in the street
referred to as intelligence. They separated the bright from the dull
and had considerable value in predicting scholastic success.
Hence, it appeared worth while to make a careful study of the
nature of these tests, to try and find out what abilities were involved
in answering them successfully, and how these abilities were inter-
related. As the experimental data were numerical test-scores, and
therefore quantitative, they lent themselves readily to mathematical
treatment. Statistical methods were rapidly adapted for this entirely
new application to psychological data, and new analytic devices soon
made their appearance. Objective methods of testing the validity of
hypotheses were thus put at the disposal of the psychologist, whereby
the analysis of the concept of intelligence was placed on an entirely
The central feature of this new approach was the attempt to
reduce the correlation coefficients*, representing the mathematical
relationships between scores on a variety of tests, to a minimum
number of independent factors. Once the existence of such under-
lying variables had been mathematically demonstrated, it became
feasible to seek psychological explanations for them. This method
A correlation coefficient states the tendency for two variables to vary to-
gether. When a person who obtains maximum score in variable A also obtains
maximum in variable B, and when the person who scores the second highest marks
in A also scores second highest in B, and so on, then variables A and B show
perfect positive correlation, indicated by a coefficient of +1. When the reverse
occurs and the person scoring highest in A scores lowest in B, and so on, variables
A and B show perfect negative correlation, indicated by a coefficient of -1.
When the top-scorer in A is as likely to get a high, as a low or medium, score in B,
and so on, then A and B show zero correlation. In most cases, the tendency for
two variables to vary together is imperfect, the degree to which it occurs, either
directly or inversely, being expressed by means of a decimal between 0 and +1,
and between 0 and -1. Correlation coefficients merely indicate relationship, not
causal relationship. Further analysis is necessary tb discover the causal factor
behind the observed relationship. The object of the psychologist, in carrying out
an analysis of this sort, is to reduce the multiplicity of abilities and traits of the
human mind to simpler dimensions, and to describe this multiplicity in terms of
a minimum number of primary attributes, evoked in the genetic structure of the
organism, and providing a framework, within which all the variety wrought by
interaction with the environment is contained.
THE NATURE OF INTELLIGENCE
differed markedly from the purely psychological analysis previously
used, where one was never sure whether such terms as "discrimina-
tion" and "judgement" were merely words, or whether they stood
for unique behavioral qualities compounded by training out of
a number of basic abilities.
The first theory to find general acceptance was the "Two-
Factor" theory of Spearman, first formulated in 19041, and gradually
elaborated to reach its final formulation in 19262. According to this
theory, the score on a test of any aptitude whatsoever is the product
of two independent factors. One of these factors is known as the
"general factor", or g, because it determines, to a greater or lesser
degree, every activity of which the individual is capable." Thus,
aptitude for mathematics or for language studies depends almost
entirely on the presence, in sufficient amount, of this general factor,
whereas the ability to play the violin is far less dependent on it,
and aptitude for tennis still less. The other factor varies from
aptitude to aptitude, primarily not in amount as g does, but in kind.
Ability to play the violin, for instance, requires quite a different kind
of skill from that required for playing tennis, piloting an aeroplane,
or writing poetry. This second factor is therefore called the "specific
factor", dr s, being unique for each separate ability. Though different
abilities are, therefore, determined by different s factors, an overlap
sometimes occurs where the activities are very similar (e.g., "eye-
hand co-ordination", which can be looked upon as a specific factor,
is likely to be a determinant both of skill in tennis and skill in bad-
Spearman also recognizes the existence of "group-factors",
i.e. "those that occur in more than one, but less than all, of any
given set of abilities"3; but he stoutly maintains that only very few
of these can be shown to exist. The most important ones to receive
Spearman's sanction are the group-factors entering into logical,
mechanical, psychological, arithmetical, and musical abilities, which
he thought "were of sufficient breadth and degree to possess serious
practical consequences, educational, industrial and vocational"4.
He suggests, however, that they are more likely to be the outcome of
special training than of native endowment.
Variations in amount ofg determine what are commonly known
as differences in intelligence, whereas variations in the number and
type of s factors determine differences in talent. By common-
sense standards, two people differing widely in their possession of
talents (such as musical, mathematical, or tennis aptitude, or a flair
for poetry) can yet be judged to be equally intelligent. In such a
case they will be found to possess an equal amount of g. When the
amount of g in any person given is low, then that person will only be
indifferently successful in tasks requiring reasoning ability, which
depends very largely on g, while he may, nevertheless, be an out-
standing musician, draughtsman, engraver, carpenter, or billiards
player. Up to a point, a lack ofg is no handicap for the latter pursuits,
which are far more dependent on s factors than on g for their execu-
tion. Itcan so happen that in a person with high the s factor required
for success in mathematical ability is poorly developed. In such a
case, the type of intellect which shines in the sphere of language
activities (the classical scholar, barrister, theologian) may result.
Actually, high g is almost always associated with sufficiently well
developed s factors for language and number to make a high level of
conceptual reasoning possible. Highg, apparently, compels the fullest
possible realization of whatever potentialities for number and lan-
guage ability happen to be innately given, whereas it does not usually
compel the complete realization of other special factors such as those
determining skill in tennis or music. This follows from the fact that
the only way in which intelligence can be brought to bear on the
major problems of existence is through conceptual reasoning, for
which language and number ability are indispensable, whereas other
skills are not. Their development is largely a matter of chance, de-
pending on what happens to be valued in a particular community.
At first, Spearman made no attempt to define the nature of g
psychologically; he looked upon it as a value which could be found
wherever certain mathematical criteria were satisfied. Later on,
however, he came to regard g very tentatively "as measuring some-
thing analogous to an energy" ; that is to say as "some force capable
of being transferred from one mental operation to another different
one".' He held that it was constant for any individual, but varied
greatly from individual to individual,,and that it was particularly
involved in the process of gaining knowledge and understanding.
The s factors, he thought, could be looked upon as the engines
through which mental energy discharged itself, each engine being a
particular neural system. People differed not only in respect of the
basic neural integration with which they were endowed, but also
according as they had, by training and experience, developed their
THE NATURE OF INTELLIGENCE 5
The "Two-Factor" theory was received with great enthusiasm
in most psychological quarters. It provided a theoretical basis for
the current practice of intelligence-testing. If what really mattered
was the presence of a sufficient'amount of g or "general ability",
rather than a sufficiently large number of specific skills, then all the
intelligence-tester had to do was to determine the level of g, through
the medium of a few abilities known to depend on g to a consider-
able extent. This was actually what intelligence-testers had been
doing all along in a crude fashion, being frequently criticized for their
presumption in judging man's intelligence on the basis of a few
paltry and rather specialized tests (such as "opposites", "analogies",
"classification", and "story-completion" tests). Spearman's work
fully justified this practice, but it also provided methods of refining
test techniques and of selecting tests most sensitive to differences
in amount of g. The hypothesis of g also gave meaning to the con-
cept of the Intelligence Quotient, a single figure to indicate the
intellectual efficiency of the mind. With this new theoretical backing,
the I.Q. could be used with confidence as predictive of the measure
of success one would be likely to achieve in those pursuits commonly
assumed to require "intelligence".
To the theoretical psychologist, the teacher, and the man in the
street, the Two-Factor theory, in one or other of its aspects, made
sense and proved useful. It was this usefulness in bringing order
into a subject where hitherto thinking had been confused, and
theory, practice, and experience had frequently been in conflict,
which largely determined the widespread acceptance ofg or "general
ability" as an explanatory principle.
The theory has been subjected, however, to a variety of criti-
cisms which, particularly of late, have grown in number and impor-
tance. These criticisms will have to be considered before the theo-
retical assumption.on which to base the discussion of the nature and
growth of African intelligence can be determined.
It has been argued that, though g may be essential to success
in activities which involve deductive or inductive reasoning, an
ability of quite a different kind-let us call it social intelligence-
is required to deal effectively with practical life situations.
This criticism, which challenges the universality of g, rests on
a misapprehension both of Spearman's position and of the function
of intelligence. According to Spearman, all possible operations of
knowing, all cognition, can be reduced to three mental functions,
namely, direct apprehension of one's' own experience, education of
relations, and education of correlates.*
Any intelligent act whatsoever must, therefore, resolve itself
into these three processes.
Spearman then proceeds to show that every type of eductive
process is involved in those 'abilities known to have g as their sole
common factor. He holds, therefore, that it is justifiable to equate g
with the ability to educe correlates and relations and to know one's
own experience. The factor g, which thus becomes universal in
nature, can therefore be justifiably equated with general ability, no
matter in what sphere, social, personal, scientific, etc., it manifests
itself. The criticism that social intelligence requires something
more than g springs from an erroneous identification of deductive
and inductive reasoning with eductive processes in general. Such
reasoning is but one of the many forms which eductive processes
One might conceivably disagree with Spearman's position on
the grounds that he has given a logical rather than a psychological
analysis of cognition. But this error, if error it is, will be similarly
involved in his analysis of the activities in which g is known to be
the determining factor. It does not, therefore, affect the universality
of g. Provided that every form of knowing is reducible to eductive
processes, as defined by Spearman, and provided that all eductive
processes involve g, g must be universal. Whether this universal
factor is described as education of correlates and relations, mental
energy, or plastic function of the nervous system, depends upon the
point of view, logical, psychological, or physiological, which one
happens to adopt. Spearman nowhere argues thatg is the sole factor
*For the non-psychological reader, these terms may require some explanation.
By "apprehension of one's own experience", Spearman means the power to
observe what goes on in one's mind. One not only knows, feels, and wills, but one
also knows that one knows, feels, and wills.
The educationn of a relation" takes place when one notes that a star in a frosty
sky glitters like a diamond, that a block of flats is larger than a house, and that, if
all Africans are black, and a Tembu is an African, a Tembu must be black. In
each case, the thinker brings to mind some essential relation springing from the
very nature of the object thought of.
The educationn of a correlate" takes place when a person looking at a pint
bottle, and thinking of "larger than", finds the image of a gallon jar coming to
mind. Spearman formulates this law as follows : "When a person has in mind any
idea, together with a relation, he has more or less power to bring up into mind the
correlative idea." 6
These laws have been fully worked out in Spearman's The Nature of Intelligence
and the Principles of Cognition 3, and are briefly referred to also in The Abilities of
THE NATURE OF INTELLIGENCE 7
determining success in everyday life situations. But then intelligence
and social adaptability are not synonymous terms. Obviously, factors
other thanintelligence, such as perseverance, which is an expression
of the individual's character, or emotional stability, which depends
on the integration of the personality as a whole, may exercise a deter-
mining influence in bringing about successful adaptation. Those
who draw a distinction between the types of intelligence represented
by problem-solving and "social intelligence" are using an unanalysed
concept which can probably be split up intog plus certain personality
attributes, and possibly a particular set of s factors. Statistical studies
have so far failed to reveal the presence of a unitary trait correspond-
ing to social intelligence9, but there is a good deal of evidence on
the important part played by temperament and personality in deter-
mining the form in which intelligence expresses itself 10. It is scarcely
a criticism of g as a general cognitive factor to say that conative and
affective factors also play a part in successful adjustment. Spearman
makes no further claim forg than that, to the extent that intelligence
is able to determine success in life situations, those with high will
be more successful than those with low g.
A far more serious criticism of the Two-Factor theory of intel-
ligence made its appearance in 1938 with the publication of Thur-
stone's Primary Mental Abilities 1. For some time Thurstone had
been perfecting a technique of mathematical analysis, based on that
used by Spearman but considerably more complex, and capable of
extracting, not one factor, but a number of factors, determining the
intercorrelations between any number of tests simultaneously 12
Applying this new method to a battery of no less than 56 tests, in-
cluding many tests believed by Spearman to be good measures of g,
he found no evidence for the existence of any general common factor
in this battery at all. Instead, he extracted 12 group-factors, each of
which entered, to a greater or lesser extent, into a small number of
tests, and not at all into others. By studying the nature of the
tests into which each factor entered, Thurstone was able to identify
seven of the twelve psychologically. They were : (1) capacity for
visual and spatial imagery; (2) capacity for finding or recognizing
particular items in a pereceptual field ; (3) numerical ability; (4)
ability to grasp verbal relations ; (5) capacity to manipulate words
with fluency; (6) goodness of memory; (7) inductive reasoning.
The available data were inadequate to enable him to name the other
8 AFRICAN INTELLIGENCE,
Acceptance of Thurstone's findings would, in effect, take us
back to the old faculty theory, with this difference that, whereas
there was no proof whatever for the existence and independence of
the former faculties, these new faculties rest on a solid basis of
experimental fact and mathematical reasoning. Obvious difficulties
encountered by the previous faculty theory-such as the lack of
unity in "memory", memory for faces, names, numbers, tunes, etc.,
being virtually independent of each other-can be provided for on
Thurstone's theory by various combinations of factors. Thus,
memory for numbers would result from a combination of the memory
with the number factor, memory for names from a combination of
the memory with the verbal factor, and memory for faces from the
memory and perceptual factors.
Thurstone believes that these factors are not merely mathematical
entities, but that they represent primary mental abilities which are
genetic in character.
Thurstone's theory lacks the attractive simplicity of the Two-
Factor theory. Differences in intellectual power, instead of reducing
themselves to quantitative variations in a single entity, general ability,
now prove to be the outcome of particular selections of primary
mental abilities, combined in various strength ratios. Hence, differ-
ences in amount of intelligence must inevitably also become differ-
ences in kind, and the useful distinction between general intelligence
and talents therefore vanishes. A difference of 20 points in I.Q.
between persons A and B, A and C, A and D, might mean vastly
different things in the three cases, so that the I.Q. would also lose
much of its scientific value. This would necessitate a revision of our
methods of intelligence-testing. Undoubtedly, greater usefulness
will have to be sacrificed for greater scientific validity, for, without
the latter, the advantage of the former may be purely illusory.
In the present state of our knowledge it is hardly possible, however,
to claim validity for Thurstone's theory. His mathematical procedure
is such as virtually to preclude the Spearman solution, and the hypo-
thesis on which this procedure rests is open to criticism. Hence, at
the present moment, it becomes very much a matter of choice which
of the two rival points of view one adopts.
Some guidance is provided by a recent assessment of the'position
made by Professor Godfrey Thomson13 in the course of a symposium
on the factorial analysis of ability. Thomson has always agreed with
Spearman that "g is, of all the factors of the mind, the most ubiqui-
THE NATURE OF INTELLIGENCE 9
tous"14, provided gis interpreted as a mathematical entity and judge-
ment is suspended as to whether it is anything more than that.
He argues that there is no proof that any physical entity exists which
can be identified with, but, as far as the experimental evidence goes,
there is some aspect of the causal background which acts "as if" it
were a single unitary factor in the usual battery of intelligence tests15.
All along, his view has been that the mind is capable of forming
vast numbers of bonds, presumably of.a neural kind.
He states : "Thinking is accompanied by the excitation of these
neurones in patterns.... Intelligence is possibly associated with the
number and complexity of the patterns which the brain can (or
He distinguishes between simple and complex patterns and
admits that it is "the hope of the intelligence tester that two
brains which differ in their ability to form readily and clearly the
comparatively simple patterns required by his test will differ in
much the same way, if, given the same educational and vocational
environment', they are later, called upon to form the much more
complex patterns there found"17. The apparent presence of a com-
mon intellectual factor in such abilities as are involved in intelligence-
tests, he ascribes to the fact that any two tests probably'draw on a
large sample of these simple patterns from the pool of the mind.
At the same time he recognizes the existence of sub-pools, drawn on
exclusively by some tests, and not by others, and thus giving rise
to group factors. Thomson has always been one of the most vigorous
critics of Spearman for his refusal to admit the existence of a sufficient
number of group factors. The following quotation from Thomson's
summing-up in the symposium already referred to states the present
"Surely the real defence ofg is that it has proved useful. It still
remains to be seen whether Thurstone's primary factors will prove
equally useful. A jury in this country would probably to-day give a
verdict to Spearman; a jury in another place might give a verdict
"The larger jury of the future will, I think, decide by noting
which system has proved more useful in the hands of the practising
psychologist. I myself lean at the moment more towards Spearman's
g and his later group factors than I do to Thurstone's, since they seem
to me more in accord with the ideas of my own sampling theory. On
that theory, g is, as it weie, the whole mind, and tests are part of g, not
g part of the test. And were that mind entirely undifferentiated, struct.
ureless, g would be the only factor we need. As the complexity of the
mind and the complexity of the upper brain is organized (partly, by
the maturing of hereditary bonds, mainly, I fancy, by education and
life) and integrated into 'pools', 'clusters', call them which you will,
so additional factors, additional descriptive coefficients, are needed.
"It seems to me at present wise to retain one coefficient to express
the general depth, starting from which the integration, the deepening
into sub-pools has gone on. But I am not sure, and I think the wiser
course is to await further papers fromworkers inThurstone's school." 18
Any study of African intelligence, particularly if the comparison
with the intelligence of Europeans is its chief object, will have to
make some choice between the various theories of intelligence "now
in the field. This difficulty might be short-circuited by taking
intelligence-test results as the basis of discussion and by tracing the
factors which, in'each racial group, influence the test results. It is
doubtful, however, whether this procedure would in the long run
prove satisfactory. If racial differences in intelligence should turn
out to be qualitative as well as quantitative, some method of evalua-
tion will have to be selected, and this would prove difficult with only
an intelligence quotient as a guide. Most discussions of interracial
differences in test-intelligence must inevitably turn on the relative
influence of nature and nurture on the growth of intelligence.
Again, there is no intrinsic connexion between malnutrition, home
environment, parental attitude, and cultural milieu, on the one hand,
and intelligence-test scores, on the other hand ; but there is between
these environmental factors and the abilities which actuallyenter into
the test performance. Some theory regarding the nature and inter-
relation of these abilities is therefore essential. And, finally, our
ultimate interest is not in intelligence-test scores, but in mental
processes, their attributes and antecedents. If there should prove to
be real and fundamental differences between the minds of Africans
and those of Europeans, it would be strange indeed if a mere index
was able to summarize these differences adequately, however useful
this index might be in predicting successful adaptation to European
culture in the case of Europeans themselves. A detailed statement
of the assets and liabilities, the potentialities and drawbacks, of the
intellectual processes of Africans might serve as a far more useful
basis for the formulation of an educational policy and for the objec-
tive assessment of the future of Africans in Western civilization than
an estimate of the range of African intelligence quotients.
THE NATURE OF INTELLIGENCE
In this book, therefore, the following hypothesis has been
adopted: differences in intelligence will be looked upon as due to
differences in the amount of general ability represented by g, and in
the number, kind, and development of certain broad special abilities,.
represented by group factors. This g will be conceived of rather in
the light of Thomson's than of Spearman's hypothesis. This means
that Spearman's s factors, the special abilities almost entirely specific
to single skills and aptitudes, will be very largely replaced by the far
broader group-factors. Any one or more of the latter can be absent
without any considerable quantitative effect on intelligence (though
such quantitative effects are by no means excluded). In each case,
genetic make-up determines the limits of development, both of the
group-factors and ofg ; but, whereas the full realization ofg is depen-
dent only on certain general environmental conditions, that of the
special abilities is dependent chiefly on specific training and cultural
1. Spearman, C. "General Intelli-
gence, Objectively Determined
and Measured." Am. J. of
Psychol., V. XV, 1904. pp. 201-
2. Spearman, C. The Abilities of
Man. London, Macmillan, 1927.
3. Ibid. p. 82.
4. Ibid. p. 242.
5. Ibid. p. 414.
6. Ibid. p. 166.
7. Spearman, C. The Nature if In-
telligence and the Principles of
Cognition. London, Macmillan,
8. Ibid. Ch. XI. 1923.
9. Thorndike, R. L. and Stein, S.
"An Evaluation of the Attempts
to Measure Social Intelligence".
Psychol. Bull. V.XXXIV, 1937.
10. Webb, E. "Character and Intel-
ligence". The Brit. J. of Psychol.
Mono. Supp., No. 3. 1915.
Garnett, J. C. M. "General Ability,
Cleverness and Purpose." The
Brit. J. of Psychol., V. IX, 1918.
Spearman, C. Op. cit. ref. 7.
11. Thurstone, L. L. The Vectors of
Mind. Univ. of Chicago Press,
12. Thurstone, L. L. "Primary Mental
Abilities". Psychometric Mono.
No. 1., Univ. of Chicago Press,
13. Thomson, G. H. "The Present
Position and the Problems Con-
fronting Us." Symposium on
"The Factorial Analysis of
Ability." Brit. J. of Psychol.,
V. XXX, pt. 2, 1939.
14. Thomson, G. H. The Factorial
Analysis of Human Ability. Univ.
of London Press, 1939. p. 240
15. Ibid. p. 227.
16. .Ibid. p. 51.
17. Ibid. p. 52.
18. Thomson, G. H. "Analysis: a
Summing Up." Op. cit. ref..
13. p. 106.
THE MEASUREMENT OF INTELLIGENCE
THE PIONEERING WORK OF ALFRED BINET
The practice of intelligence-testing is closely associated with the
name of the French psychologist, Alfred Binet, whose first intel-
ligence-test was published in .1905. It was designed to pick out
from the school-going population of France those pupils who stood
in need of special tuition, because they were by nature incapable of
profiting from the usual class-room instruction. The success of his
efforts encouraged him to elaborate his test in such a way as to make
it suitable for the quantitative diagnosis of individual differences in
In 1908, he published a scale consisting of a number of tests
arranged in groups for various age-levels from 3 to 13 years, there
being from four to eight tests for each age-level. Binet arrived at
the location of a test at a particular age-level by a process of trial and
.error. Starting with a large number of likely tests, varying greatly in
difficulty, he applied these to an unselected group of children be-
-tween certain agreed age-limits, assuming that, as a result of the
randomness of his choice, this sample represented a more or less
true cross-section of the population as a whole. By rating the pro-
portion of children of each age which passed a given test, he was able
to locate that test at its appropriate age-level. If, for example, all
children aged 6 failed in a particular test and all 10-year-olds
passed it, then, obviously, that test was too difficult for the former
age and too easy for the latter. It could, however, be considered a
fair index of average ability in that age-group where from one-half
to three-quarters of the children passed it. Unless a test was passed
by a regularly decreasing number with increase in age, it was rejected
as not being diagnostic. Thus in the 1908 scale 58 tests were retained.
To score the tests the following procedure was followed. First,
the highest year for which all the tests were passed was taken.
To this age, one-fifth of a year was added for each additional test
passed in any higher age-level. Thus, if a child of seven passed all
the tests for the 6th year, three tests for the 7th year, two for the
THE MEASUREMENT OF INTELLIGENCE 13
8th year, and one for the 9th year, its total credit would be 7, years.1
(This procedure of allowing fractions was not adopted until the
publication of the 1911 scale, in which there were five tests for each
age-level with the exception of the 4-year level, at which there were
To the total credit Binet gave the name of 'mental age'..
Let us be quite clear as to what this mental age represents.
It indicates the degree of mental development which one can expect
a normal, or average, child to possess at a particular age-level.
A normal, or average, child is one with such ability as is most fre-
quently encountered in the community as a whole, and above and
below which, at the same age-level, there is an equal number.of
superior and inferior children respectively. If a child with a chrono-
logical age of 10 has a mental age of 7, his ability is such as one
would expect to find in a normal child of 7. He-is, therefore, three
years retarded below what is normal for his age. With a mental age
of 13 he would be 3 years advanced.
THE GROWTH OF INTELLIGENCE
Binet assumed that the growth of intelligence continued until
the 16th year. The difference between the mature adult and the
adolescent, he held, was not a difference in the absolute capacity to
invent, to direct, to comprehend, and to judge, but a difference in
knowledge and experience. As a rule, the mature adult would have
far more wisdom than he had when his intelligence reached its full
stature at the age of 16, depending, of course, on the way in which
that. intelligence has been used to acquire knowledge during the
intervening years. He would be able to achieve a deeper compre-
hension and truer judgement, only because his intelligence could
exercise itself on more relevant material. Provided, therefore, that
the correct distinction was made between intelligence and wisdom,
there was for Binet nothing ridiculous in the statement that the
average adult had the same intelligence as the average child of 16..
Whilst this point of view is generally accepted by all psycho-
logists, there is some difference of opinion regarding the precise age
at which the growth of intelligence comes to an end. The age most
commonly accepted as that .at which intelligence reaches its full
stature is 16 years. Some hold that in the case of those with inferior
ability growth may cease well below 16 years, so that their develop-
ment not only proceeds at a lower level, but also ends sooner.
14 AFRICAN INTELLIGENCE
Superior children may continue to develop up to 18 and possibly
even beyond that age. It follows that when a study of the intelli-
gence of a social group, nation, or race has to be made, it will suffice
to examine representative samples of the youth of the population
concerned. Nothing is gained by the inclusion of adults ; not only is
it difficult .to collect sufficiently large groups of the latter to make
testing worth while, but their attitude towards the intelligence-test
is usually unfavourable, and the results obtained from them far less
reliable than those from children.
THE INTELLIGENCE QUOTIENT
By itself, the statement of a child's mental age is of little value.
Though the mental age indicates its mental capacity, it tells one
nothing of the normality or abnormality of that capacity. An accom-
panying statement of the chronological age is also required.
This drawback was remedied by Professor L. M. Terman, of
Stanford University, California, whose revisions of the Binet Scale,
commonly referred to as the Stanford-Binet, have been most widely
used. He combined mental and chronological ages into one coeffi-
cient by expressing the former as a percentage of the latter. It was
to this coefficient that the name "Intelligence Quotient" was given.
The normal child, i.e., the child whose mental age is equal to his
chronological age, will have an I.Q. of 100. Retardation is indicated
by I.Q.s below 100, advancement by I.Q.s above 100.
This way of expressing the result of an intelligence-test is a
great step forward, because it enables one to see, at a glance, what
potentialities for intellectual development a child possesses, no
matter at what age it is tested. Instead of indicating amount of
intelligence, it indicates inferiority or superiority with reference to
some common standard.
This concept has proved so useful that it has been retained for
all subsequent intelligence tests, even those for which no mental
age can be directly calculated, as scoring is not done in terms of
months but in number of items correct. The statistical devices
employed to turn raw scores into an Intelligence-Quotient scale
need not detain us here. Suffice it to say that the general idea of the
I.Q., as an index of relative intellectual superiority or inferiority,
has remained the same for all tests, regardless of the method of
THE MEASUREMENT OF INTELLIGENCE
THE CONSTANCY OF THE INTELLIGENCE QUOTIENT
The usefulness of the Intelligence Quotient as an index of the
ultimate intellectual capacity of an individual obviously depends on
This constancy is in the first place determined by the mainten-
ance of a constant rate of growth of intelligence, relative to that of
the mean of the population. If a child with an I.Q. of 80 at age 6
could suddenly put on a spurt at age 15 to finish strongly at age 18
with an I.Q. of 120, then the concept of the I.Q. would lose all value
and there would be no point in going beyond a statement of mental
age, purely as a measurement of current intellectual status. Actually,
there is good evidence that, normally, the rateof growth is more or
less identical in all individuals. The curve of growth being a smooth
parabola of negative acceleration, it would appear that no hereditary
factor could make the constancy of the I.Q.*
To what extent this statement is true only of a tendency,
observable in the mass, and not of individual cases, it is impossible
to say. The occurrence of individual acceleration or deceleration of
hereditary origin, altering the I.Q. to any marked extent, is, how-
The second assumption underlying the hypothesis of the
constancy of the I.Q. is that the natural rate of growth, as established
by heredity, is incapable of being influenced by environmental
circumstances, no matter how extreme these might be.
The extent to which the I.Q. is the product of both nature and
nurture is one of the most controversial issues in contemporary
psychology. The tide has, however, set strongly against the nativists.
By means of a number of carefully conducted investigations,
it has been established beyond all doubt that numerous environ-
mental influences do materially affect the magnitude of the I.Q.
Quantitative estimates have been made of the extent to which
nurture can depress or enhance the realization of congenital poten-
tiality. These estimates will be discussed in detail in a later chapter.
For the moment it will suffice to note their general trend.
One of the most outstanding of these investigations is that con-
ducted by Barbara Burks2 on the intelligence of a group of foster-
The only exception to this statement occurs in the case of certain rare heredi.
tarily determined forms of mental defectiveness, such as Tay-Sachs Disease and
Amaurotic Family Idiocy, where a progressive deterioration in intelligence occurs,
accompanied by physical changes which end in death.
children. These children were placed before the age of twelve months
into foster-homes varying in cultural quality. The relationship be-
tween their I.Q.s and the cultural level of their foster-home environ-
ment after residence of from 5 to 14 years in it was compared statis-
tically with the relationship obtaining in a control group of parents
and their natural children. The findings were as follows:-
"Measurable environment one standard deviation above or below
the mean of the population does not shift the I.Q. by more than 6 to 9
points above or below the value it would have had under normal en-
vironmental conditions. In other words nearly 70 per cent of school
children-American White-have an actual I.Q. within 6 to 9 pointsof
that represented by their innate intelligence.
"TTle maximal contribution of the best home environment to intel-
ligence is apparently about 20 I.Q. points. Conversely, the least cul-
tured, least stimulating kind of American home-environment may
depress the I.Q. as much as 20 I.Q. points But situations as extreme
as either of these probably occur only once or twice in a thousand times
in American communities." 3
These findings are substantially borne out by Freeman, Hol-
zinger, and Mitchell," who found a correlation of .48 between the
child's I.Q. and the cultural rating of its foster-home; an average
increase of 9 points in the intelligence of one sibling, placed in a
superior foster-home, over that of another placed in an inferior oie
(data obtained from 130 pairs of siblings) and who concluded that
superior environments may increase intelligence by 10 points or
more. Lithauer and Klineberg5 tested 120 Jewish children before
admission to an orphanage where environmental conditions were
superior to those the children were used to, and re-tested them after
a stay of from 3 to 7 months. The mean improvement in I.Q. for
the whole group was 5.9 points.
These investigations all concentrated on measuring the effect
of home-environment on the growth of intelligence; but quite
clearly that is not by any means the only environmental influence to
which this growth is subjected. (The indirect effect of parental
intelligence is included under the general heading of home-environ-
ment.) Burks admits that a further contribution can be made by
"random somatic effects"6 of environment, such as nutritional and
health factors, though this contribution is not considerable. She also
stresses the fact that her data are relevant only for "White American
children living in ordinarily variable circumstances. If variation
THE MEASUREMENT OF INTELLIGENCE
is greater in another cultural milieu, correspondingly greater shifts
in I.Q. may occur. In a community with schools varying widely in
quality, the measure of the home-environment alone would also be
misleading. The total possible effect of environment on test-intel-
ligence is therefore likely to be considerably greater than the effect
measured by Burks. We shall return to a detailed consideration
of this matter in a subsequent chapter.
Here in South Africa, the findings of the Carnegie Commission
of Investigation on the Poor White problem offer further general
corroboration. Wilcocks8 found a steady decline with age in I.Q.s
of Poor White children.
Whereas the average I.Q. for a group of 356 children, aged 120-
123 months, was 98.5, it was 94.2 for those aged 152-155 months.
This fall undoubtedly continued until cessation of development,
yielding an ultimate I.Q. probably well below 94. After extrapolating
a logarithmic curve fitted to the observed data, Wilcocks suggested
that the fall must have been slower before 120 months, and that the
I.Q. at birth might have been in the neighbourhood of 107, though
he added that, for statistical reasons, much value could not be attached
to this figure. In a group of rural Non-Poor-White children, the
average I.Q. was found to be 104.5 at age 120-125 months and 105.2
at age 152-155 months, which suggests that his calculated I.Q. is
probably not far out.
The general result of these investigations is that, though there
may be some disagreement about the extent to which nurture can
influence the rate of growth of intelligence, there can be none about
the fact that it is a relevant determining factor. Hence, to a hitherto
unknown extent, the I.Q. must vary with the particular environ-
mental circumstances to which an individual or group has been sub-
jected during the years of growth.
WHAT DO INTELLIGENCE-TESTS MEASURE ?
It can be safely assumed that the intelligence of man derives
ultimately from some innately determined potentiality, different
from individual to individual. This potentiality is presumably given
in the germ plasm. From the moment of its existence environmental
influences are capable of affecting this germ plasm, more particularly
after the union of sperm and ovum, when a new organism begins, to
react to its intra-uterine environment. These environmental factors,
though far less variable than post-natal ones, are of vital importance
in determining development. It is a mistake to look upon growth,
particularly embryological and foetal growth, as the mere unfolding,
in an inevitable and relentless fashion, of that which was innately
given. Growth is the product of interaction between an organism
and its environment, and as the nature of that interaction changes,
so will the growth processes change. Thus mongolism is surmised
to be due to an environmental influence operating at the germinal
stage9, while feeble-mindedness can also result from the intra-
uterine syphilitic infection of an otherwise normal foetus. These are
gross effects, produced by drastic abnormalities in pre-natal environ-
mental conditions. There are bound to be less marked variations in
the influences to which the foetus is normally subjected, and it is
likely that these, too, materially affect its growth. Severe malnutrition
on the part of the mother during pregnancy, for instance, is recog-
nized* to have an effect on the physical growth of the foetus.
The development of the nervous system, which is the ultimate
physical basis of intelligence, may suffer from the resulting general
stunting, leading to a depression of congenital potentiality which may
have permanent effects. The congenital is, therefore, always a
product of the innate and the environmental, and it is a difficult
task to indicate to what extent it represents the normal realization of
genetic make-up. This difficulty is increased a hundredfold by the
neonate's subsequent interaction with the post-natal environment.
Growth continues after birth as before, but under far more variable
Even those, therefore, who genetically possess the same poten-
tiality for intellectual development and in whom at birth that poten-
tiality is equally realized, may still considerably diverge, during the
first years of life, according to the circumstances under which they
We are, therefore, compelled to look upon an intelligence-test score,
not as a direct measure of innate ability, but as a measure of hereditary
potentiality as it happens to have been realized by specific environmental
circumstances. It must, furthermore, be remembered that the test-
situation has a refractory index of its own, affecting different people
differently. There is, as yet, no test which is equally suitable for
people differing widely in their cultural background. It is possible
that emotional factors, such as apprehensiveness in the test-situation,
play a part in depressing the score of those liable to suffer in this way.
THE MEASUREMENT OF INTELLIGENCE
Suspicion about the motives of the tester may produce an uncoope-
rative attitude, which is bound to influence the test results adversely.
Because intelligence-test results are therefore the product of a
multiplicity of variables, of which innate intelligence or general abi-
lity is but one, it has become customary to speak ofscores as indicating
differencesin test-intelligence, rather than in intelligence. The signifi-
cance of this test-intelligence is a matter for interpretation, the possible
effect of environmental influences being assessed in the light of the
available facts regarding the individual and group concerned.
THE SIGNIFICANCE OF TEST-INTELLIGENCE
It is, nevertheless, possible to use the I.Q. as an index of innate
potentiality, but only subject to certain conditions. When the
individuals tested are drawn from a closely homogeneous group,
within which environmental influences are approximately identical
and equivalent in their operation, it is legitimate to ascribe observed
differences in test-intelligence to heredity. In this case, the differ-
ences between I.Q.s, rather than their magnitude, are important.
Great care must be taken to make sure that the group is in fact
homogeneous and that only minor individual, variations in home
circumstances and upbringing do occur within it.
The grouping together of children from middle- and working-
class suburbs in the same town would be inadmissible, despite the
facts thit these children went to Government schools, belonged to
the same race, and had all been subjected to urban influences.
Admittedly; the differences in this case would not be great; but a
difference there would be, as we have already shown, and as will
be shown in greater detail in subsequent chapters. This difference
may not be regarded as irrelevant because it' is slight, or because it
is difficult to control.*
When the groups to be compared differ only socio-economically,
and not culturally and racially, the average observed difference in
It cannot be too strongly emphasized that, if psychology wants to be a
science, on a par with the natural sciences, it must accept the disciplines which are
the essence of these sciences. Precision and careful control of conditions are as
essential in the psychological as in other scientific fields; even more so, as the
subject matter tends to be more vague and elusive, and slipshod work can get by
The temptation to measure in a rough and ready way, to get things "more or
less", should be strongly resisted. It is better that a problem should be left alone
until precise methods of dealing with it are found, than that it should be tackled
in a manner which makes a show of being scientific. Such a course can only lead
to the ultimate discredit of psychological experimental method.
test-intelligence between them may legitimately be looked upon as
a difference in actual, available, as distinct from innnate, intelligence.
Such a statement can be quite valuable for various purposes, provided
its meaning is clearly defined.
When, however, a cultural divergence is added to the socio-
economic one, the interpretation of the data becomes increasingly
It does not quite follow that, when there is a high correlation
between test-intelligence and intelligent adaptation to a particular
culture, those having a low test-intelligence must inevitably show an
equally low adaptation to the intellectual demands of that culture.
For, as we shall see, the test-situation may exercise such an adverse
effect on the test-intelligence of those belonging to another culture,
that even such insight into life situations as they can normally muster
fails to show itself in the answering of the test problems. Unless
one has a clear realization of the significance of test-intelligence,
an extraneous influence of this type may easily be overlooked, and
the test group appear to be more stupid than it actually is.
Thus Fick's'1 statement that "the inferiority of the native in
educability... limits considerably the proportion of natives who
can benefit by education of the ordinary type beyond the rudimen-
tary" might have serious practical consequences. It will be shown,
that among many other missions, Fick failed to allow for the dif-
ference in attitude on the part of European and Native children
towards the intelligence-test situation. Actually, Fick made the case
considerably worse by adding a corollary to the effect that "the
inferiority does not appear to be of a.temporary nature"". Here, in
fact, lies the real danger of interracial comparisons. The motive
behind these investigations is, as a rule, to discover racial, and there-
fore genetic, differences. And where it'is impossible to disentangle,
with any degree of scientific certainty, what is innate and what is
acquired, the temptation to interpret the data in accordance with
political bias or emotional needs often becomes too strong.
The only safe procedure is to unravel, one by one, the various
environmental influences which might have a measurable effect on
the growth of intelligence; to study the magnitude of that effect,
and then either to hold these influences constant for the two groups
examined; or, else, where this is not possible, to make a suitable
correction in the results obtained.
THE MEASUREMENT OF INTELLIGENCE
THE CONTROL GROUP
For this purpose, a device known as the "control group" is
usually employed. An experimental group is carefully matched with
another in such a way that the only difference between them is that
one concerning which information is wanted. To achieve equi-
valence is by no means an easy task, as it demands precise know-
ledge of all the factors relevant to the problem under investigation.
In the case of interracial intelligence studies, differences of opinion
regarding the importance of certain environmental influences have
led to the use of control groups which failed to equalize important
factors, such as nutrition, certain aspects of the home-environment,
Apart from that, the environmental differences may well be so
great as to make the matching of two completely representative
groups an impossible task. In this case, the only way out is to make
a correction in the scores obtained, in accordance with what is
known to be the quantitative effect of various environmental factors
on the growth of intelligence.
But, this merely brings us back again to the old difficulty of
the lack of unanimity regarding the nature of the environmental
determinants of intelligence and the magnitude of their effects.
Before plunging into such problems as interracial differences
in intelligence, psychologists should therefore first solve the more
general questions. To tackle a problem in applied psychology while
.the general principles are not yet clearly established is to put the
cart before the horse.
INTERRACIAL INTELLIGENCE STUDIES IN SOUTH AFRICA
Both theoretical interest and practical necessity have prompted
a considerable number of interracial intelligence studies in South
Africa. Intelligence-tests have been used to provide scientific proof
for the' commonly held belief that, on the average, Africans are in-
nately less intelligent than Europeans, even when full allowance is
made for the difficulties of the transition stage.
Most notable among these investigations are those conducted
by Dr. M. L. Fick, Psychologist of the National Bureau for Educa-
tional and Social Research, who maintains that, around the ages of
13 and 14, Native children are from 4 to 5 years inferior to European
children in educability as gauged by the results of intelligence-tests.12
It is impossible to assess the validity of Fick's work without precise
knowledge of the effect of various environmental factors .on test-
intelligence. Fick has been severely criticized in the past, notably
at the New Education Fellowship Conference in Joharnesburg in
1934' but, while the extent to which the magnitude of the I.Q.
could be influenced by environment remained a matter of opinion,
this criticism was not as compelling as it might have been. It did
not deter Fick from publishing the results of another investigation,
to which very much the same criticism could be applied.
It is the purpose of this book, first of all to state clearly, in the
light of all available data, to what extent the growth and measure-
ment of intelligence can be determined by factors other than heredity.
These factors will be studied under the following headings : Cultural
Milieu, Home-Environment, School Environment, Nutrition, and
Temperament. Thereafter the situation of the Union African
population will be studied with reference to each of these factors,
and the possibility of using the control-group technique will be
Only then will it be possible to assess the validity of Fick's use
of this technique, and of the manner in which he has interpreted
1. Peterson, J. Early Conceptions and
Tests of Intelligence. Chicago,
World Book, 1925.
2. Burks, B. S. "The Relative In-
fluence of Nature and Nurture
upon Mental Development".
In: 27th Yearbook of the Nat.
Soc. for the Study of Education,
1928. Ch. X.
3. Burks, B. S. Ibid. pp. 300-09.
4. Freeman, F. N., Holzinger, K. J.;
and Mitchell, B. C. "The In-
fluence of Environment on the
Intelligence, School Achievement
and Conduct of Foster-Child-
ren". Op. cit. ref. 2. Ch. IX.
5. Lithauer, D. B. and Klineberg, O.
J. of Genet. Psychol., V. XLII.
6. Burks, B. S. Op. cit. p. 303.
7. Ibid. p. 308.
8. Wilcocks, R. F. "The Poor White
Problem. in South Africa".
In: Carnegie Commission Report,
V. II. Stellenbosch, Pro Eccle-
9. Rosanoff, A. J. Manual of Psy-
chiatry and Mental Hygiene.
New York, Wiley, 1938. Ch.
IX, pp. 227-29.
10. Fick, M. L. "The Educability of
the South African Native." In
S.A. Council for Educational and
Social Research, Pretoria. Re-
search Ser. No. 8. 1939. p. 56.
12. .Ibid. p. 54.
13. Malherbe, E. G. ed., Educational
Adaptations in a Changing So-
ciety. Cape Town, Juta, 1937.
THE SAMPLING PROBLEM
In order to determine group-differences in intelligence, suitable test-
samples of the groups concerned must first be selected. It stands.
to reason that the application of intelligence-tests to all Africans and
Europeans in the country would be an impossible task. It would
furthermore be a grievous waste of time. An accurate estimate, not
only of the differences between the average performance in the two
groups, but afso of the probable upper and lower limits in each case,
can be obtained from representative samples.
Before dealing with the statistical methods available for this
purpose, let us first consider what makes a sample representative.
Numerical size and randomness of selection are often held to be
first considerations. Whenever a certain quality is present in a popu-
lation in widely varying amount, a small sample drawn from that
population is not likely to reveal the full range and nature of its
distribution. How large, however, with respect to thisquality, should
our random sample be, in order to give a faithful reflection of the
composition of the population as a whole ?
Some writers have endeavoured to justify the size of their sample
by demonstrating the identity between the mean and standard
deviations of the sample as a whole and parts of the sample.*
Only then does the close correspondence between these two
measures indicate that nothing can be gained by increasing the size
of the sample, when the original sample has been selected in a truly
random fashion or is known to be fully representative of the com-
munity as a whole. For we shall always get close correspondence
in the case of homogeneous samples, and it is comparatively easy
The S.D., or "standard deviation", is a statistical measure which indicates
the limits between which approximately two-thirds of the individuals in the sample
are to be found. Thus, if the mean I.Q. of a particular sample is 105, with an
S.D. of 6, we know that approximately two-thirds of the cases in our sample lie
between the limits 99 and 111. Approximately the entire distribution lies between
the limits established by laying off 3 S.D.s above and below the mean. In addition
to the S.D., which gives an idea of the range of that distribution, we can also
calculate the S.D. of the mean of a sample. It indicates the limits between which
two-thirds of all subsequent means of samples, similarly selected, are to be found.
From the S.D. of the sample-mean we can therefore get some idea of the probable
magnitude of the population-mean.
to select a sample in such a way as to achieve great homogeneity
without representativeness. For example, the scatter of I.Q.s in
a University class is far smaller than it is in Standard I of the Pri-
mary School, as all the lower I.Q.s have been eliminated by exami-
nation barriers. A regiment of Guardsmen will show remarkable
homogeneity as regards height, again because variation has been
deliberately excluded. Even where no obvious selective factors
such as these can be discovered, it cannot be assumed that, when
means and S.D.s of the whole and part of a sample agree, that sample
is representative of the community as a whole. It may or it may not
be. The test merely shows that there is no point in enlarging the
sample by means of the same selection procedure as was adopted to
obtain the initial sample.
Before he applies the above test, the investigator must therefore
be able toprove that his sample is truly random, and this is no easy task.
Suppose, for example, that we were required to find the range
of I.Q.s of all 10-year-old Europeans in this country. To be fully
representative, the experimental sample should contain boys and
girls, urban and rural pupils, in the exact proportions in which they
occur in the entire population of 10-year-olds. The sex distinction
is necessary because the growth of intelligence does not proceed at
exactly the same rate in the two sexes. With regard to the differen-
tiation between town and country, not only are the relevant environ-
mental conditions sufficiently different in the two cases to affect the
growth of intelligence, but there is also a possibility that the more
intelligent country-dwellers have gradually migrated to the towns,
where the scope for their diverse abilities is so much greater. This
would lead to a relative lowering of the mean rural I.Q., the gap
being made more permanent by in-breeding within the two groups.
In South Africa, the drift, of the rural poor to the towns would
provide a complicating factor.
As there is a significant relationship between socio-economic
status and intelligence, the sample to be selected should also be pro-
portionately representative of the chief employment groups in the
community. In the case of 10-year-olds, selection would be com-
paratively easy. Compulsory education ensures that every White
10-year-old child goes to school.* Hence one knows both the size
Mental detectives are an exception. The majority are to be found in special
institutions, and can therefore be taken into account. A very small percentage,
children of the well-to-do, escape detection.
THE SAMPLING PROBLEM
of one's group, and where to find it, while from the school record
at least some information is obtainable as to its socio-economic
'It would be otherwise in the case of 17-year-olds, of which
only a superior percentage are to be found in schools and colleges,
while the remainder is scattered about in a manner which would
make selection and testing very difficult.
Quite clearly, a representative sample, deliberately selected in
this way, is very far from random ; but it is doubtful whether with
known inhomogeneity in a community, randomness of choice can
be an effective selection method, unless huge samples are used.
Systematic selection for representativeness in respect of all those
factors known to influence the quality under investigation is the
only fully satisfactory and scientific way.
Once- a representative sample has been obtained, calculation
of the S.D. of the mean of that sample will indicate the limits be-
tween which the means of all other samples, similar selected, would
lie. Suppose for instance, that we find the mean I.Q. of 10-year-
olds in our sample to be 101, the S.D. of the mean being 2.5. This
enables us to say that 68 per cent of all subsequently found means
(provided the samples continue to be selected in the same manner)
will lie between 103.5 and 98.5. If a greater certainty is wanted, we
canr say that 98 per cent of all similarly determined means will lie be-
tween 108.5 and 93.5. When there is a difference between the means
of two samples representative of two different populations, it is possi-
ble, once again through the medium of the S.D.,to discover whether
this difference is chance, or whether it is likely always to occur in
the observed direction.* If the difference is three times the size of
its own S.D., the chances of its occurring in the opposite direction
are only one in a hundred, while if it is four times its own S.D., the
chance of a reversal is virtually nil. Such a difference is said to be
a real, or significant, one. If, for example, the average I.Q. of urban
10-year-old European children is 103, and of rural children 98,
with an S.D. of 1.2, we can say with confidence that the difference
is significant, and that urban children have a higher test-intelligence
The formula to determine the S.D. of a difference between two means is :-
S.D. -MM = VS.D.2 1 + S.D.2M
M1-M V M I M2
where S.D. is the S.D. of the first mean, S.D. is the S.D. of the second
mean. 1 M2
than rural children, as the chances are approximately one in a thou-
sand that a random sample should be found in which urban children
actually score a lower mean than rural ones.
How SHOULD A REPRESENTATIVE SAMPLE OF THE UNION AFRICAN
POPULATION, FOR THE PURPOSE OF INTELLIGENCE-TESTING,
BE CONSTITUTED ?
The various tribes which together make up the Union African
population display, wherever tribal life has remained intact, cul-
tural characteristics which to some extent are unique for each tribe.
Apart from differences in ritual and material culture, well-defined
differences in personality make-up may be noted. Whether these are
due to innate differences in temperament, to the type of socializa-
tion inherent in each culture, or to historical causes, it is difficult
to say. Do these personality differences also involve variations in
One frequently hears it stated, on the basis of experience, that
the Basuto are intellectually superior to other tribes. Can statements
such as these be scientifically substantiated ? The task should not
be a difficult one and could be carried out by means of existing intel-
ligence-tests. Provided the test used is sufficiently discriminative
to give a fairly wide scatter of scores, it should be able to serve the
purpose of intra-group comparison, as whatever imperfection' it
possesses will affect all tribes equally. Popular belief regarding
differences in intelligence between the tribes may be, and probably
is, unfounded; but until we know that it is, the experimenter who
desires to make statements regarding "Native Intelligence" will be
compelled to use representative samples of all the chief tribes. Un-
less this is done, there can be no certainty that a sampling error has
not crept into his data.
Of the total African population of the Union, approximately
1,361,000 live in urban areas, the rest, numbering 5,158,131, are in
Reserves, and on farms.* The majority of those living with their
families in urban areas are completely detribalized, the majority
of those in the country are not. -A fairly large number of men em-
Data from 1936 Census Report. The urban figures include 386,858 Africans
in Mine Corpounds, of which a number are from outside the Union. Data include
Africans on alluvial diggings, in road gangs and "other areas", numbering approx-
THE SAMPLING PROBLEM
played in the cities, either on the mines or in domestic service, keep
family establishments in some Native village in the Reserves. Fun-
damentally, therefore, their orientation remains tribal. In the Re-
serves,.on the other.hand, there is a number of families-those of
teachers, ministers, and clerks in Government service-who live
more like Europeans and who have lost all contact with the tribal
way of life. Consequently, the demarcation line between tribal and
detribalized does not correspond entirely with that between rural
and urban population.
A representative sample should contain both the thoroughly
tribal and the thoroughly detribalized in appropriate numbers. The
factors leading to detribalization are extremely complex, often for-
tuitous. There is a possibility that those who elect to settle in the
cities, or those who return to the country to practise there some trade
or profession they have acquired in the city, are somewhat superior
to those who, unaltered by their contact with a different culture, per-
sist in following time-honoured paths. According to the theory of
selective migration, the more intelligent, ambitious, and energetic
Africans, knowing that the cities offer greater opportunities for
work, for the earning of money, for education, and for the acquisition
and enjoyment of the amenities of European civilization, gradually
leave the tribal areas. After a few generations of intermarriage of
duller elements in the Reserves, and of brighter elements in the
towns, the latter stock would become progressively superior to the
It is extremely doubtful whether in'reality the progress of mi-
gration is determined in this straightforward way. In any tribal
community, there exist considerable differences in rank and status.
which may play an important part in deciding whether a man shall
migrate or not. Sooner or later, most men spend some time in the
cities, largely because it is only there that money-incomes can be
earned. Money is scarce in the tribal communities. It is needed for
the payment of taxes and for the purchase of such articles and food-
stuffs as have become essential to a tribal culture profoundly in-
fluenced by its contact with European civilization, Employment in
the towns may also be resorted to in order to save enough money to
purchase cattle for lobola. The significant fact, therefore, is not
whether a man goes to the towns from time to time, but whether
he returns to the country, and in what capacity he returns.
On the average, one would expect migration to be less frequent
among near relatives of chiefs, sons or brothers of headmen, and
others possessing important status in the tribal community. Though
these people may not be superior to the rank and file in ability, they
are not likely to be inferior, so that for themselves and their families
the theory of selective migration does not hold.*
Cattle are the chief form of private property. A man's wealth
is usually measured in terms of the number of cattle he owns. The
inheritance of cattle is usually from father to the first son of the
principal wife, except where for some reason or other this son has
been disinherited. Other sons may receive one or two beasts each,
and the principal heir usually gives them some at least of his father's
property, but there is nothing to compel him to do so. Though in
the tribal communities a man is only granted the use, and not the
ownership, of land, he has the right to pass on this use to his eldest
son, while other sons have other land granted to them by the head-
man. Here again, therefore, the eldest son is likely to be better off.
Though land tenure is somewhat different to-day, it is still true that
younger sons and sons of subsidiary wives will always be poorer and
may be compelled to go to the Cities sufficiently frequently to be
tempted to stay there.
Consequently, cultural factors, as well as personality qualities
such as adventurousness and intelligence, may determine migration
from tribal to urban communities. There may, therefore, be very
little difference in inborn ability between the tribal and the detribal-
ized. To be on the safe side, however, in the present state of our
knowledge, the investigator should include a sample group of tribal
Africans. This would, as we shall see, considerably add to his experi-
mental difficulties, but it would meet a criticism which, however
unlikely it is to be valid, cannot be finally disposed of except through
recourse to experiment.
2. URBAN-RURAL DIFFERENCE
A good deal of time has been devoted to determining differences
in intelligence between urban and rural populations. That differen-
ces in test-intelligence do exist has been amply demonstrated (Figure
1). The consensus of opinion to-day is that these differences are
environmental rather than innate. The opportunities available to
the rural child tq develop those abilities required for the answering
SThe writer was unable, through lack of time, to verify this statement. The
argument is put forward, not as a fact, but as a possibility.
2 1- 11.1 -1 -
\1 DISTRIBUTION OF 1L.s.
RURAL AND URBAN GROUPS.
\-- UnsAN CAS[S.
*--- ----------- -- -i--T-- l ------ .. ------ --'
...... Ru ...' cass
35- 45. 5 65- 75- 85. S95. 0.. as5 2is. I .35 t4S- 1 t65.
44 54 64 74 84 94 104 114 24 1 4 154 1(4 174
(PRo, TsrSmA ASo MER.u.1
of the usual types of intelligence-tests, particularly verbal tests, are
not as favourable as those enjoyed by the town child. It is even
likely that, apart from qualitative differences in intelligence, there
is a slight quantitative difference, innate potentialities not being
realized as fully under rural as under urban conditions. But this
difference is not a difference in the quality of the stock.
There are, of course, exceptional cases where the composition
of rural communities is definitely innately inferior. In isolated
mountain villages or semi-desert areas, where there is little scope for
meaningful activity, the average level of intelligence may have been
gradually lowered by selective migration, inbreeding of inferior
remnants, and possibly a progressively depressing effect by the phy-
sical environment on the physiological foundations of intelligence.
Provided that care is taken not to overemphasize such groups, there
is no reason why the representative sample of the community as a
whole should contain town and country samples in appropriate
proportions. Whether the sample should be drawn mainly from the
urban or mainly from the rural population is a matter to be decided
on the basis of convenience and simplicity. The experimenter will
naturally choose that group which is most readily accessible and
in which environmental influences can be more easily controlled.
In the case of Africans, it may be particularly convenient to be
able to ignore urban-rural differences, as it is extremely difficult to
draw a clean distinction. Virtually all rural Africans have spent
some time in the cities. Many urban Africans have rural antecedents.
In practice, it will probably be found most convenient to use sub-
jects available in the city and town locations, regardless of the perma-
nency of their residence in these areas.
3. ECONOMIC DIFFERENCES
In the selection of a. representative sample of the European
population, economic status cannot be ignored (Table I). While
differences in test-iltelligence between economic groups are again
to a considerable extent environmental, as will be shown in a sub-
sequent chapter, some innate differences in intelligence are bound
to exist as well. Though it is a notorious fact that lack of character,
rather than the possession of high intelligence, is frequently the
determining factor in securing economic success in Western society,
a considerable relationship between intelligence and success never-
theless remains, particularly if success is measured in terms of social
THE SAMPLING PROBLEM
MEAN I.Q.S OF A REPRESENTATIVE SAMPLE OF AMERICAN CHILDREN,
ACCORDING TO FATHER'S OCCUPATION (TERMAN AND MERRILL) 2
Father's Occupational Classification 2-5 6-9 10-14 15-18
I Professional 116.2 114.9 117.5 116.4
II Semi-Professional and Managerial 112.4 107.3 112.2 116.7
III Clerical, Skilled Trades, and Retail
Business 108.0 104.9 107.4 109.6
IV Rural Owners 99.1 94.6 92.4 94.3
V Semi-Skilled, Minor Clerical, Minor
Business 104.3 104.6 103.4 106.7
VI Slightly Skilled 95.1 100.0 100.6 96.2
VII Day-Labourers, Urban and Rural 93.6 96.0 97.2 97.6
and professional, as well as economic, status. Some difference in
the quality of stock, constituting the various socio-economic groups
of the European community, is therefore likely to exist, though this
difference is probably less marked than it is commonly believed to be.
The existence of the Colour Bar makes the position among
Africans radically different. As the possession of superior ability
causes the African to become a competitor of the European in the
higher-wage market, its expression is generally denied an outlet.
What the European requires of the African is hard manual labour, a
commodity which he does not himself produce, and this at a low
There are, of course, a few unusual individuals who succeed
in triumphing over all obstacles; but they constitute an almost
infinitesimal percentage of the African population.
Furthermore, the range of incomes is very narrow and selection
according to economic status therefore quite meaningless. More
significant may be the differences between various occupational
groups-the professionals (comprising teachers, clerks, ministers,
and occasional doctors); those in business on their own account;
the semi-skilled workers in industry ; the domestic servants, includ-
ing cooks; those employed in the distributive trades, including motor
drivers; the mine labourers ; the vast horde of casual, unskilled
labourers and farm workers. Whereas the differences in intelligence
between some of these groups are probably slight, it might be as
well to select samples from each, roughly in proportion to their
total number in the population as a whole. The statistics compiled
by the Municipality of Johannesburg would provide an excellent
guide to the research worker in this sphere.3
4. AGE DIFFERENCES
As we have seen, intelligence grows with age, very rapidly at
first, then slowing down markedly between 10 and 14 years, until
development ceases for the average individual at 16 years. In the
case of inferior ability, growth ceases well below 16 years, while con-
tinuing to 18 years and even beyond in the case of superior ability
A representative sample need not, therefore, contain individuals
older than 20 years. This considerably simplifies the testing prob-
lem. In the first place, children and adolescents are more readily
obtainable in sufficiently large numbers, as the majority of them can
be tested in schools, colleges, night classes, etc. In the second
place, their attitude towards the test-situation is far more favourable
than that of adults. The latter often resent being tested, or consider
the test childish. They are frequently not prepared to make the
effort necessary to return a good score, or they cannot achieve that
single-minded concentration on the task which comes easily to child-
ren interested in the test and imbued with a spirit of rivalry.
If it is our object to compare the intellectual capacity of two
races, we cannot content ourselves with the measurement of differ-
ence at one or two age-levels, or at the age-level where the growth
of intelligence stops for the average individual. We should deter-
mine the shape of the entire growth curve for each race for the fol-
lowing reasons :
1. The rate of growth may not be identical in the two races.
Knowledge on this point is of some practical importance. Marked
differences might necessitate some changes in the educational system
for Africans, which is at present adapted to the developmental pro-
cess of the European child.
2. It is generally alleged that at about adolescence a marked
deceleration occurs in the rate of intellectual growth among Afri-
cans. If this phenomenon had been observed only in intelligence-
test results, it might be argued that it is due to the existing tests at
higher age-levels being unsuitable for Africans. For example, it
might have been necessary, in order to increase the difficulty of the
test problems for older children, to state these problems in a form
which handicaps the African relatively to the European, in that,
because of his educational background, the medium itself becomes
too difficult for him.
Arrested development has, however, been noted under circum-
stances where this explanation is definitely excluded. No thorough
study of this change has yet been made. It may be closely tied up
with cultural factors, such as the change in outlook which accompa-
nies the transition period between adolescence and manhood, still
marked in many tribes by the circumcision ceremony. It may also be
related to a personality-difference which is the outcome of a greater
interest in the sexual life on the part of the African child. It may
be truly physiological in origin, in which case further research into
its organic causation is needed. It is also particularly important to
know whether this stage of arrested development is permanent or
temporary, and if temporary, whether the setback it produces can be
Made up, or whether all those subject to it must finish at a lower
apex than those not so afflicted.
3. Even though there may be a difference between the mean
intelligence of two races, there is bound to be a considerable overlap
between the two distributions of individual I.Q.s. This is often
forgotten when some statement, such as "the Native is less intelli-
gent than the European", is made. Such a statement might be true
of the mean, but not of every individual. As we have seen, the growth
of intelligence does not stop at the same age for all persons. It is
Important to know how many Africans, compared with Europeans,
develop beyond the average racial level, and up to what point.
By using one or two age-groups, some idea of the average infer-
iority or superiority of one race over another can undoubtedly be
obtained. Only a few of the questions raised above can be answered
from such limited material. The partial information obtained isdan-
gerous, as it tempts people to make generalizations which are only
partly true, but which may have serious practical consequences.
Particularly in view of the wide divergence in the environmen-
tal conditions to which Africans and Europeans are subjected, it is
most important that the fullest possible information should be at
hand regarding the growth of intelligence in the African race. This
information can be obtained only by examining samples at each age-
level up to the age of twenty.
5. SEX DIFFERENCES
There is virtually no difference in intelligence between Euro-
pean males and females (Figure 3). For a while girls lead boys by
36 AFRICAN INTELLIGENCE
a trifle, probably as a result of the somewhat earlier maturation of
the female. The boys pull up at a later stage, however, so that adult
men and women probably finish up level at the end.
We cannot assume without some proof that African boys and
girls develop at the same rate, and up to the same level. It might be
advisable, therefore, to include equivalent samples of each sex in a
representative sample of the African race.
Whether it is possible to select two representative samples of
the African and European races respectively, and, at the same time,
to satisfy the control-group criteria, is open to serious doubt. The
possibility of equating the two requirements will constantly be borne
in mind throughout the argument developed in this book. When
all the relevant environmental facts are before us, we shall return
to a final discussion of this problem.
1. Terman, L. M. and Merrill, M. A.
Measuring Intelligence. London,
Harrap, 1937. p. 49.
2. Ibid. p. 48.
3. Johannesburg, City of, Non-Euro-
pean and Native Affairs, Dept.
of, The African in Industry. City
of Johannesburg, 1939.
4. Cattell, R. B. Your Mind and Mine.
London, Harrap, 1934. p. 82.
5. Terman, L. M. The Measurement
of Intelligence. London, Harrap,
1922. p. 69.
THE INFLUENCE OF GENERAL CULTURAL
MILIEU ON TEST-INTELLIGENCE
The cultural differences which obtain between rural and urban,
tribal and -detribalized, backward and progressive, civilized and
primitive, communities exercise a measurable effect on test-intelli-
gence. In a well-controlled investigation, Klineberg' neatly de-
monstrated the occurrence of such an effect in the case of Negro
school children after varying periods of residence in New York.
Having divided his experimental group into sub-groups with from
1-2 years, 3-4 years, 5-6 years, 7-8 years, over 8 years, residence
in New York, after immigration from the South, he applied a num-
ber of language and performance intelligence-tests to them. On the
National Intelligence-Test, the average scores for these sub-groups
were 72, 76, 84, 90, and 94; on the Stanford-Binet, 81.4, 84.2,84.5,
85.5, 87.4. On a performance-test, less affected by environmental
differences, the scores were 39, 26, 31.88, 37.5, 37.5. For New-York-
born Negro children the score on-this test was 41.61.
Anastasi2 aptly remarks : "The experience of people living in
different cultures may vary in such a way as to lend a different
meaning to their actions, stimulate the development of totally differ-
ent interests, and furnish diverse ideals and standards of behavior."
The manner of living of the Union African population has, as
a result of contact with Western civilization, become so diversified
that it would not be correct to speak of one cultural milieu common
to them all. Yet as a result of the Colour Bar and the segregation
policy only a very small number has been able to become thorough-
ly Europeanized. As a people, therefore, they remain distinct,
and despite detribalization certain general traits and attitudes tend
to linger on. The European's individualism, which makes of com-
petition a social virtue; his "hustle", which makes "Time is Money"
an article of faith with him ; his tendency to be restless and strenuous
even in his moments of relaxation-these are all culturally condi-
tioned attributes which even the city-bred African has not thorough-
ly assimilated, foreign as they are to the tradition of his people.
Intelligence-tests are, however, constructed and applied on the
assumption that these attributes are inevitably present in all testees.
This brings us to the first of the cultural environmental influences
affecting the test-intelligence of Africans.
1. ATTITUDE TOWARDS TEST-SITUATION
'(a) FAMILIARITY WITH THE TEST-SITUATION AND ITS REQUIREMENTS
Most European children get accustomed to test-situations in
the course of their school experiences. Hence, the answering of an
intelligence-test evokes primarily a number of routine responses,
while it also attracts interest as a game and immediately arouses the
impulse to rivalry. The fact that some tests have a time-limit merely
enhances the competition atmosphere. The motives behind the
testing are not likely to be questioned, provided the test is adminis-
tered in a standard way by trained testers. Variations in individual
reactions, in comparable age and environment groups, are, therefore,
bound to be of a random kind, not affecting group averages in a
constant manner. To the African child who has never been to
school, or vho has only passed through the first two standards, the
test-situation appears in quite a different light. Even if the feeling
of awe can be overcome, there still remains the difficulty of getting
the child to give his undivided attention and to try his utmost to
succeed. Rivalry will not be so potent a motive as it is in a European
group, partly because the African is less of an individualist than the
European, partly because the situation 'does not evoke the same
amount of interest. In the case of the school-going African child,
where some of these problems no longer arise, new and even more
serious attitudinal difficulties present themselves.
(b) THE PRESENCE OF A EUROPEAN TESTER
The European child is always tested by a member of his own
race, either a teacher, or a special tester, who is often mistaken for
a school inspector by the class. The African child, on the other
hand, is usually tested by a European, and, in view of the dominance-
submission relationship between the two races, this introduces a
number of features which cannot be paralleled in the European
group. Younger children who have never been taught by Europeans,
are thrown into a state of apprehension, or experience an emotional
inhibition which precludes single-minded concentration on the
CULTURAL MILIEU AND TEST-INTELLIGENCE 39
task inrhand. In older students, apprehension is gradually displaced
by suspicion and even antagonism. In the course ot an inquiry
into the African's attitude towards European ideas of right and
wrong, the writer had occasion to visit most institutions for Native
secondary education in the Union. Again and again, hostility and
suspicion were encountered, partly because of the nature of the
questions asked, but more particularly because the questioner had a
white skin. This reaction is to some extent an immediate and un-
reasoned one, somewhat like a conditioned emotional response. It
is partly determined, however, by uneasiness about motive. "What
is the European now after again ?" and "How can what I say be
used against me ?" are questions frequently asked or implied.
This suspicion is particularly strong with regard to intelligence-
testing. The publication of results from poorly controlled investi-
gations, purporting to prove the innate intellectual inferiority of the
Africans and asserting the futility of higher education, has done a
vast amount of harm. Without being able to point to all the flaws
in these investigations, the African feels, nevertheless, that they are
unfair to him. While 15-year-olds may not have clearly formulated
these suspicions to themselves, a generalized doubt about European
sincerity probably exists even in their case. This doubt becomes
more and more explicit as they grow older. A non-co-operative
attitude is adopted, lest it be said that the African himself accepted
the test as a valid one and did his best to shine at it. Allowance
must also be made for a quite natural uncertainty in the African's
own mind about his intellectual equality with the European. This
fear, perhaps never consciously admitted, may have an inhibitory
effect, or lead to the adoption of a "go-slow" policy as a defence
In order to meet some of these difficulties, some investigators
have used an African to give the instructions, while they themselves
either supervised or took over at the appropriate time. This device
is unsatisfactory as it involves both a change in attention and a change
in attitude on the part of the testees. The mere presence of a Euro-
pean supervisor introduces, furthermore, an element which cannot
be paralleled in the European group. The ideal is to use a fully
trained African tester who can be trusted not to give any assistance,
either consciously or unconsciously; to the members of his race
being tested. While this would put the pupils more at ease, it would
not necessarily remove their doubts and suspicions. Special efforts
would have to be made to win the ready co-operation of the pupils
in the form of a frank, though tactful, explanation of the nature of
the investigation. Whether such a talk would achieve its purpose is
an open question, but it is the best that can be done in the circum-
stances. The possibility that an altitude of suspicion towards the
test-situation may have influenced the measurements obtained,
despite all precautions, must always be borne in mind.
(c) THE AFRICAN'S ATTITUDE TOWARDS EDUCATION
The Native child's attitude towards education is very different
from that of the European child, and in so far as intelligence-tests
are usually applied during school hours, complications not met
with in a European school are apt to arise. It came as a surprise to
the writer that African scholars preferred doing the attitude-tests
in their free time after school, rather than in school time. The rea-
son for this is that to the African education is a privilege, as yet
within the grasp of a minority only. Is is the one avenue which offers
an escape from the inferior position in which Africans find them-
selves. At the present stage, this education is valued less for its own
sake than as a means to an end, and for the material and prestige
advantages which it may bring. The senior, matriculation, or
teachers' certificates are the coveted, concrete realizations of this
ambition, and, with great single-mindedness of purpose, school
time is exclusively devoted to a preparation for examinations. School
time therefore literally means money, not only because it imposes
a very heavy financial burden on the African parent, who has to
forego the'earning capacity of the child and spend a large propor-
tion of his money-income on board, fees, and travelling expenses;
but also because every minute of school time means a step nearer to
the goal. This concentration on the examination to the exclusion
of an interest in the cultural value of education is considered to
be necessary because the African pupil feels, himself handicapped in
his preparations for the examination by lack of background, language
difficulties, and often poor tuition. The examinations, therefore,
become relatively more difficult for him than for the European can-
The European child is far less self-conscious and often far less
responsible in his attitude towards education. It is too much a mat-
ter of course and too easily obtained, often without cost, for him to
consider it a privilege. Nor is he, on the whole, as sensitive as the
CULTURAL MILIEU AND TEST-INTELLIGENCE 41
African child to the financial sacrifice which may be involved. This
is partly due to the fact that schooling is compulsory up to a certain
age, and partly to the better economic status of most European fami-
lies. Consequently, one finds that, whereas in the European school
an intelligence-test is looked upon as a pleasant diversion, appre-
ciated by some because of the change it brings from the boredom of
school routine, by others not so motivated because of its intrinsic
interest, in the African school it becomes a wasteful interruption.
It follows that frequently little enthusiasm will be evinced, so that
on this score alone many an African child approaches the test in a
frame of mind different from that of the European. It will, there-
fore, always be preferable to administer the test on a Saturday or
other free day; and if this raises complications, an attempt should
be made, either to create sufficient interest in the test as a diversion,
or else to convince the students that some advantages, relative to the
examination, are likely to accrue from taking it.
2. CULTURAL UNFAMILIARITY OF INTELLIGENCE-
TEST CONTENTS AND TESTING ROUTINES
It has been generally recognized that tests devised for people
reared in one cultural milieu, and measuring ability to function
efficiently within that 'milieu, may be quite unsuitable for those
reared in a different cultural environment. In the first place, the
type of problem set, or the medium in which it is set, may be un-
intelligible to a person unfamiliar with its cultural background ; in
the second place, habits providing for efficient adaptation to one set
of environmental circumstances may be absent in people reared
under different conditions, where these particular adjustments are
Difficulties of the first type are not merely relative to language
differences, but also to differences in that socialization process where-
by special expectations and meanings are built up. In the Otis Self-
Administering Mental Test, there occurs a question : "A meal al-
ways involves-(1) a table, (2) dishes, (3) hunger, (4) food, (5)water."
The majority of a group of Indian children chose "hunger" because
they were in the habit of eating only when hungry and not at stated
times. To the question: "Why should all parents be made to send
their children to school ?", which occurs in Form 6 of the Alpha
42 AFRICAN INTELLIGENCE
Test, the correct answer of the various alternatives given is that
"school prepares the child for his later life" ;,but this is only partly
true for the urbanized African and quite untrue for the Reserve
African. One of the tests for Age III in the Individual Scale of Ge-
neral Intelligence for South Africa is to give the surname. Nomen-
clature and the use of names is so different in a tribal community
that this question is certain to be unintelligible to its members. At
Age VII in the same test, the child is required to tie a bow in a shoe-
lace. At that age a negligible percentage of African children have
ever worn shoes. In the attitude-test already 'referred to, the writer
included an item : "You must wash yourself at least once every day",
asking for a "yes" or "no" to this statement. These plain alterna-
tives proved unsatisfactory, such queries being raised as "What are
we to do if there is a drought and the little water we have has to be
carried for six miles ?"
The speed factor in intelligence-tests illustrates a difficulty of
the second type. There is a premium on speed and.hurry in West-
ern society. Few of us are without a watch, and catching trams and
trains, being "up to time", "getting things done"-even if they are
useless-becomes habitual with us. Even our natural competitive-
ness is cast in a mould of speed. This "hustle" is quite foreign to
African culture. Africans in the Reserves have plenty of time "to
stand and stare", and even the urban African population has by no
means assimilated the European's restless way of living. It follows
that the intelligence-test with time limits tends to handicap the Afri-
can child, which does not respond to the injunction to work as rapid-
ly as possible. If he does, he is likely to become confused, as his
training has not provided him with the necessary habits.
These general difficulties have, of course, been recognized by
investigators, but it cannot be said that a satisfactory solution has
yet been found.
Yet here lies the crux of the whole problem of interracial intel-
ligence-testing. Either we shall have to find an intelligence-test which
is equally difficult and diagnostic for both Europeans and Africans;
or else some special test will have to be designed capable of measur-
ing African intelligence as efficiently as current tests measure Euro-
pean intelligence. This second alternative raises the further problem
of some common standard by which to assess the results obtained
from these two instruments.
Owing to the vital importance of this question, it will be dealt
CULTURAL MILIEU AND TEST-INTELLIGENCE 43
with in a special chapter, in which the suitability of existing tests will
be exhaustively examined.
Before proceeding with this task, a distinction must be drawn
which is implicit in all that has been said regarding the significance
Environmental influences can be held to exercise both an extrin-
sic and an intrinsic effect on test-intelligence.
Extrinsic Effects are those which arise as a result of unsuitability
of the test and of the test-situation. The test-intelligence of Africans
may be quantitatively affected as a result of the use of test-material
which is strange and unfamiliar to them and of testing techniques to
which they are not accustomed. They may not be fully conversant
with the language medium used, strange to the symbolism, and un-
acquainted with the cultural content.
Intrinsic Effects are those which influence intelligence itself, and
not merely test-intelligence as in the above case.
This influence may be either qualitative or quantitative.
It is qualitative when it leads to the establishment of thought
processes, perceptual habits, specific skills, and interests which are
peculiar to a particular culture, and are not found in other cultures.
It is quantitative when it affects not the form in which intelli-
gence expresses itself, but its rate of growth and ultimate capacity.
Cultural milieu usually exercises qualitative intrinsic effects,
but differential quantitative ones may also occur. In those primitive
communities, for example, where education is repressive, and pa-
rents conventionally adopt an attitude of aloofness towards their
children, unfavourable emotional conditions and lack of formative
stimulation may permanently retard the normal development of
Test-intelligence can, of course, be affected in one way only
by these three environmental conditions.
It should be the aim of the intelligence-tester to construct his
test, and to arrange the testing, in such a way that neither extrinsic
effects, nor qualitative intrinsic effects, can adversely influence the
scores. Quantitative intrinsic effects are, of course, unavoidable,
and must be carefully allowed for when test-intelligence is used as
an indication of innate capacity.
Do existing intelligence-tests measure up to these require-
44 AFRICAN INTELLIGENCE
1. Klineberg, O. Negro Intelligence
and Selective Migration. Colum-
bia Univ. Press, 1935.
2. Anastasi, A. Differential Psycholo-
gy. London and New York,
Macmillan, 1937. p. 503.
OF EXISTING INTELLIGENCE-TESTS
FOR THE MEASUREMENT OF
The various types of intelligence-test at present in use can be dis-
cussed under three headings :
(a) Group-Tests of the Army Alpha type;
(b) Scales of the Binet type;
(c) Performance-Tests and Scales.
Most group-tests, though not all, are language tests, the Ameri-
can Army Alpha Test being their prototype. This test was de-
signed in 1917, to enable the American Army authorities to deter-
mine the fitness of recruits for various types of Army training. The
examination consisted of eight tests, viz. (1) Following Directions;
(2) Arithmetic Problems; (3) Practical Judgement; (4) Synonym-
Antonym; (5) Disarranged Sentences; (6) Series Completion;
(7) Analogies; (8) General Information. The test broke entirely
new ground in content, manner of presentation, answering, and
scoring. Particularly noteworthy was the amount of verbal material
it contained. Reasoning ability, for instance, was measured by ability
to see relations of a verbal sort in the analogies test. (Lion is to
Animal as rose is to-smell, leaf, plant, thorn; a line to be drawn
under the appropriate word). The Synonym-Antonym test was
designed to assess ability to apprehend relations of likeness and
difference. (A pair of words was given, such as "fallacy-verity" and
behind it the words "same-opposite", one of which had to be under-
lined in answer to the problem. The test contained a large number
of such pairs.) In the Disarranged Sentences test, a jumble of
words was given. When rearranged, these words made a sentence
which was either true or false. By underlining "true" or "false"
behind each sentence, the testee indicated whether or not he had
46 AFRICAN IN 'ELLIGENCE
succeeded in rearranging it correctly. This test was intended to
measure ingenuity and cleverness.
Verbal group-tests of the Army Alpha type (of which the South
African Group Intelligence-Teit is one) are useless if the language
medium is not completely familiar to the testee, for a number of
1. Instructions may not be fully and clearly understood.
2. Finer shades of meaning may not be fully grasped, and
vocabulary may be limited.
3. More careful thought way be needed to comprehend fully
the meaning of words in "andogies", "opposites", and "classifi-
cations" tests, before the actud problem set through the medium
of these words can be tackled, thmi slowing up the test-performance.
(Group-tests of this type are always time-limit tests.)
4. These language sub-tests also presume an analytic mode
of verbal thinking, which may be quite unfamiliar to one speaking,
like the Bantu, a polysynthetic tongue. Where we have words, the
Bantu as often as not have roots, stems, prefixes, and sur'iSxe. By
means of a complicated mood and tense system and the use of a
large number of auxiliaries which form compound tenses, a wealth
of subtle meanings may be given to a verb, indicating situation,
possibility, intention, implication, manner, etc. Whether a quali-
fying word is the adjective,relative, numeral, or possessive, is deter-
mined by the form of concord used. Concordial agreement is indeed
an outstanding feature. To convey : "Our wise Chief bought the
white cows", the Zulu would have to say: "Chief he-our he wise he-
them-bought the-cows them-white.'' Furthermore the entire mean-
ing of a word may be changed by a mere change in intonation. It is
doubtful whether even a matriculatedAfrican thinks spontaneously,
and by preference, in a foreign medium from which his own differs
so markedly in structure. Not until Africans are taught English or
Afrikaans from birth as their mother tongue will existing language
group-tests be applicable to them.
5. The question of unfamiliarity of context through lack of
I culturally determined associations must also be taken into account,
though this is of less importance in'the case of urban Africans who
have received at least a primary-school education.
6. Finally, there remains the handicap of poor vocabulary, as
distinct from unfamiliarity with the structure of the language. On
the whole, the African child's detribalized environment,ascompared
SUITABILITY OF EXISTING TESTS
with a tribal way of living on the one hand, and an average Euro-
pean environment on the other hand, is poor and uncultured.
Opportunities to talk anything but the vernacular are severely
limited. Contact with European teachers is rarely possible beyond
the school situation. In any case, it is only the privileged few who
are ever taught by Europeans. African teachers suffer from simi-
lar disabilities themselves and are therefore hardly in a position to
assist their pupils in getting a firmer grasp of European culture
and language. Obviously, only continuous and intensive conversa-
tional contact with educated Europeans would be of any use; and
this the social Colour Bar prevents in all but a few cases. There
remains reading as a means to vocabulary development. Again
only a small percentage can effectively avail themselves of this
method. The majority lack sufficient education as well as the cul-
tural background, necessary to enable them to appreciate what they
are reading, or to read those books which would enrich their lan-
SCALES OF THE BINET TYPE
These scales, which are all revisions and re-standardizations
of the scale published by Alfred Binet in 1908 and revised again in
1911, consist of a wide variety of tests located at different age-levels
from 2 years to "superior adult". Many depend on familiarity with
a particular language medium, on: wealth of vocabulary, and on
the appreciation of verbal concepts. What has been said regarding
the language element in group-tests applies here equally. The
assumption underlying other tests in the Binet Scale is that a normal
child, at each age-level, should have acquired sufficient familiarity
with certain objects and social situations to be able to react adapt--
ively to them in the test-situation. Identification of coins, stamps,
objects in daily use, giving correct change, answering such ques-
tions as: "What would you do if you were cold, hungry, sleepy,
etc.?", giving the months of the year, and tying a bow-knot, are ex-
amples of the type of problems set. All these items presume a specific
cultural milieu, and their use is only justified when thetestee has had
normal opportunities to familiarize himself with that milieu. The
"Ball and Field" test, located at Age XII, in which the subject
must show by means of a pencil-tracing how he would find a ball
lost in a field represented on the test paper by a circle, evoked some
interesting responses from Noh-European children. Thus Porteus
AFRICAN INTELLIGENCE ,
and Babcock' report that they had to eliminate this test when
working on the intelligence of Japanese and Chinese children, be-
cause "the presentation of finding an imaginary ball in an imaginary
field (represented by a ring) was usually met with a blank stare,
being too much for the practical-minded Orientals".
Included in the Binet Scale are also a number of tests of the
performance type, .such as a simple form board*, copying a figure
from memory, pointing out omissions from pictures, uniting the
halves of a divided rectangle, etc. Performance-tests will be fully
dealt with in the next section. Suffice it to say at this stage, that
these tests usually demand manipulative and perceptual habits
which are culturally determined. Hence, they can only be used
for testing African intelligence if these habits are common to the
two cultures. .
The remaining tests in the scale (the counting test and discri-
mination of weights test) are probably least unsuitable for interracial
comparisons, but, by themselves, they are quite useless. It is hardly
surprising, therefore, that a number of judges "thoroughly acquaint-
ed with the language, culture and life generally of the Natives"
found that 69 of the 77 items in the scale were totally unsuitable
for Native children, and of the remaining .8, 4 were too easy for the
ages under consideration.2
Attempts have been made to re-standardize the Binet Scale for
primitive and oriental culture groups. Dr. Herbert Rice, of the For-
man Christian College, Lahore, has constructed a Hindustani Binet,
which was standardized and revised on a thousand children. This
revision included a good deal of performance material.3
Beatrice Blackwood reports an attempt to apply the Binet
'Scale to a Zulu group in Natal. "Starting with the Goddard revi-
sion of 1911, the investigators constructed a version adapted to
local'conditions, translating such tests as were suitable, and putting
in substitutes for others. Some, for which they were unable to find
any kind of substitute, were omitted altogether. They had to choose
other pictures for description, because the regular pictures imply a
cultural status which even the most educated Zulus have not attained.
The form board usually consists of a wooden board with apertures, into
which have to be fitted one or more pieces of various shapes.
t The scale referred to is the South African standardization of the Stanford-
Binet, known as the "Official Mental Hygiene Individual Scale". It has since
been replaced by a new standardization, known as "An Individual Scale of General
Intelligence for South Africa", published by the S.A. Council for Educational and
Social Research, Research Series No. 7.
SUITABILITY OF EXISTING TESTS
All tests involving dates, time and money had to be omitted. lhe
objects prescribed in Test 2 for age 4, viz. key, knife and penny
were regarded as being unfamiliar, and beads, a Native pot and a
loin cloth were substituted. Those prescribed for 'memory discri-
mination' at age 8 had also to be replaced. Time limits had in most
cases to be extended to allow for the normally slow reaction time
of Zulus. These examples will serve to indicate the general trend
of the alterations required." The investigation referred to, carried
out by Loades and Rich4, was apparently of a preliminary nature ;
but it seems never to have been followed up, as no further reports
regarding it can be found. While it may be possible to work out
an adequate scale in this way, it could only be used for intra-group
comparisons. A particular score on the one scale might represent
a much higher or much lower achievement than the same score on
the original standard. There is at present no means whereby two
different standardizations can be equated. As neither language
group-tests, nor scales of the Binet type, can therefore be used for
the comparison of African and European intelligence, performance-
tests are our sole remaining hope.
The common principle underlying these tests is the use or
non-verbal material. As the test-set is usually of a manipulative of
perceptual type, which can easily be demonstrated, these tests have
become the stand-by of interracial intelligence students. Perform-
ance-tests usually consist of Form Boards, Block Tests, Drawing
Tests, Puzzles, Picture Completion Tests, Maze Tests, and the
like. Most of them are individual tests, though there are group-
tests, such as the Army Beta Test, the International Group Mental
Test and the "Drawing a Man" Test.
As a rule, a certain amount of verbal instruction is necessary,
but there is no objection to its being given in the vernacular. In
some tests, notably the Army Beta and the International Group
Mental Test, use is made of pantomime. Pantomime directions are
1. Pantomime is not our normal means of communication.
Apart from the fact that it introduces an artificial atmo-
sphere, it leaves considerable scope for misunderstanding.
2. Pantomime directions cannot be adequately standardized,
so that no two applications of the test would be identical.
50 AFRICAN INTELLIGENCE
3. It has been found that attention easily wanders when pan-
tomime is used.
The Leiter International Performance Scale is almost entirely
independent of pantomime instructions. The principle of the test
is the matching of appropriate designs, ability to see essential rela-
tions being one of the chief factors necessary for success. There is
no time limit and it is claimed that the test is independent of cul-
tural background. As the test uses abstract designs, with'which
some cultures are more familiar than others, and as the process of
matching is a common one in Western parlour games, it is doubtful
whether this claim is justified. Undoubtedly, performance-tests of
this type are far more independent of specific cultural associations
than the tests discussed under the first two headings; but this inde-
pendence is by no means complete.
Even in the form-board and block tests, where the medium
merely demands familiarity with the handling of blocks and shapes
of wood, and with the dimensional qualities of shapes and solids,
African children are at a cultural disadvantage. The majority of
European children have an opportunity of familiarizing themselves
with blocks and jig-saw puzzles during their first years of life, when
they are given these materials as toys. Many indeed get in intensive
training in the use of these materials in nursery schools, and, as a
result of compulsory education, every child has at least its kinder-
garten experience to rely on. Similar experience on the part of
African children is the exception rather than the rule. Kindergarten
classes, which are attended by a fair percentage of African children,
are usually too large and too poorly equipped to provide much train-
ing in this direction, and it cannot be doubted that African children
are somewhat handicapped in these performance-tests.
Proof for this statement is provided by an investigation
conducted by Hugo and Biesheuvel on 48 delinquent African
boys at Diepkloof Reformatory near Johannesburg. These boys,
aged between 13 and 14, were divided into two equivalent groups.
One group received training in the handling of coloured blocks.
This training was given to three boys simultaneously, the
average training-time being 3 hours. After an interval of a week,
both the trained group'and the untrained group were given the Koh's
Blocks and Cube Construction Tests. Five scores were computed,
the average score of the untrained group in each case being lower
than that of the trained group. The fact that these differences were
SUITABILITY OF EXISTING TESTS
all in the same direction makes up for their rather low statistical
significance. If even so short a training period, at an age when the
optimal conditions for the establishment of these perceptual and
manipulative habits are well past, can have a measurable effect on
test scores, it can be safely assumed that normal experience, extend-
ing over some years, and at the appropriate age-levels, would have
had a far more marked effect.
Consequently, even this type of performance-test cannot be
accepted as a valid measure of African intelligence, equally suited,
to both Africans and Europeans. If these tests are used at all,
some preliminary practice with kindred material should be given.
Thus when putting the Ferguson form boards to a number of Afri-
can delinquents, the writer noticeA one child whjch scarcely knew
how to pick up and handle the pieces to be fitted. Its manipulative
dexterity was almost nil, and hence its performance extremely poor.
Yet at a second attempt this child was already at ease and its perform-
ance quite creditable in terms of the norms for the second attempt.
Such practice, of course, does not by any means dispose of the entire
handicap ; it merely removes the grosser disabilities.
There is a further objection to the form-board test which it
will be difficult to dispose of. The score is always based on time taken
to complete each board, though number of moves made are some-
times also taken into account. The introduction of a speed factor
imposes a handicap on those reared in a leisured milieu, which no
amount of practice can remove. It springs from a basic personality
tempo, which is firmly established at an early stage. The alternative
of a scale of tests of increasing difficulty is not practicable in the case
of performance-tests. Such a scale would of necessity have to include
complex spatial and manipulative problems which would be tests of
some special aptitude, rather than of general intelligence. Perhaps
it might be possible to apply a correction to time scores by multi-
plying them by some constant, based on the tempo of the subject's
movements. There are, furthermore, numerous ways in which num-
ber and quality of moves made to achieve a solution could be used
for scoring purposes. Further research into these questions is urgent-
Pictorial test-material is far less suitable than blocks or form
boards for interracial comparison. Though a picture is a more con-
crete symbol than a word, it yet remains a symbol. Anastasi5 makes
this point as follows : "A two-dimersional reproduction of an object
is not a perfect replica of the original; it simply presents certain
cues which, through the influence of past experience, lead to the
perception of the object. If the cues are highly reduced, as in a
simplified or schematic drawing, or if the necessary past experience
is absent, the correct perception may not follow."
Nadel6 gives the following example: "When testing young
Yoruba boys in Northern Nigeria, I discovered that they could not
identify outline drawings on paper of a man, or even of such common
objects as a hut, a crocodile or a pot, although exactly the same out-
line figures were at once identified and described when they appeared
in carvings or on Native leatherwork, i.e. in their familiar, established
cultural context." Another difficulty was experienced by Margaret
Mead7 who found Samoan children unable to adopt an explanatory,
theoretical attitude towards pictures she asked them to interpret.
Their point of view remained entirely aeshetic and they would only
comment on the beauty of the scenes. In the natural state, the Afri-
can child's contact with pictures is negligible, and even the school-
going child has far less opportunity to familiarize himself with pic-
torial material than the average European child, who usually has a
number of picture books and gets accustomed to seeing pictures
about the house. Their significance as symbols is learned'at an early
Finally the contenit-error must be guarded against when using
pictorial material. Only pictures of objects with which the child has
had a reasonable chance to familiarize itself should be used. Most
of the existing pictorial tests, such as the Army Beta and Pintner-
Paterson Tests, would have to be thoroughly revised before they
could be applied to African children. A test of some promise is
Goodenough's Drawing Test, which does not seem to be influenced
by school training, though it demands familiarity with pencil and
paper. Bladkwood8, who used a test similar to it in the course of
her studies among Negroes and Indians, states that even those who
had only recently entered school, had had no instruction in drawing,
and whose "introduction to a pencil was clearly not far behind them",
performed very successfully on this test. As far as instruction and
procedure are concerned, the Goodenough Test is simplicity itself.
The sole instruction given to the subject is to make a drawing of a
man, ample time being allowed. A detailed scoring scale, with nu-
merous sample drawings fully analysed, is provided. A certain
amount of practice in scoring is needed before the test can be used
SUITABILITY OF EXISTING TESTS 53
with confidence. It is well standardized, has a reliability of .94,
correlates .78 with the Stanford-Binet, and is easily and quickly
administered, even to a group. Unfortunately it can only measure
mental ages up to 10 years. It is strongly recommended for inter-
racial comparisons at lower age-levels, where it may prove to be a
valuable research instrument. Modification of technique and of the
medium in which the drawing is done may be necessary before the
test can be applied to primitive African children.
Our discussion so far suggests that, though the use of perform-
ance-tests involves certain extraneous difficulties-such as manner
of presentation, need for preliminary practice, and method of scor-
ing-none of the problems is insurmountable. Nevertheless their
removal would not mean that our difficulties were over.
Let us suppose a performance-test could be applied to African
children without imposing any of the handicaps so far discussed.
Would it even then be as efficient and valid a test of African as of
European innate intelligence ?
At this stage, the reader is advised to turn back to the last
paragraph of Chapter I in order to refresh his memory on the way
in which we decided to define intelligence in this book. We shall
adhere to the concept of g, interpreted in Thomson's rather than in
Spearman's sense, and admit also the existence of anumberofgroup-
factors. We shall assume that it is our primary object to measure
amount of g, by which all subsequent development is ultimately
conditioned. We-shall have to be careful not to select tests which
are more a measure of some group-factor or other, than ofg. Group-
factors are not the whole of intelligence. Individuals may differ in
their possession of them, either through innate or through environ-
mental causes, without differing in the amount of g they possess. It
remains a necessary criterion of a good intelligence-test that varia-
tion in score should be due more to variation ing than to variation in
a group-factor, or group-factors, or to those undefinable specifics,
through which g happens to be measured. Spearman holds that in
group-tests of the verbal type this condition is satisfied by the selec-
tion of abilities in which specifics play a comparatively unimportant
part, and in which group-factors are absent. This selection is made
possible by virtue of the fact that the extent to which an ability is
saturated with g rather than its specifics can be mathematically
determined. Spearman's claim that group-factors are absent in
the usual test for g also rests on mathematical argument. If one
54 AFRICAN INTELLIGENCE
accepts it, then it can be safely assumed that a good or bad perform-
ance on an analogies, opposites, or classification test, is more likely
to be determined by amount of g than by the possession of a specific
Actually there is some doubt regarding the correctness of Spear-
man's claims. There is a distinct possibility that a verbal group-fac-
tor does exist. Higher mental activity is, however, so closely linked
to verbal facility that it would be most unusual to find high g un-
accompanied by a fair degree of development of the verbal factor.
The type of stimulation which is necessary to allow g to develop
to its full stature can only be provided in a language form. In any
case, group-tests usually also contain tests which involve number-
ability and perceptual ability, so that, notwithstanding individual
variations in the possession of group-factors, a fairly accurate measure
of g is bound to be obtained in the end.
Binet Scales attempt to achieve the same result in a different
way, by offering scope for a wide range of abilities. The Stanford-
Binet even provides for a number of alternate tests at each age-level,
so that if one test happens to be unsuitable, a further opportunity
to score is given through another. No one is, therefore, likely to
be credited with an I.Q. which is too high or too low, because he is
endowed with, or is deficient in, that particular aptitude which
happens to be the one through which g is being measured.
By comparison, performance-tests are less efficient instruments,
for the following reasons :-
1. The skills utilized by them are less saturated with g. Garrett
and Schneck9 state the case admirably when they say of the Pintner-
"Considerable scatter will ordinarily appear in the Mental Ages
calculated for the same child from the separate tests. Part of this varia-
bility is due to the fairly specific functions measured by the test; and
part arises from the instability inherent in simple manipulation tests
of this sort. Because of this, the M.A.s obtained from separate tests
in the series are admittedly rough measures. Chance and luck are
decidedly important factors in the simpler form boards, and the puzzle
type of test, once solved, is never the same test the second time. For
thirteen tests, ten of which were taken from the Pintner-Paterson Series,
Gaw obtained reliability coefficients of .76 and .54. : When the
reliability of the composite is no greater than this, the reliabilities of
the separate tests must be quite low. Johnson and Schriefer report
SUITABILITY OF EXISTING TESTS 55
that the correlation between M.A.s obtained from the separate Pint-
ner-Paterson Tests and the median M.A. from the whole'scale range
from .13 to .70 with a median of .50. Such a wide range of correlations
between separate tests and total score indicates considerable specificity
in the abilities measured by the single tests."
2. The construction of a battery of performance-tests, while
partly overcoming this difficulty, immediately introduces another.
Such a battery, in addition to measuring g, inevitably also measures
some special skill. Statistical analysis of results obtained from
Alexander's scale, consisting of the Passalong Test, Cube Construc-
tion Test and Koh's Blocks, revealed that "in addition tog a practical
factor (F) was essential to success in the tests"10
Garret and Schneck show that, when maturity is held constant
(i.e. when single age-groups are examined), the correlations between
performance-tests and the Stanford-Binet are roughly between .45
and .65. (The Goodenough "Drawing a Man" Test is an exception.
For single age-groups, ranging from a 4-year group to a 10-year
group, the correlations varied from .56 to .86 with a median at .72.)
They conclude that a performance-test of the Pintner-Paterson type
measures a concrete and manipulative type of "intelligence", which
is in most respects different from that measured by the Stanford-
The occurence of a group-factor in a test-battery of this type is
a far more serious matter than it is in a language group-test of intel-
ligence. As we have seen, the verbal group-factor is not only one
which is fairly generally developed, because it is an essential feature
of our everyday life, but it is also closely connected with the growth
and expression ofg. A practical manipulative factor is, however, not
by any means so essential for the achievement of intricate adjust-
ments to reality. It is therefore neither as universally developed, nor
as dependent on g for its successful exercise. Hence, it is possible
for a person with high g and low F to score far lower on these tests
than one with medium g and high F.
3. Whereas language tests and tests of the Binet type can be
adequately graded for all age-levels, performance-tests are usually
too simple for older persons and superior intellects. By making them
more difficult, one sets, as a rule, a harder task to some special ability,
rather than to g. Hence, the higher general intelligence, the less
performance-tests are able to measure it adequately.
The first difficulty (i.e. low reliability) does not necessarily
56 AFRICAN INTELLIGENCE
invalidate the performance-test for the purpose of interracial intelli-
gence studies. Unless a test is less reliable in the case of one race
than in the case of another-which is, of course, a possibility to be
reckoned with- both group-averages will be vitiated in like manner
and a valid comparison can still be made.
The second and third difficulties (i.e. differential validity) are
far more weighty. Performance-tests are really tests of special intelli-
gence, only partially measuring g, and measuring it unequally at
different age-levels. This seriously limits their usefulness for our
purpose, in view of the fact that races may differ markedly in their
possession of particular talents.
It would be idle to deny that there is such a thing as" the genius
of a people", even though this "genius" is, to a considerable extent,
culturally determined. To do or not to do a particular thing may
become second nature to a member of a cultural group to such an
extent that subsequent training in a different direction may be of
little avail. It would be strange if such group proclivities did not
initially spring from some preference or innate predisposition, and
if they merely perpetuated themselves by their own inertia as habits
and nothing more. The tendency for musical ability to run in fami-
lies is a case in point. It is fostered by a musical atmosphere in the
home, which creates suitable opportunities for the acquisition of
skill, stimulates interest, and develops taste. Yet, no amount of this
musical conditioning will be of anyavail unless certain basic capaci-
ties, pre-conditions for musical talent, are present. These capaci-
ties-such as a sense of pitch, intensity, time and consonance, musi-
cal memory, muscular control, to name but a few-are genetically
determined, and their occurrence in the children is more probable
when the parents also possess them than when the latter are funda-
Actually it makes very little difference, from the point of view
of mental testing, whether particular talents or deficiencies are
mainly culturally determined, or mainly innate ;test-scores will be
equally affected in either case. Great care must therefore be exerr
cised in the selection of performance-tests which are in keeping with
the genius of the group and which adequately allow for the expres-
sion of its g.
Observation and experience suggest that differences in respect
of special abilities do, indeed, exist between Europeans and Africans.
The Bantu have a flair for words and a remarkable ability to master
SUITABILITY OF EXISTING TESTS
languages. Many an African knows a few Bantu languages in addi-
tion to his own, while possessing a working knowledge of English
and Afrikaans. He does not seem to experience the same difficulty
in mastering the quite different structure of the Indo-European
languages, as the European has in learning a Bantu tongue. Latin
and English are amongst his best subjects in the High School
curriculum. Shakespearean plays are read with great relish, and .
large sections are learned by heart and recited for the sheer joy of
rolling the verses off the tongue, and despite the fact that much of
the content and many of the allusions must be unintelligible. Con-
sequently, the criticism by Dr. W. Eiselen, Chief Inspector of Native
Education in the Transvaal, of the use of non-language tests for
testing African intelligence, is very much to the point. He says :-*
"I think that the Natives should in the first place be tested in their
own language, as is done in the case of European children. The Bantu
are great speakers and it seems to be a pity that the 'speech medium'
should be excluded when their intelligence is being appraised."
As a group, Africans also excel in musical ability. Special
mention is made of this in a recent study of Native intelligence, based
on teaching experience in an African teacher-training college. The
author states :-
"that the speed with which Native students learn songs is remarkable.
We use the tonic solfa system in our singing lessons. It takes them about
ten minutes to learn an unknown piece of four lines in four parts. It is.
their custom after study and before they go to bed to sing some famil-
iar hymn. Night after night we hear them singing, but only in two
voices, namely bass and tenor. They say that the girls are singing the
other two parts and so they need only sing their two. It is most remark-
able that they succeed in singing in this manner, because the first voice,
which introduces the tune, is not heard at all. This makes it difficult
for us to make out what hymn they are singing. That it is possible for
the Native to do this shows that he is capable of representing imaginally
the voices which are not sung. If the first two voices were not imaginally
present they would not be able to keep their parts. Nearly every Native
is an amateur composer. Numerous new songs are sung annually here
by students in small choirs. It happens sometimes, when they are busy
working outside, that one of them hits on a tune-usually one he has
made up himself-and after he has repeated it a few times, one after
the other of the other students present joins in and, after a while, the
In a Preface to The Educability of the South African Native, by M. L. Fick,
1939, Research Series No. 8, S.A. Council for Educational and Social Research.
58 AFRICAN INTELLIGENCE
simple tune has become a part song with a perfect harmony, sometimes
in as many as five parts."
This shows that at least in the auditory sphere, the African's ability
to grasp, work out, remember, and create intricate new relations of
a most abstract kind is by no means inferior to that displayed by
the European in the visual'or conceptual sphere. Endemann further
Shows that African students are markedly deficient in ability to handle
visual spatial relations. Such concepts as "circumference", "length",
'width", are not clearly grasped. They find it difficult to abstract
these concepts from the objects, of which they are attributes. Hence,
African students make a very poor showing at certain types of arith-
metical problems, a subject which is usually their weakest. Though
Endemann's specific findings need corroboration, his general con-
clusions that the African attempts to solve arithmetical problems
without any.real understanding of what he is about, is fairly generally
accepted. He suggests that language difficulty, poor tuition, and
inferior intelligence might all be responsible, though he prudently
declines to commit himself to any one of these in the absence of con-
Performance-tests do not usually set problems of an arithmetical'
kind; but, they frequently demand ability to grasp dimensional
relations, as for instance in Alexander's Passalong Test, where a
solution achieved in Test 6 can immediately be transferred to Test
7, though different blocks are involved. Information as to the manner
in which African children do solve form boards, what difficulties
they encounter, and whether these are due to inability to abstract,
from the material given, the requisite relations, is urgently needed.,
Until this has been done, performance-tests of the form-board or
Passalong type cannot be used with complete confidence as equiva-
lent tests of innate intelligence for both Black and White. The
:skills involved appear to be foreign to the African people's genius, if
not fundamentally, then at least culturally.
Lest it be thought that this implies the African's innate infer-
iority in respect of conceptual thought, usually considered to be
the highest manifestation of human intelligence, it must be stressed
that his inability to make the requisite abstractions was noted only
in connexion with manipulative problems involving chiefly visual
clues and spatial relationships. Even if this conceptual weakness
were general, it need not be innate.
This opinion is shared by Nel, who conducted an exhaustive
SUITABILITY OF EXISTING TESTS 59
experimental study of the phantasy life of European and Native
school children.'3 He comes to the conclusion that, although at birth
the psychological process is qualitatively identical in both White and
Native, Native thinking ultimately proceeds..on a different level,
namely, "in a less developed and undifferentiated form"'4, "concrete
representational", rather than abstract.'" He attributes this differ-
entiation in the first place to the demands made by the environment,
and secondly to the opportunities provided within each environ-
ment to raise thinking to the abstract level.16
Hence, "the Native child, though completely adjusted to its
own more natural and, therefore, simpler environment, which does
not demand higher forms of thought and which does not present
such intricate situations as are common in modern civilization, is
less adjusted, or let us rather say adjusts itself less rapidly* than a
European child to a situation which requires higher mental func-
tions .... I do not wish to argue that [its] 'intelligence'-whatever
we may mean by that-is defective17 and therefore permanently
inferior to that of the European."
Though from a developmental point of view he considers the
present thinking-level among Natives to be lower than that found
among Europeans as a group, he holds that "there is a chance to
raise [it] to higher levels, so that in agreement with Loram we can
say that: 'the spread of civilisation, selective breeding, improved
environment and better teaching*, will undoubtedly tend to lessen
the mental differences between Europeans and Natives' ".s1
SUITAIIILITY OF EXISTING INTELLIGENCE-TESTS
From the foregoing discussion of the various types of intelli-
gence-tests at present in use, it should be clear that current attempts
to measure interracial differences in intelligence are premature. We
do not possess an instrument suitable for that purpose. The atti-
tude of "doing the best we can with existing means" is unscientific,
and cannot be too strongly condemned. What is required at present
is further intensive research into the abilities, response mechanisms,
and manner of thinking and of feeling, of the African, without
Dr. Nel's italics.
60 AFRICAN INTELLIGENCE
making evaluation and comparison the chief objects of the inquiry.
Nel's work on the imagination and general mental structure of the
African school-children, and Endemann's study of the quality and
determinants of African intelligence (both published in Afrikaans)
are examples of painstaking and objective research of the right type.
Much more of this kind of knowledge about the quality of the cogni-
tive process in Africans is required, before we can begin.to think of
measurement, more particularly of comparative measurement. In-
stead of concentrating on the quantitative aspects of intelligence-
test results, a detailed analysis should be made of the manner in
which the problems set are tackled and solved. Scores can have no
meaning in the absence of this knowledge. It is quite legitimate to
rank lead shot in order of heaviness by measuring the diameters;
but nobody would choose this method if the objects to be ranked
were a tennis ball, a balloon, an apple, and lead shot. Yet this is
the type of error which is constantly being made when innate intel-
lectual ability is to be measured. Just as scores in the diameter test
no longer measure differences in weight, if the diameters are obtained
from qualitatively different materials, so intelligence-test scores will
fail to reveal differences in innate intelligence, if qualitative differ-
ences in cognitive process are not taken into account.
As performance-tests suffer less from the extrinsic effects of
cultural milieu than any other tests at present in existence, they
provide probably the best starting point for a study of the intrinsic
influences, which, together with innate bent, determine African
intelligence. The "Drawing a Man" Test, Ferguson Form Boards,
Koh's Blocks Test, "Gottschaldt Blocks", Porteus Maze Test,
Knox Cube Test, C. Bihler's Vienna Test System, the Leiter
International Performance Scale, and the International Group Men-
tal Test, could all be used, but as research instruments, rather than
as tests. Manipulative and perceptual skills should also be carefully
examined by such means as are employed by students of vocational
aptitude. Those special aptitudes in which Africans appear to excel-
such as language ability, musical ability, and others which may still
emerge as a result of intensive study-should be analysed as possible
avenues through which general ability might be measured. The
necessity for careful control of those numerous factors already men-
tioned and still to be mentioned, which may influence any test result,
hardly requires to be stressed. In this study the use of the method of
factorial analysis elaborated by Thurstone,19 whereby the deter-
SUITABILITY OF EXISTING TESTS
minants of any activity can be isolated more fully and accurately
than has been the case up to now, is strongly advocated.
The ultimate construction of an interracial intelligence-scale
will only become a practicable proposition when it is possible to
validate the scale separately for the two racial or cultural groups
for which it is intended. This will mean the construction for each
group of a separate criterion of educability, as determined by the
needs of its culture, with which the proposed interracial scale should
correlate positively and highly. This task could obviously never be
carried out if educability in one environment should prove to have
nothing in common with educability in another. This contingency
is, however, so unlikely that it need hardly be discussed.
A far more probable difficulty is the differential development
of group-factors and special abilities in two cultures, to such an ex-
tent as to leave'an insufficient number of abilities in common, through
which g could be measured.
In such a case, intelligence would become a relative concept,
and the measurement of quantitative differences an impossibility.
A comparison of qualitative differences does, of course, also
invite judgements of value; but here one has to be exceedingly
careful. To meet the need of a particular cultural environment may
call for abilities which, however much we may despise or look down
upon the ends which they subserve, are in themselves difficult of
achievement. If only we could find a common criterion, we might
find as much g required for aptitudes common to Western culture,
as for those practised by what we choose to call 'primitives'.
It might be argued that cultural environment could be placed
into some sort of hierarchical order on the basis of wealth of insights
into reality, and diversity of demands made on adaptability. Educa-
bility for the environment placed at the top of the hierarchy could
then be made the final touchstone of superiority and inferiority.
As we are all, however, children of some culture, and we are
likely to be biassed in favour of that culture to which we are adapted,
who would be competent to construct such a hierarchy ? Further-
more, what proof have we that those who are perfectly adjusted to
a culture low down in the hierarchy, could not have achieved the
higher measure of adjustment demanded by the culture at the top,
if they had been fully initiated into it from birth ?
Throughout this discussion, cultural milieu has been spoken
of as if it were the same for all Africans. While this is probably true
62 AFRICAN INTELLIGENCE
in the case of Africans living in the Reserves, at least in respect of
the most -dominant features,, the cultural milieu of farm labourers
and particularly of the wholly detribalized town-dwellers has
markedly diverged from tribal pattern. We do not yet know whether
the African in the town has more in common, mentally, with the
tribal African or with the European. Until we have such knowledge,
the assumption that there is "an African cultural milieu" is a danger-
ous one. The differences may be great enough to compel the con-
struction of a number of-distinct scales.
CAN THE CONTROL-GROUP DEVICE BE USED ?
The conclusion is inescapable that, so far as interracial intelli-
gence studies in this country are concerned, the control-group device
cannot be used in its usual form, as the most important element of
all, the measuring device, cannot be equally controlled and standard-
ized for the two groups. Even if we were to concentrate on the group
which is closest to us, and if we used a test least affected by cultural
factors, complete equivalence could not be achieved. We would
furthermore be quite in the dark as to whether a detribalized sample
represented a superior or an inferior selection of the African popula-
tion as a whole.
For better or for worse, interracial comparisons have, however,
been made. In assessing the observed differences, we should note
particularly the possible occurrence of the following extripsic influ-
ences of cultural milieu on test-intelligence :-
1. Were steps taken to control the testee's attitude towards the
test-situation ? This could be done by :-
(a) Allaying suspicions as to the motives behind the inves-
(b) The utilization of a trained and trustworthy African
(c) The selection of a suitable testing time.
(d) The use of some incentive to excel.
(e) Giving instructions in the vernacular wherever possible
2. 'Was unfamiliarity with the meaning, content, symbolic signi-
ficance, or manipulative attributes, of the test-medium elimina-
ted or reduced to a minimum?
Control should take the form of:-
.SUITABILITY OF EXISTING TESTS
(a) Elimination as far as possible of the speed element in
the test to be used.
(b) Exclusion of culturally foreign material, verbal or
pictorial in form and European in content.
(c) Training in the purely manipulative responses required
The extent to which current investigations have controlled
the intrinsic influence of cultural milieu on test-intelligence cannot
be assessed until more is known about the quality of the African's
thought processes, about his perceptual habits, and his possession
of specific skills and interests. As we have seen, control is here as
yet impossible. We can safely state that wherever a difference in
test-intelligence is found in favour of a European control group, part
at least of this difference, even in the best controlled investigation,
must be due to adverse influences caused by the African's cultural
milieu. It is impossible to assess the extent of this influence on test-
intelligence in terms of I.Q. points.
1. Porteus, S. D. and Babcock, M. E.
Temperament and Race. Boston,
Badger, 1926. p. 219.
2. Malherbe, E. G. ed. Educational
Adaptation in a Changing Society.
Cape Town, Juta, 1937. p. 449.
3. Blackwood, B. "A Study of Mental
Testing in Relation to Anthro-
pology". Mental Measurement
Mon. No. 4. Williams and Wil-
kins, 1927. p. 48. (Quoted.)
4. Loades, H. R. and Rich, S. G.
"Binet Tests on South African
Natives, Zulus". Ped. Sem.,
V. XXIV, 1917. pp. 372-80.
5. Anastasi, A. Differential Psycho-
logy. London, Macmillan, 1937.
6. Nadel, S. F. "The Application
of Intelligence Tests in the An-
thropological Field". In: F. C.
Bartlett, ed. The Study of So-
ciety. London, Kegan Paul,
1939. Ch. VIII, p. 190.
7. Mead, M. Coming of Age in Sa-
moa. London, Cape, 1929.
p. 297. (Quoted by Nadel op.
cit. p. 190.)
8. Blackwood, B. Op. cit. p. 55.
9. Garrett, H.E. and Schneck, M.M.R.
Psychological Tests, Methods and
Results. New York, Harper, 1933.
10. Blackburn, J. M. "Intelligence
Tests". Op. cit. ref. 6. Ch. VIII,
11. Garrett, H. E. and Schneck, M.M.R.
Op. cit. pp. 87-90.
12. Endemann, T. M. H. Die Intelli-
gensie van die Naturel in dieLig
van Pedagogiese Bevindings aan
die Bothsabelo Opleidingskool vir
Naturelle, M.Ed. Thesis. Pre-
toria, University, 1927.
13. Nel, F. B. Die Fantasie van Blanks
en Naturelle Skoolgaande Kin-
ders. Amsterdam, Swets &
14. Ibid. p. 303.
15. Ibid. p. 309.
16. Ibid. p. 201.
17. Ibid. p. 303.
18. Ibid. p. 308.
19. Thurstone, A. A. The Vectors of
Mind. Univ. of Chicago Press,
The assessment of the relationship between home-environment and
the growth of intelligence is no easy task. The term "home-environ-
ment" comprises a multitude of influences, from purely material
ones, such as number of people per room, to intensely personal
ones, such as emotional relationship between parent and child.
In view of these many difficulties, it should not occasion sur-
prise that the current findings on the extent to which home-environ-
ment can influence the growth of intelligence, vary appreciably.
Careful consideration of the data (to which reference has already
been made in a general way in Chapter II) reveals, however, a
marked underlying agreement, which makes it possible to state that
extreme home-environmental conditions may raise or depress an
individual's test-intelligence from 10 to 30 points above or below the
valueitwould have had under normal environmental conditions, while
lesser deviations from the normal average may shift the I.Q. byfrom
3 to9points. Itwillbeworthwhilereviewingtheevidenceonwhichthis
assessment is based, more particularly for the light which may be thrown
ontherelativeimportanceofvariousaspectsof the home-environment.
The most noteworthy recent investigation into the effect of
home-environment during the first years of life is that of Skeels.'
His experimental group consisted of 73 foster-children, all of whom
had been placed in foster-homes before they were 6 months old,
their mean age at placement being 2.5 months. Case histories of
the true parents showed that these children came from rather low
educatiorial and occupational levels of society (Figure 4). As their
antecedents were known to the placement agency, though not to the
adoptive parents, it was possible to discover whether there was any
tendency towards selective placement. Skeels found that placement
was entirely random. When the children were tested, at a mean
age of 24.4 months (S.D., 14.2), the mean I.Q. turned out to be no
less than 115.3. This is not only vastly superior to the mean I.Q.
of their true mothers (a sample of which was tested, yielding a mean
I.Q. of 83.3), but also to the mean of an unselected sample of the
population (Figure 5). Taking their somewhat inferior heredity
(PFRO SxEsLS. PAGE 96t
DISTRIBUTION OF OCCUPATIONAL LEVELS OF
TRUE AND FOSTER FATHERS.
ILv.L 1 PR rS1160oH* OCCuPvIe. LCVct VII 01 LAkOUEI)
IFTO Or1t1 Lo VEts sE TABLte 11
s. ... .. i--. ..---
--- TFo-- r-UE P- -----
I ii III IV V VI Vi
OCcuPnohlat G0oup Leverl
NO. T7ur FATNEs O0
No0 Fosrt FArTatEI B
4 4 2 12 9 14
4 27 15 14 4
Ia SX;ELS floe 981
into account, they would under normal environmental circumstances
probably have ranked just below the mean for the population as a
whole, though obviously not as low as their true mothers, whose
test-intelligence had itself been depressed by unfavourable environ-
mental influences. The probable gain in their mean I.Q. as a result
of placement in these foster-homes was, therefore, nearly 20 points.
What attributes of the home-environment were responsible
for this considerable increment above the normal level ? The foster-
group consisted of families representative of all socio-economic
levels of the community, from the unskilled to the professional.
The criteria applied to determine the suitability of applicants who
wished to adopt a child, ensured, however, that within each group
only families above the average were selected. No family was en-
trusted with the final adoption of a child until it had convinced the
agency of its suitability over a probationary period. The bricklayer
had to be as successful in his trade as the doctor in his profession.
This meant that he should have no debts; that he should have
been able to maintain his home satisfactorily over a period of years ;
that he should be able financially to incur the additional obligation
of an extra member qf the family; and that his adjustment to the
community should be satisfactory. References were required from
banker, minister, doctor, and at least one other, favourably recom-
mending the placement of the child in the home.
It seems also reasonable to assume that people who actually
take the trouble to adopt a child when they have none of their own,
are above average in their keenness to have children. They are likely
to take a more lively interest in the child's development and to by
more sensitive to its need for attention and affection. They may
make a greater effort than the average parent to provide suitable'plae
material, to stimulate the child to take an interest in the environ-
ment, and to answer its questions.
In addition to the upward selection occupationally within each
group, it must have been this superior educational and parental
attitude which was responsible for the observed upward shift of the
I.Q. As Skeels found zero correlation between socio-economic status
of foster-parents and test-intelligence of foster-children, it appears that
these attitudinal qualities cut right across the socio-economic differences
between the members of the foster-group (Tables IIa and IIb).
The selection of none but well-adjusted homes in each occu-
pational group ensured that this favourable parental attitude could
TABLE II a
RELATIONSHIP BETWEEN EDUCATION OF FOSTER-PARENTS AND. I.Q. OF FOSTER-
Number Corr. Coeff. Prob. Error
Education of Father 72 +.01 .08
Education o Mother 71 -.10 .08
Mid-Parent Education 70 -.04 .08
Mean I.Q. of
Occupational Levels Foster-
of Foster-Fathers Number Children S.D. of I.Q. Median I.Q.
Groups I, II, III com-
bined (all children) 39 115.7 11.4 114
Groups IV, V, VI, VII
combined (all children) 34 114.9 13.2 116.5
Groups I, II, III com-
bined (children below
24 months) 24 117.1 11.7 114.5
Groups IV, V, VI, VII
below 24 months) 26 118.3 11.5 119.5
Groups I, II, III com-
bined (children above
24 months) 15 113.5 10.2 113
Groups IV, V, VI, VII
above 24 months) 8 103.7 12.3 102
Classification of Occupations, according to Fourteenth Census of the United
States, Vol. 4, 1920.
I-Professional; II-Semi-Professional and Managerial ; III-Clerical, Skilled
Trade, and Retail Business; IV-Farmers; V-Semi-Skilled Occupations,
Minor Clerical Positions, and Minor Business. VI-Slightly Skilled Trades
and Other Occupations requiring Little Training or Ability; VII-Day
Labourers of All Classes.
realize itself to the full, unencumbered by economic worries and
social and personal conflicts. Provided, therefore, that certain
material needs are reasonably met, differences in parental solicitude
and attention, rather than differences in parental occupation, edu-
cation, and culture, determine the unfolding of innate intellectual
potentiality, at least within the first two or three years of life.
There is a good deal of evidence in support of the view that
parental attitudes and general stimulativeness of the home can have
so pronounced an effect. An experiment was conducted in which
"two groups of two-year-old children living in the same institution
were segregated from each other and subjected to two divergen,
types of treatment. One group was given very little tendernesst
although adequately cared for in every other respect. In the other
group a nurse was assigned to each child and there was no lack of
tenderness and affection. At the end of half a year the first group
was mentally and physically retarded, in comparison with the
second." 4 Gindl and Hetzer tested 20 children in an institution,
20 in their own family where they were very much neglected, and
20 in foster-homes. All were between 1 and 2 years of age. Table III
shows their respective placings both in developmental tests and in
GENERAL RETARDATION AS A RESULT OF UNSTIMULATING ENVIRONMENT
(GINDL AND HETZER)
Environmental Situation Hetzer-Koller Hetzer-Reindorf
Developmental Test Language Test
Institution Children 47 36
Neglected Family Children 78 80
Foster-Children 90 95
Though the institution children were well cared for physi-
cally, their mental retardation was pronounced. This must be attri-
buted to the fact that institutions generally present a highly simpli-
fied and standardized environment which provides little stimulation
and few social contacts, most of which tend to be of an impersonal
type. Table IV, based on observations which extended through the
entire 24 hours of the day, illustrates the difference in respect of
social contacts between the institution and the family situation.
TABLE IV *
NUMBER OF CONTACTS BETWEEN CHILD AND OTHER PEOPLE DURING ONE DAY (STURM)
Age in Years Institution Family
0-1 25 137
1-2 62 261
2-3 169 243
3-4 178 279
4-5 133 263
5-6 177 226
As to kind of contact, Biihler's summary of Sturm's findings
shows that "a five-year-old institution child receives 85 orders in
the same period of time during which the family child receives 21 ;
.the two-year-old family child asks 69 questions during one day,
while the institution child has so much less opportunity for that,
that he has only 10 questions answered; while the family child has
things explained in 28 situations during the day, the institution child
70 AFRICAN INTELLIGENCE
receives only three explanations With the two-year-old children
it was still possible to make up for the loss suffered because of their
institutional life when they were brought into favourable milieu
(Hetzer), but with children older than that, it seems impossible to
compensate for their early privation."7 Table V is of some interest
in showing the effect of home-environment on language develop-
ment, which is always highly correlated with test-intelligence.
RETARDATION IN THE DEVELOPMENT OP SPEECH IN CHILDREN IN POOR MILIEU
Age, Years and Months Good Milieu Poor Milieu
Percentage of Children using Meaningful Words
0:8-0:11 65 0
1:0-1: 2 91 40
1:3-1: 5 100 71
1:6-2: 0 100 100
Number of Words which are used
1:0 7 0
1-:3 49 1
1:6 91 4
1:9 121 8
2:0 216 27
2:6 Too many to be scored 92
Although homes differing in socio-economic status, but sharing
an attitude of interest in child-rearing, may be equally stimulating
to children during the first three years of life, thereafter higher pa-
rental intelligence, the greater material and mental culture of the
superior home, and the more plentiful economic means at the dispo-
sal of the family for amusements, travelling, and education, are
likely to make themselves felt. The children in homes higher up
the socio-economic scale will be taken more frequently to parties,
on outings, and for holidays. They are more likely to be sent to good
nursery schools, and they are able to interact with a material environ-
ment of far greater diversity. As the child becomes responsive to
the conversation of adults, to pictures, books, and music, it. will
begin to enter a cultural world which the parents of lower education
and slender means will be unable to provide. Hence, it is reasonable
to expect that differences in socio-economic status, at first of little
importance, will assert themselves in due course. Though the child
in a lower cultural environment will, therefore, begin to lag behind
after the fourth year, it is doubtful whether the ultimate difference
will ever be more than 10 points in those cases where a sound found-
ation has been laid in the early years. This suggestion is supported
by the findings of Leahy, who studied the relationship between the
test-intelligence of 194 foster-children and various indices of their
adoptive homes.9 She, too, reduced the chances of selection to a
minimum by only including those cases in her experimental group
which had been adopted before they were 6 months old. (The mean
age at placement was 2.5 months.) She found the average difference
in I.Q. between children adopted by professional men and those
adopted by slightly skilled workers to be about points. This finding
is practically identical with that of Barbara Burks. She found that
the average I.Q. of a group of foster-children, similar in composition
to' Leahy's group as far as age of adoption and duration of stay in
foster-home were concerned, was raised by from 3 to 6 points through
placement in foster-homes from half to one standard deviation above
the "average" home-environment of the population as a whole.10 In
other words, Burks found that 662 per cent of all White American
homes did not raise, or depress, the I.Q. of children within those
homes by more than 6 points above, or below, what it would have
been in an average or standard home-environment. This range is, of
course, quite considerable, making a shift of 12 points possible,
according as a child of normal innate ability had the good or bad
luck to be born into a home-environment one standard deviation
above or below the mean.
Burks and Leahy differ in that the latter holds this amount to
be the total possible effect of home-environment on the I.Q.11, where-
as the former states that if deviation from the mean exceeds 1 S.D.,
greater variation in the I.Q. is possible, amounting to over 20 points
in extreme cases."2 By "extreme cases" she means those 3 S.D. above
or below the mean, which occur once in 400 cases. Taking a varia-
tion of 20 points as a limit, this means that a child of normal innate
intelligence might have had an I.Q. of 80 or 120, according as it had
been reared in a very inferior or a very superior home-environment.
Leahy ignores the fact that her sample of foster-parents by no means
covered the entire range of homes in the United States ; and that
even those in her group with the lowest occupational status were
probably by no means inferior in respect of attitude towards the
child, otherwise they would not have been permitted to take a child
for adoption. Her figures do not take into account the common in-
crement observed by Skeels. Yet the fairly high mean I.Q. of her
foster-group (110.5, S.D. 12.5) shows that it too must have profited
from such an increment. (We are assuming that the ancestry of the
two groups was identical. An examination of the occupational status
and education of true fathers in the two groups shows that this
assumption is justified ; see tables on pp. 95-96 in Skeels, p. 295 in
It might be objected that while Burks did take into account
environmental conditions in the home more extreme than were re-
presented in her foster-group, the limits of .variation which she
thought possible were calculated and not actually observed. This
objection can be countered to some extent by the work of Freeman,
Holzinger, and Mitchell,13 in whose group the' range of foster-home
environments was far wider than in the other investigations, as a
number of coloured foster-homes were included, and the percentage
of professional, semi-professional, and business, foster-parents was
higher than that of Leahy or Skeels. (See Freeman p. 144, Leahy
p. 285, Skeels p. 96.)
Tables VI, VII, VIII, IX, and X, show the relationship found
by Freeman between grade of foster-home and intelligence of foster-
children under various conditions of adoption. Table IX is parti-
cularly striking, showing a difference of no less than '17.9 points
between the mean I.Q. of children reared in good and in poorfoster-
homes. Table X gives only those cases who spent at least 4 years in
foster-homes and who entered (a) before the age of 5 years, (b)
before the age of 2 years. The means for legitimate and illegitimate
children have been given separately. Because the latter had a some-
what better heredity and happened to be placed more frequently
in homes of higher grade, they proved to be superior in each of the
RELATION BETWEEN THE INTELLIGENCE OF CHILDREN
AND THAT OF THEIR FOSTER-MOTHERS
Mean I.Q. of
Foster-Mothers' Intelligence Foster-Children Number
Superior: Mean I.Q. 115 104.9 64
Range in I.Q.s 108-132
Average: Mean I.Q. 100 97.8 91
Range in I.Q.s 92-107
Inferior: Mean I.Q. 83 94.3 100
Range in I.Q.s 63-92
RELATION BETWEEN THE INTELLIGENCE OF CHILDREN
AND THAT OF THEIR FOSTER-FATHERS
Mean I.Q. of
Foster-Fathers' Intelligence Foster-Children Number
Superior: Mean I.Q. 122 104.9 68
Average: Mean I.Q. 102 97.0 61
Inferior: Mean I.Q. 87 93.8 51
TABLE VIII "1
INTELLIGENCE OF CHILDREN ACCORDING TO THE OCCUPATIONAL STATUS OF THEIR
Mean I.Q. of
Occupational Class Foster-Children Number
Professional 106.8 61
Business 101.1 160
Skilled Labour 91.6 149
Semi-Skilled to Slightly
Skilled Labour 84.9 19
Unskilled Labour 5
INTELLIGENCE OF FOSTER-CHILDREN ACCORDING TO GRADE OF HOME
Type of Home
Mean I.Q. of
TABLE X '
INTELLIGENCE OF FOSTER-CHILDREN ACCORDING TO GRADE OF HOME, LENGTH OF TIME
SPENT IN HOME, AND AGE AT COMMITMENT
Entered before 5
Entered before 2
Legitimate Illegitimate Legitimate Illegitimate
Mean I. Q. No. Mean I. Q. No. Mean I. Q. No. IMean I.Q. No-
101.3 36 109.4 48 106.5 10 112.4 45
94.5 44 104.4 35 98.3 19 104.9 39
88.8 34 96.1 29 83.4 16 95.6 27
Combining early commitment with superior grade foster-home,
there is a difference of 16.3. I.Q. points between most and least
74 AFRICAN INTELLIGENCE
favourably'placed illegitimate children, 17.7 I.Q. points between
most and least favourably placed legitimate children. In the group
of legitimate children adopted before they were 2 years old, the
difference is no less than 23.1 points, though little significance can
be attached to this figure as the number of cases is too small. As
Freeman's superior, average, and poor homes probably cover a
range more than 1 S.D. above and below the mean, the corres-
pondence with the findings of Burks is remarkably close.
Freeman applied various checks to his data to make sure that
selection was not responsible for the observed relationships. There
were 59 cases in his group about whose parentage practically
nothing was known. The mean age at which these children were
placed was 2 years 4 months. At this age it is not so easy to pick out
the intelligent child merely by its looks and general demeanour. In
this group, the possibility of selection was therefore much reduced.
Nevertheless, the I.Q. of these children correlated .51.06 with
foster-home rating, which is higher than the correlation for the
*entire group. As Negro children were placed in Negro homes, there
was a certain amount of selection according to race ; but omission of
theNegro childrenleftthe correlation between child's I.Q.and foster-
home rating for the remaining group unaltered, so that this selective
factor obviously had no bearing on the observed relationship.19
It is important that we should know which of these influences
have a measurable effect on the development of intelligence, and
to what extent they can affect it, either singly or together. Without
'such knowledge it is impossible to arrive at a criterion of home-en-
vironment. Nor will it be possible to state in any given case to what
extent an observed I.Q. is higher or lower than it would have been
under "normal" home-environmental conditions. It has proved
extremely difficult, however, to obtain precise factual information
on the relationship between various aspects of home-environment
and the growth of intelligence.
A first difficulty springs from the elusive nature of some of
the home-environmental qualities themselves. Parental attitude
towards the child, intellectual and cultural atmosphere of the home,
the power of the home to stimulate the child, these are all influences
which are likely to be judged differently by different observers.
The use of the correlation technique to determine relationship
between test-intelligence and home-environment requires, further-
more, that these judgements be stated in quantitative form. This
necessitates the construction of elaborate rating scales. The margin
of error in these ratings is obviously fairly wide, particularly where
more than one rater is used.
The elimination of the influence of heredity presents a second
difficulty. The inheritance of intelligence is such a complex affair,
that one cannot say more than that the chances of begetting intelli-
gent offspring are higher in the case of intelligent than of unin-
telligent parents. If the quality of home conditions depends, in
part, on the innate intelligence of the parents founding that home,
then the difference between the I.Q. of children born in good and
poor homes must be partly environmentally, partly hereditarily,
determined. Complex experimental and statistical methods have
to be resorted to in order to discover how much is contributed by
each. A favourite device is to use foster-children and to study the
relationship between the magnitude of their I.Q. and the quality
of home into which they were adopted. The same relationship is
thereafter determined in a similarly constituted group of parents
and their own children. This device removes the heredity difficulty
only to re-introduce it again in another form. For, let us suppose
the findings are that children adopted into superior homes have,
on the average, a higher test-intelligence than those placed in a less
favourable home-environment. The issue then becomes to what
extent superior parents succeeded in selecting children with superior
endowment for adoption. Investigators have'found this selection
factor most difficult to control.
1. Skeels, H. M. "Mental Develop-
ment of Children in Foster
Homes". J. of Genet. Psychol.
V. XLIX, 1936. pp. 91-106.
2. Ibid. p. 102.
3. Ibid. p. 103.
4. Buhler, C. From Birth to Maturity.
London, Kegan Paul, 1936.
5. (Quoted) Handbook of Child
Psychology. Worcester, Mass.,
Clark Univ. Press, 1931. p. 422.
6. (Quoted) Ibid. p. 422.
7. (Quoted) Ibid. p. 422.
8. (Quoted) Ibid. p. 423.
9. Leahy, A. M. "Nature, Nurture
and Intelligence". Genet. Psy-
chol. Mono. No. 17, 1935. pp.
10. Burks, B. S. "The Relative Influ-
ence of Nature and Nurture
upon Mental Development".
In: 27th Yearbook of the Nat.
Soc. for the Study of Education,
1928. pp. 306-07.
11. Leahy, A. M. Op. cit. p. 304.
12. Burks, B. S. Op. cit. p. 309.
13. Freeman, F. N., Holzinger, K. J.,
and Mitchell, B. C. "The In-
fluence of Environment on the
Intelligence, School Achieve-
ment and Conduct of Foster-
Children". Op. cit. in ref. 10.
14. Ibid. p. 177.
15. Ibid. p. 177.
16. Ibid. p. 178.
17. Ibid. p. 173.
18. Ibid. pp. 174 and 175, from tables
XXXII and XXXIII.
19. Ibid. pp. 179-85.
Having determined the total effect of home-environment on test-
intelligence, we can now proceed with a detailed consideration of
the factors which, together, make up this influence. So far, only
parental attitude and socio-economic status of the home have been
mentioned. The latter not only comprises parental intelligence and
education as, separate entities, but also their material and spiritual
products, auch as tasteful furnishing; an atmosphere of neatness
and distinctiveness ;presence of books, newspapers, and periodicals
.of various types; interest in music and art; intelligent and stimu-
latiig conversation; etc.
In Table XI, the relationship between a variety of home-en-
vironmental influences and test-intelligence of foster-children is
.stated in correlation form. Though these correlations are low, all
INTELLIGENCE QUOTIENTS OF FOSTER-CHILDREN CORRELATED WITH
FREEMAN BURKS LEAHY
Corr. Corr. Corr.
FACTOR Coefficient No. Coefficient No. Coefficient N
Foster-Father's Intelligence .37 .04 180 .09 .06t 178 .19 .06 178
Foster-Mother's Ihtelligence .28 .04 255 .23 .061 204 .24 .06 186
Foster-Father's Education .01 .05 173 .19'.06 193
Foster-Mother's Education .174 .05 194 .25 .06 192
Mid-Parent's Education .42 .03 .24 .06 193
Foster-Father's Vocabulary .27 .05 152 .14 .06f 181 .26 .06 177
Foster-Mother's Vocabulary .37 .04 224 .25.06t 202 .24.06 185
Occupational Status .37 .04 394 .14 .06 194
Income .26 .06 181
Culture Index .29 .06 186 .26 .06 194
Whittier Index .24 .06 206
Foster-Parents' Rating .49 .03
Environmental Status Score .48 .04 401 .421--. 164 .23 .06 194
Compiled by the writer from the monographs of Freeman, Holzinger, and
Mitchell; Burks and Leahy.
t Corrected for attenuation.
Multiple Correlation, using coefficients corrected for attenuation
but four are significant. Parental intelligence was determined by
means of the Otis Self-Administering Test, and in the case of
Burks's investigation by means of the Stanford-Binet. This test is
not very suitable for adult subjects, which may account for the low-
ness of Burks's correlations. Parents' Education Score was based
on the number of years of formal schooling they had received. Free-
man used a five-point scale, one point being given for less than four
years' schooling, five points for 16-21 years. Freeman constructed
a special vocabulary test by selecting words from various fields of
knowledge, while Burks used the Vocabulary Test in the Binet
The foster-parents' personal rating was carried out by trained
field-workers, who based their judgements on congeniality and har-
mony between parents; the degree to which they co-operated in
training the child; their apparent intelligence and disposition;
their refinement; speech ; personal appearance ; kindliness; personal
interest in the child; and their methods of discipline.
Occupational status was rated : (1) according to Taussig's Five
Point Scale, as Professional, Semi-Professional and Business, Skilled
Labour, Semi-Skilled to Slightly Skilled, and Unskilled Labour;
(2) according to the sevenfold classification of the Fourteenth Cen-
sus of the United States, which makes separate provision for farmers
and the slightly skilled (Leahy); (3) by means of a still more detailed
nine-point scale of a rather unconventional type.1 For the subtler
cultural indices, a variety of rating devices was used. The Whit-
tier Home Scale, used by Burks, provides ratings under five head-
ings, namely Necessities, Neatness, Size of Home, Parental Condi-
tions, and Parental Supervision. Points are granted on a five or six
point scale under each of these headings, detailed examples being
provided to serve as a guide to the rater.* Burks measured the more
specifically cultural attitudes, of the'home by rating Speech, Educa-
tion, Interests of Parents, Home Library, and Artistic Taste, on
specially constructed five-point scales. These ratings were then com-
bined into a Culture Index. The information on which the ratings
were based was obtained by field-workers actually visiting the
homes, as well as from information blanks which the parents
were asked to fill in. The "Artistic Taste Rating Scale" may serve
as an example of the method of rating and of the criteria employed :
For details, see Whittier, Social Case History Manual, Calif. Bureau of
Juvenile Research, Bull. No. 10, 1931 ; or Burks, op cit., pp. 231-233
78 AFRICAN INTELLIGENCE
1 point-No musical instrument, no pictures or only those of
the most inartistic type; no taste in furnishing.
2 points-Possibly a musical instrument, but furnishing and
pictures distinctly inharmonious; trashy ornaments, such
as gaudy bric-a-brac scattered about.
3 points-Nondescript; nothing offensive to the eye. Usually
some kind of musical instrument, but music listened to not
of high type. Pictures are cheap prints, poorly framed.
Family photographs abundant.
4 points-Noticeable effort to make home somewhat beautiful.
Rugs and hangings blend well. Fairly good music. Fur-
niture and pictures do not clash. No trashy ornaments.
Family photos absent or present in small number.
5 points-Effect of interior beautiful. Fine taste used in blend-
ing curtains, rugs, furniture and pictures, though these
things are not necessarily of the expensive sort. Musical
selection from standard composers, though a little jazz may
be included. Ornaments selected with discrimination and
Leahy used three information blanks, entitled The Child and
his Environment, on which answers to questions put by field-wor-
kers were entered. Her monograph provides no information on the
nature of these questions and the method of rating.
Freeman's Environmental Status Score was a compound rating,
comprising assessment of material environment, evidences of cul-
ture, occupation of foster-father, education of foster-parents, and
social activity of foster-parents. The Material Environment Rating
was based on geographical situation of residence, on type of neigh-
bourhood, type of house, number of rooms per head of family, type
of rooms, ventilation and light, size of grounds around house, com-
forts and conveniences, adequacy and quality of furnishings, and,
finally, care and upkeep of house and grounds. The Culture Rating
was based on number and quality of books, newspapers, magazines,
and pictures in the house, as well as on the type of music listened to
and musical life generally. Social Activity covered use of leisure
time and membership of church, clubs, and societies.
The final placement of a foster-home on this Home Rating
Scale represented the relative superiority or inferiority of the tota-'
lity of home-environmental influences to which the foster-child
was subjected. Of the correlations on Table IX, that between the
Environmental Status Score, as thus determined, and the I.Q. of
the foster-child is, therefore, most important. After the correlation
with Foster-Parents' Rating, it is the highest obtained by Freeman.
Statistically, Burks's coefficient of .42 cannot be directly compared
with that of .48 obtained by Freeman, as the former is a multiple
correlation, based on distinct correlations between various home-
environmental factors and foster-child's I.Q. Burks did not compile
an all-inclusive Home-Environmental Index, similar to that of Free-
man, and had, therefore, to combine statistically the effect of a
number of separate influences. Ultimately, however, these two
devices, though different in character, aim at measuring the same
existing relationship. This difference in procedure makes the close
agreement between the findings all the more impressive. When it
is remembered that the compositions of the experimental groups
studied in these investigations differed considerably, that each in-
vestigator had his own method of assessing environmental factors,
and that stress was not always laid equally on the same aspects,
the discrepancies in the findings become less significant than the
agreement. The conclusion that the relationship between test-
intelligence and the totality of home-environment can be stated in
terms of a correlation coefficient of magnitude between .4 and .5
is therefore justified. It is unnecessary to state the precise physical
meaning of this statistical relationship, as the extent to which the
I.Q. can be raised or depressed by home-environment has already
been indicated. The main purpose of the examination of the corre-
lation material was to indicate which of the attributes, into which
the general term "home-environment" can be resolved, were posi-
tively related to test-intelligence, and approximately to what extent.
Without such knowledge, it is obviously impossible to equate home-
environments for control-group purposes.
It appears that there is hardly any aspect of the home-environ-
ment which does not at some time or other exercise a measurable
influence on test-intelligence. While a certain degree of material
well-being is a pre-condition for normal development, its chief
function is the provision of a favourable medium in which parental
solicitude and interest in education can operate. Parental intelli-
gence and education, as well as a multitude of material and mental
cultural influences, may cause further differentiation. The effect
of any one of these influences singly, with the exception of parental
atittude and general material well-being, is never very great. It is
their joint effect which is important. The magnitude of this effect,
in any given case, can only be determined by checking the presence
or absence of all possible cultural influences singly, as was done by
means of the rating-scales discussed above.
The question whether home-environment merely influences
test-intelligence, without affecting quantitatively the development
of innate intellectual potentiality, must be considered before we
turn to the home-environmental conditions obtaining among the
Union's African population. It may be argued that an inferior per-
formance on the Binet Scale on the part of a child reared in a poor
home-environment, is perhaps less due to lack of general ability
than to lack of the necessary skills and culturally established asso-
ciations. A number of facts suggest that only part of the inferiority
in test-intelligence, displayed by a child reared in a poor home, can
be accounted for on these grounds. In addition to the Binet Scale,
Freeman also applied the International Group Mental Test to his
foster-children. This test, a performance-scale which dispenses with
the use of verbal material and scholastic knowledge, is relatively
free from the effect of extrinsic environmental influences within
the same cultural milieu. Freeman found that it agreed with the
Binet Scale in "showing the influence of good home environment
upon intelligence."2 The difference in test-intelligence between
those reared in good and those in bad homes, though not quite as
large on the International Test as on the Binet Scale, was yet quite
All investigators agree that the earlier the age of adoption, the
higher will be the I.Q. ultimately. Though more evidence is required
before Hetzer's statement, that it is impossible to compensate for
early privation in children older than 2 years, can be accepted as it
stands, it is at any rate certain that, after the age of 5 years, a change-
over to a good environment is comparatively ineffective. Such a
change may stem further deterioration, but it can have no bearing
on those past stages in the maturation process at which the rate of
mental growth of the organism was determined. Hetzer's finding of
general retardation on her developmental tests, as a result of un-
stimulating environmental conditions, is particularly significant in
It appears, therefore, that inadequate environmental stimula-
tion may reduce the I.Q. absolutely. A poor home-environment not
only retards the rate of development of innate intelligence, but also
brings it to a close at an earlier age than would normally have been
the case. Those levels of development at which subtler environmen-
tal influences might provide further stimuli for growth will never
be reached. The effects of home-environment on intelligence are
therefore largely quantitative and intrinsic, leading in the case of
adverse conditions to a non-realization of natively given potentia-
lities for intellectual development, which is permanent in its effects.
1. Burks, B. S. "The Relative Influ-
ence of Nature and Nurture
upon Mental Development". In:
27th Year Book of the Nat. Soc.
for the Study of Education, 1928.
2. Freeman, F. N., Holzinger, F. J.,
and Mitchell, B. A. "The In-
fluence of Environment on the
Intelligence, School Achieve-
ment and Conduct of Foster-
Children". Op. cit. in ref. 1.
3. Ibid. pp. 132-33 and 176.
THE CONTROL OF
HOME-ENVIRONMENTAL INFLUENCES IN THE
AFRICAN EXPERIMENTAL GROUP
From the evidence adduced in previous chapters," it is clear that,
for control-group purposes, a rough and ready division of home-
environments according to locality (rural or urban), education of
parents, or income, is inadequate.
The European control group must match the African experi-
mental group in respect of the following home-environmental fac-
1. Parental Intelligence.
2. Parental Education.
3. Parental Occupation and Economic Status of Home.
4. Material Culture of Home.
5. General Culture of Home.
6. Personal Attributes of Parents, with special reference to
The attempt to apply the control-group technique in respect
of home-environment to the Union African population immediately
meets with a serious difficulty. Home-environments can only be equated
within the same cultural milieu. Every cultural milieu represents a
unique equilibrium-attempt between human needs and capacities,
on the one hand, and the specific attributes of the geographical
environment, including other men, on the other hand. Consequent-
ly, a home-environment which is admirably suited to develop to
the highest degree qualities necessary for adjustment to one cultural
milieu, may rank exceedingly low when the criteria of another cul-
ture are applied to it. It is impossible to assign to specific elements
in the home-environment an absolute value from a developmental
point of view. Each of these elements gains its significance from the
cultural totality within which it functions. Whereas, for example,
subservience to authority, rather than a critical and questioning atti-
tude, had adjustment value in untouched tribal communities, based
as they were on tradition, the opposite is true in our rapidly changing
Western civilization. African tribal family life develops intelligence
in directions entirely different from those of the European home,
The goodness or badness of its attributes can only be judged in terms
of its effectiveness in achieving its own purposes. In the absence of
some final culture-independent criterion of innate intelligence, it
might be possible to make a rough assessment of tribal family life,
on the basis, both of intimate knowledge of the dynamics of African
society from within, and of abstract psychological considerations.
No such study has as yet been made, so that we cannot even assign
a general value to tribal home-environment as a whole, in terms of
its effectiveness in developing innate intelligence, regardless of the
cultural type of the intellectual end product. Far less still can we do
this for its specific attributes, and for the range of internal variations
in quality. For the time being, tribal Africans will, therefore, have
to be excluded from the experimental group, as the necessary con-
trol-group conditions governing interracial comparisons cannot be
fulfilled. While this will facilitate the construction of a suitable
intelligence-test, it raises serious sampling problems, to be discussed
in a later chapter.
The exclusion of the tribal section of the African community
is not, however, as simple as it seems. Where does "tribal" end and
detribalizedd" begin? All degrees of culture contact and detribaliza-
tion are met with. At the one extreme there is the Native village
tucked away in a deep valley of the Zoutpansberg. It possesses no
school, and only a certain number of the men have been to the larger
urban centres. The majority have been no further than the European
farms in the neighboring districts.
Superficially, the material culture of such a community shows
few signs of contact. There is no European furniture. The men wear
European garments. The women have adapted manufactured cloths
to their own purposes. There are blankets, an odd paraffin tin, occa-
sional bits of building material and tools, and here and there some
food bought at the nearest store about thirty miles away. There is
no Church, and no signs of Christianity are evident. Though com-
munity life has been interfered with, its structure, particularly as
regards social relationships, remains intact.
The town locations are at the other extreme. The better ones
are similar in type to European Sub-Economic Housing Schemes,
though on a lower level. The worse ones are incredible slums, n ith
squalor, general overcrowding, and tumble-downness of structures
being the sole distinguishing features. Though the material environ-
ment of the kraal has entirely vanished in both types, traditional
relationships and Native institutions and customs may linger on in
a watered-down form. Houses may be kept as airless and closed as
the Native hut, and the children may continue to sleep with their
heads completely under a blanket so as to be safeguarded against
spirits. The excessively long suckling period customary in tribal
areas may be retained. Lobola is still paid in the majority of cases,
and the traditional husband-wife relationship frequently remains
unaltered. Home brewing and drinking of beer remain prime social
institutions, wherever the authorities permit it or where the law can
be evaded. The number of Africans living entirely as Europeans,
with a standard of living equal to that of the European lower middle
class or middle class, is very small. The mass, materially on or below
the level of unskilled workers in European countries, still differ from
the latter in their social life, having lost most of their own culture,
and having gained little from the European to put in its place. As
they stand nearer to the European working class than to their tribal
fellows, European home-environmental standards can, I think, be
legitimately applied to them. Where, however, must the line be
drawn between these two extremes of tribal life and detribalization?
Let us consider some of the intermediate groups. There are,
in the first place, the squatters on European Company Farms. Very
often they live in typical Native villages, owe allegiance to a chief,
and observe the practices of their people.
Labour-tenants are in a somewhat different position. They
usually have their own establishment on a European's farm, for
which they and their dependants give from 90 to 180 days' service
per year in lieu of rent. In the Northern Transvaal, they generally
live in kraals, though elsewhere any type of dwelling may be used.or
erected. Some farmers have compounds where they accommodate,
for the period of service, men who have their own establishments
on labour farms, kept by the employer specially for the purpose of
ensuring a ready labour supply. Even in those cases where a tenant
lives in a kraal with a number of wives, there is some measure of
detribalization as a result of absence of village life, and hence of
the larger social unit with its various institutions and functions.
It stands to reason that a general weakening of traditional social
relationships, customs, and observances must follow. A class of
labourers has already come into existence who are employed on a
full-time basis for wages in cash and in kind. All types of accommo-
dations, from tin shanties to reasonably well-built houses, are pro-
vided for them, while Native huts also occur. These labourers have
lost all connexion with the Reserves, owe no allegiance to a Chief,
and maintain a dwindling number of Native customs.
The differences between these various environments are more
marked than those obtaining in Western Europe between town and
country, elite and proletariat. There, at least, there is a common
social structure, identical kinship systems, the same religion, and a
common core of mores. Here two social systems are in conflict.
Chaos and formlessness, as well as all kinds of new syntheses and
partial assimilations, are the products of this clash.
Apart from this variety of distinct cultural milieus, there is also
the factor of movement between them, another unique quality of
African society to-day. The mines are chiefly responsible for the
circulation between town and tribal area. A considerable number of
tribal Africans spend some years of their early manhood in mine
compounds, from which they get some glimpses of the life of the
towns. They become familiar with a daily routine and mode of life
vastly different from that to which they were accustomed in their
village. A further number get their first taste of European civilization
as house boys in flats or private residences. It is said that even after
a number of years in town, as house boys or as miners, Africans will
settle down again to the ways of the Native village, rapidly shedding
such European habits as they may have acquired. Whether this
return to traditional life is as complete as it appears to be on the
surface is open to question. Furthermore, a large number, though
establishing a home in a Reserve, keep on travelling to and fro
between it and the town, where they work for a certain number of
months every year. This in itself, of course, markedly deflects the
home-environment from its original pattern.
,It also leads to the setting-up of irregular households in the
towns. Belonging neither to tribal culture, the standards of which
the men obey when back within its fold, nor to European culture,
these households constitute a veritable "no man's land" of human
relationships. Conduct within them is beyond the sway of tribal
sanctions, while barely subject to European customs and morals,
which, for this type of African, have as yet but little binding force.
The search for education constitutes another cause for move-
ment from one milieu to another. Numerous African students at
present in colleges and high schools, where all but an insignificant
few are boarders, have spent their early youth in the Reserves. Many
will return again, perhaps as teachers, agricultural demonstrators,
police constables, or clerks in the Native Affairs Department.
Though they will build themselves houses of a European type, fur-
nished on the hire-purchase plan, they will, nevertheless, retain close
contact with the tribal culture. While that is so, complete European-
ization is out of the question, despite membership of the Church
and the possession of a Matriculation Certificate. Thus, .quite a
number of students at a high school in Johannesburg insisted on
being circumcised when the time came for their age-group in the
Reserve to undergo this ritual operation. The majority will ulti-
mately marry in Church, but pay lobola and feast in the time-honour-
Clearly, therefore, home-environment cannot be classified as
detribalizedd" on the basis alone of material culture and the house-
holder's contact with European ways of living. The ideal procedure
would be to judge each case on its merits ; but as this is impossible
in a large-scale investigation, a more rough and ready criterion will
have to be employed.
The suggestion is, therefore, put forward that of the rural
groups all those living in Native villages or in kraais be excluded,
however closely some of these people may have been in contact with
European civilization ; also those squatters and labour tenants who,
though they may have departed from the tribal material culture,
still are intrinsically part of the tribal organization and subject to
its customs and attitudes. Students in boarding-schools should be
excluded if their early home-environment was tribal, or if that is
the home-environment to which they return during vacations.
Included may be those families which, though living in a Native
Reserve, dwell in houses furnished, however primitively, in the
European way, both husband and wife wearing European clothing
and following, superficially at least, a Western mode of life. The
husband should have regular daily employment as clerk, teacher,
agricultural supervisor, police constable, or labourer. The union.
should be monogamous and solemnized in Church or Native Com-
missioner's Office, payment of lobola to be permissible. Non-
attendance at school on the part of the children should be no dis-
qualification, though participation in an initiation school should be.
In the towns, those at least should be excluded who have only
recently migrated from a tribal area, where they lived in a tribal way,
as well as those who keep up polygynous establishments. It seems
advisable also to exclude the large number of customary unions
where lobola is paid, but no church or civil ceremony is gone through.
This argues close adherence to tribal custom in social relationships
and hence the continuance of an atmosphere and of attitudes foreign
to Western home-environment. Families which are the product of
irregular unions, solemnized neither by European nor by customary
rites, might be included, as the home atmosphere is certainly not
tribal, whatever else it might be. It is doubtful, however, whether
control cases could be found for them in the European group.
Here, then, we come across another stumbling-block. We saw pre-
viously that is was virtually impossible either .to control cultural
influences adequately or to construct a test suitable for both groups.
It was stipulated in Chapter III, that a sample of the African popu-
lation, to be representative, should contain tribal as well as detri-
balized individuals. It now appears that this condition cannot be
fulfilled. Although the exclusion of tribal Africans somewhat light-
ens the task of controlling cultural environment, our final result may
be in error to an unknown extent.
Special experiments will be necessary to provide proof of the
equivalence of tribal and detribalized groups,' if we are to generalize
from the data to be obtained.
THE HOME-ENVIRONMENT OF
1. THE GENERAL SETTING
The majority of urban African families live in locations, outside the
European residential and business areas of the towns, but under the
control of Municipalities, or in townships administered by Health
Committees. The remainder, in certain Municipalities a not incon-
.siderable number, dwell in areas of the European town, specially
set aside for their use and the use of other Non-Europeans. Though
there is nothing to prevent Europeans from living in these wards,
and a number in fact do so, Africans are debarred from living in
European areas, even when they own property there.
In the locations, both land and houses are owned by the Munici-
palities. Elsewhere, private ownership of site, house, or both, is
In none of these African communities is there any regional
division on the basis of class, despite thesfact that Bantu society is
not the homogeneous mass of "just Natives" which in the eyes of
the average European it appears to be.
A middle class, consisting of teachers, clerks, ministers, and a
few business men, is slowly beginning to enierge. The desire to own
property often keeps them out of the Municipal locations, where
overcrowding is generally prevented and slum yards do not exist.
As conditions in other townships and African urban areas, where
private ownership is permitted, are not so well controlled, they fre-
quently live in the midst of slum conditions, cheek by jowl with new
arrivals in the city, with the humblest of labourers or with the illi-
terate and the dissolute.
Characteristic of most locations is the dreary monotony of row
upon row of identical small houses, each surrounded by its small
plot of ground, and housing indiscriminately families at diverse!
levels of cultural development. Hence, here, too, the social environ-
ment of one is the social environment of all. Although the unsavoury
moral and material conditions of the slum yard are absent, there is
HOME-ENVIRONMENT OF URBAN AFRICANS 89
yet no means of separating those who have acquired more elaborate
standards of life from those still living in a primitive way. Such
differences as exist between Municipal location and private township,
or between locations among themselves, do not, therefore, affect any
particular class, but all classes equally.
In these circumstances, differences in home-environment,
related to the status and education of particular families, tend to be
levelled out by the pressure of the immediate social setting, in which
the home happens to be located. Houses are far too small to make it
possible for the children to remain inside, away from the jetsam and
flotsam of the streets, and the little gardens are usually put to some
productive purpose and are therefore not available for play. Play-
grounds there are none, so that for the smaller children there is only
the street in which to amuse themselves. In Sophiatown, a pre-
dominantly Native township within the Municipal area of Johannes-
burg, where the street is also the repository of garbage and dish-
water, the recreational environment is most unsalutary, only sur-
passed in this respect by conditions in slum yards which, though
rapidly disappearing, are still to be found in parts of the town.
Hellmann' has given a graphic account of life in such a yard. Its domi-
nant features are dirt, squalor, overcrowding, and constant breaches
of the peace. In the New Brighton Location at Port Elizabeth,
indubitably one of the best in the country, the evils of street-playing
have been mitigated by a novel grouping of houses into units of six,
around a small grassed and tree-lined open area. There are other
locations, such as Langa in Cape Town, where the lay-out is spacious
and airy and where the streets therefore lack those backyard features
so common in most townships. School children, which in Johannes-
burg constitute only 37.6 per cent of the estimated number of Afri-
can children of school-going age in the town, are somewhat better
off, as recreational activities in the shape of sport, Pathfinders, Way-
farers, and social clubs, are provided for them. Nevertheless, even
for them, the public precincts of the townships or locations remain
the setting for most of their leisure time. The smaller children play
games in the street, the older ones loaf around street- and shop-corners,
try to earn some money doing odd jobs, or, more frequently, gamble
or act as runners for Chinese gambling-games which have a great vogue
among urban Africans. They rub shoulder with dagga-smokers, get
involved in drinking-parties, and, not infrequently become members
of gangs where they engage in petty annoyances and worse.
It stands to reason that it is well-nigh impossible for individuals
,or groups to keep aloof from these prevailing conditions. Even at
the best of times, and in well-ordered Municipal locations, the school
,or play companions of children from better-class families are drawn
indiscriminately from all types of homes. The absence of residential
segregation according to class orculture makes this inevitable. Hence,
:at all times, each child is liable to be influenced by others lest
-cultured and socialized than himself. This is bound to have a des
pressing effect on the average cultural level, and to nullify to a con-
siderable extent the efforts made in individual families to.interes-
,children in educative activities.
This, then, is the general setting within which such individual
differences in culture as may be able to maintain themselves must
What kinds of home-environment are actually to be found
-within this framework ?
2. MATERIAL CONDITIONS OF URBAN AFRICAN HOMES
From a material point of view, the worst conditions are those
found in the slum yards. We can do no better than quote Hellmann's
description of one such yard.
"Rooiyard consists of 107 rooms and covers an area of five stands
with a total extent of 1,183 square yards. As a result of the large number
of rooms which are built on this confined space, a state of extreme
congestion prevails. The yard is roughly triangular in shape. Fifty-
seven rooms are built on the boundary and face the yard, and 15 rooms,
7 on one side and 8 on the other side of the triangle, face the street.
In the centre of the yard there is a double line of 35 rooms, built back-
to-back and facing the rooms which skirt the yard, thus dividing the
yard into two sections with rooms on either side and alleyways, about
15 feet to 20 feet in width, in the centre. The 15 outer rooms and 14
of the inner rooms are built of brick and have cement floors, the remain-
der of the rooms being ricketty constructions of old corrugated iron
.and thin wooden planks. The brick rooms vary in size from 10 feet
by 11 feet to 11 feet by 12 feet. The partitioning walls, about 10 feet
in height, do not reach the roof, which at its apex is about 15 feet high.
The other rooms vary in size from 8 feet by 11 feet to 11 feef square,
with a height of from 8 to 10 feet. The flooring boards are, in the majo-
rity of these 78 rooms, rotten. Some rooms have no flooring at all and
the bare earth forms the floor. The doors of the rooms are badly fitted
and have no proper locks, being fastened from the outside by a padlock
HOME-ENVIRONMENT OF URBAN AFRICANS 91
and from the inside by a bent nail or rough contraption of wire. Each
room is fitted with two windows, but as one window often gives access
to an adjoining room, it is usually covered with a plate of tin. Cross-
ventilation is not possible in the 64 rooms which are built back-to-back.
In summer the rooms are unbearably hot, and in winter the cold winds
which enter through the gaps and holes in the walls necessitate the
constant burning of large coal-braziers, introducing an element of dan-
ger and rendering the atmosphere in the rooms extremely unhealthy.
Very few of the roofs are rainproof, window-panes are often missing,
and the level of the floor is, in a number of rooms, below the level of
the yard. In wet weather the rain-water flows into the rooms, carrying
with it the debris from the yard, and the discomfort of the occupants
under these miserable conditions requires very little emphasis. The
yard has a narrow entrance. Flanking the entrance inside the yard
stand two cement garbage-bins which serve the whole yard. The occu-
pants are served by 6 latrines, three for men and three for women, but
they are usually in such a bad state of repair and so neglected that the
children shun them, as is amply testified by the condition of the alley-
ways inside the yard and of the pavements surrounding it. There is a
'washing-room' adjoining the lavatories, consisting of 4 corrugated
iron walls with a cement floor and containing two water-taps, one or
the other of which is never in working order. This single tap serves
all the residents of the yard.
"The alleyways in the yard and the pavement on to which face
the outer rooms are cluttered with an important part of the essential
possessions of the Rooiyard Natives. Here stand the motley tins, rang-
ing from one-gallon oil-tins to large petrol-drums, which are used for
the preparation and storage of beer. The cooking-braziers are placed
outside the rooms as the smallness of the rooms, which have to serve
the needs of the whole family, does not permit of cooking operations
being performed inside. Large packing-cases used for firewood occupy
much of the available space outside each room. The repeated requests
of the Health Inspector that the yard be cemented have remained
unheeded, and after rains it is like a quagmire. In dry weather it is
usually littered with an assortment of refuse and debris. Six Sotho
women have each constructed a lapa (courtyard) of clay and cow-dung
in front of their rooms, and these little courtyards form oases of clean-
liness and order in the midst of the general litter of the yard.
"The interiors of the greater number of the rooms present a
striking contrast to the unsavoury disorderliness of the yard. Although
the ceilings are often covered with cobwebs, the floors are well scrubbed
and the belongings of the family tidily arranged. That this cleanliness
is achieved only by the tireless expenditure of energy and labour is
conclusively proved by the constant preoccupation of the Rooiyard.