LaAf Oordt, Joan Frederik van

597nx (1856-1918)
The origin of the Bantu,
A preliminary study. 1997,
 ( \PE OF GOOD HOPE.

THE

ORIGIN OF THE BANTU

PRELIMINARY STUDY

j. F. VAX <

)ORDT, B.A. (C

i
 n.X

CAPK OF GOOD HOPE

COLONIAL SECRETARY'S MINISTERIAL DIVISION.

\

THE

ORIGIN OF THE BANTU

A PRELIMINARY STUDY

BY

].
F. VAN OORDT, B.A. (CAPE).

'il LI I'jtli //c'HXt's- .'/ r,ifluinu-nt hv Cumnunui i>t Hi^ t: \ct'!lfiu\ tin

C'Al'K TOWN:
LAHl'J Tl.MKS M.MITKD, OOVKkX.MKNT I'KIXTKK'S.
[907.
H.l.'io

Aft
V-

JUL 11 1945
 PREFACE.

In submitting this Essay on "The Origin of the Bantu,"

for the purpose of having it printed as a Report to both
Houses of Parliament, I have the honour to make the follow-
ing remarks. This work is the first tangible result of three
years' close and unremitted study of the Bantu question.
During two of these three years, I enjoyed, through the liber-
ality of the Government and the Parliament of this Colony,
a monetary grant,, which if it did not supply all my needs, enabled
rue to carry on my researches and to devote nearly all my
time to them. Without this liberality of Government and
of Parliament my work would have been impossible, and
hence I consider it not only my duty to submit this Essay
to Government, but I do so also in a spirit of sincere thank-
fulness.

It is not for me to say anything about the value of the

work contained in the following pages. I gladly leave this in

the hands of others, in the consciousness that I have tried to

do honest and lona-fide work. No doubt there is much in it
which will require correction in the course of time much will
;

require rounding off. But the material at my disposal was

scant. Beyond the valuable library which Sir George
(in-y left to this Colony, I had little or nothing to consult ;

indeed, to enable me to become acquainted with the discoveries

of the last 30 years, I have been compelled to spend a con-
siderable sum (over 100) in obtaining the latest \\<>rks from
Europe. My ineome scarcely justified such extravagance, but
rather than court disastrous failure, 1 have gladly made Hie
sacrifice.

This essay purports to be a mere Pniiniinarv ^tndv on the

Un^iu of (lie Bantu. Though I flatter myself that I have
louud the keys which will unlock most of the safes containing
the secrets of Bantu philology, and of the-mstitutions, customs,
and religious ideas of that widespread race, yet the threat task
remains to lit these keys, and as safe after safe is opened, to
spread the treasures before the scientific world, as well as
before the general public.
 I am determined to continue the work once
begun, and to
form at least a small
portion of the immense task that still
requires to be done. am
doing the work from a fn-lin* of
I
love towards the
country of my birth, as well as from a desire
serve Science I am ;
prepared to devote: the rest of my life
to the solution of a few of the
many questions which still re-
main unanswered.
But in order to do so, I require assistance, and principally
such assistance as will me
while thus engaged beyond the
put
sordid wants of life, and will also enable me to
procure several
books which I
urgently require.
For these reasons hope that the Parliament of this Colony
I
will enable me to prosecute
my researches, and to continue a
work which is merely
begun, and which to finish will
require the energies and the time of at least a dozen more able
men than myself.
I trust that I have, at all events,
proved to the Govern-
ment and to Parliament that the financial
support they have
thus far so kindly granted me has not been wasted.

J.F. VANOORDT
Cape Town, 28th February, 1907.
 Ill

AUTHORITIES CONSULTED AN 1) MADE USE 1

OF.

i. Skeat and Blagden. The Pagan races of the Malay

Peninsula.
?..
Hodgson. Essays relating to Indian subjects.
3. Bellew. From the Indus to the Tigris.
4. Mockler. Grammar of the Baloochee language.
5. Pierce. Description of the Mekranee-Beloochec Dialect.
6. Seymour, (irammar of the Sindhi language.
7. ('lark. Ao Naga Grammar.
8. McCabe. Outline (irammar of Angami Naga.
9. Soppit. Short account of Kachcha Naga.
10. Witter. Outline Grammar of Lhota Naga.
11. Arden's Telugu Grammar.
32. Mudaliyar. Anglo-Tamil Primer.
13. Knight, Spauldings, and Hutchings. English-Tamil Dic-
tionary.
"~
14. Percival. Telugu Dictionary.
15. Foucaux. Grammaire de la langue tibetaine.
16. Das' Tibetan Dictionary.
17. Jaeschke's Tibetan Dictionary.
1 8. Ramsay. Western Tibet.
19. Henderson. Tibetan Manual.
20. Mainwaring. Lepclia Dictionary. >*-.
21. ,, Lepcha Grammar.
22. Roberts.Anglo-Khassi Dictionary. ^
23. Elmslie. Kashmiri Vocabulary.
24. Wade, (irammar of the Kashmiri language.
25. Vaughan. (irammar and Vocabulary of Pushtu. "
26. Elliot. Memoirs of the Races of the North WeMmi
Provinces of India.
27. Hunter. Imperial Gazetteer of India.
28. Rivers. The Todas.
29. Dirr. Annamitische Spiachlehrc.
30. Wershoven. Siamesische Sprache.
31. Hepburn. English-Japanese Diction, u\
32. Seidel. Japanesische Schrift-Sprache.
Jaj)aiK sische I'mgangssprache.
>

33. ,,

34. Elliot. Finnish (irammar.

35. Wellevvill. Finnische Si)rachlehre.
36. Meiinnann. Dictionaire FranQais-Finnois.
r
37. Donnrr. V'ergleichendes Wortrrbuch <ler l
innish-ugri-
Sprachen.
 IV

38. Weske. Untersuchungen zur vergleichenden Gramma-

tik der Finniseheii
Sprachstammes.
;'i Wiedcmann. Syrjanish-Deutsches Worterbuch.
40. ,, Grammatik der Wotjakisrhen Sprache.
41. drun/el. Fntwurf einer vergleichenden (irammatik
der Altaischen Spraelx 11.

42. Redhouse. Turkish Dictionary.

43. Wied. Turkische Grammatik.
44. Vambery. Catagaische Sprachstudien.
4.5. Ktvmologiscb.es Worterbuch der Turko-
Tartarischen Sprache.
46. Radloff. Phonetik der Nordlichen Tiirksprachen.
47. Castren. Griindziige einer Tungusischen Sprachlehre.
48. ,,Burjatische Sprachlehre.
49. ,, Yorlesungen iiber Finnische Mylhulugie.
50. .,
Ostjakischen Sprachlehre.
51. liotlin^k. Jakutische Sj)rachlehre.
52. Zwick. Handbuch der Westmongplischen Sprache.
53. Harlex. Manuel de la langue Manchoue.
54. Vitale and Senvy. (irainmaire de la langue Mongolr.
35. Ballagi. Worterbuch der I'ngarischen Sj)rache.
5^>. ("n">rg. Grammatik.
I'ligarische
57. l.'jfalvy. Etudes Altai'ques.
58. Lenormant. La langue primitive de la haldce. (

59. Delitzsch. Assyrisches Handworterbuch.

60. Norris. Assyrian Dictionary.
61. Delitzsch. Assyrian Grammar.
62. Sayce. Assyrian (irammar.
d ;.
Rosenberg. Assyrische Sprachh'hre.
64. King. Letters of Hammurabi.
(>> King. Assyrian language.
66. Prince. Stmierian Lexicon.
^7. Berlin. The Cuneiform languages.
68. Lenormant. L'accadien el les langiics touranienncs.
Oppert. Le ])( u})le et la langue des Medo.
>

69.
70. Bleek. Comparative Grammar of the S.A. langnag-
71. Torrend. Com])arative Grammar of the South African
Bantu languages.
72. Brincker. Dcntscher \\"ortfiihrer fiir Otjiherero, O^hin-
donga und Oshikuanjama.
7;;. Brincker. Lehrbuch des Oshikuanjama.
74. Kolbc. A language study based on Bantu.
75. Koelle. I'olyglolta Africana.
7'>. Seidel. (Trammalik der Hauptsprachen D< utsch Siid-
\vest Afrikas.
77. Yocabnlaire adouma-frangais.
Dahin.
78. Zimmermann. irammatical sketch of the Akra langu
(

79. Isenberg. Dictionary of the Amharic language.

80. Mondont-Vidqiilhet. Grammaire de la langue amhariquc.
 81. Christallcr. Dictionary of Asante.
82. Reinisch. Die Barea Sprache.
83. Mitterrutzner. Die Sprache der Bari. *
84. Crisp. Notes towards a Secoana Grammar.
85. Brown. English Secwana Dictionary.
86. Jacottet. Practical Method to learn Sesuto.
*"-
87. Sesuto-English Vocabulary.
88. Whitehead. Grammar and Dictionary of Bohangi.
89. Koelle. Grammar of the Bornu language.
90. Nylander. Grammar and Vocabulary of Bullom.
91. Mitterrutzner. Die Dinka Sprache.
92. Seidel. Leitfaden zur Erlernung der Dualla Sprache.
93. Goldie. English-Erik Dictionary.
94. Guiraudon. Manuel de la langue foule.
95. Reichardt. Vocabulary of the Fulde language.
96. Tutschek. Dictionary of the Galla language.
*
97. Payne. Grebo-English Dictionary.
98. Marre. Die Sprache der Haussa.
99. Robinson and Brooks. Dictionary of the Hausa language.
TOO. Schoen. Grammar of the Hausa language.
4
101. Crowther. Vocabulary of the Ibo language.
102. Sims. Vocabulary of Kibangi.
103. Velten. Die Sprache der Wakami.
*
104. Sims. Vocabulary of Kilolo.
~~"
105. Sims. Vocabulary of Kiteke.
106. Bentley. Grammar of Congo. '-
Dictionary and
107. ,, Grammar and Vocabulary of Luganda.
-
108. Blackledge. Luganda Vocabulary.
109. Hollis. The Masai language and Folklore.
no. Elliot. Dictionary of the Tebele and Shuna languages.
in. Kroenlein. Wortschatz der Khoi-Khoin.
112. Schils. Grammaire complete de la langue des Namas.
113. ,, Dictionaire etymologique de la langue des
Namas.
114. Reinisch. Die Nuba Sprache.
115. Henry. Grammar of the Chinyanja.
116. Scott. Dictionary of the Mang'anja language.
117. Grammar and Vocabulary of Mpongwe.
118. Junod. Grammaire ronga.
119. Smith-Delacour. Shironga vocabulary.
120. Seidel. Handbuch der Shambala Sprache.
121. Larajasse. Somali-English Dictionary.
122. Kirk. Somali Grammar.
123. Reinisch. Die Somali Sprache.
124. Madan. English-Swahili Dictionary.
'

125. Seidel. Grammatik der Suahcli Sprache.

126. Krapf. The Kisuaheli language.
127. Stover. Observations on llic Umbiindu.
128. Sanders. Umbundu Vocabulary. ~**^
I2Q. Knrllc. (iramniai "1 tli-' Yei language-.
130. Dictionaire volof-francais.
131. Boyce. Grammar of the Kafir language.
132. McLaren. Kafir Grammar.
133. Davis. English-Kafir Dictionary.
134. Grout. The Isizulu.
'

135. Roberts. English-Zulu Dictionary.

136. Colenso. Zulu-English Dictionary.
137. Bryant. Zulu-English Dictionary.

Besides these, use has been made of the Vocabularies found

in the works of Barth, Schweinfurth, Stanley, and Johnstone,
as well as of Livingstone. The above list naturally does not
comprise any works on general philology, or general works
on Africa, a large number of which have been perused.
 CAPE OF GOOD HOPE.

THE ORIGIN OF THE BANTU.

Presented to both Houses of Parliament by Command of His Excellency the Governor.

1907.

SECTION I.

When Dr. W. H. J. Bleek died on the lyth of August, 1875,

near Cape Town, the Science of Philology lost one of its greatest
and most ardent students. His loss to the scientific world was
particularly lamentable for more than one reason. Cut off in
the midst of a great work, at an age when most men are still
capable of mental and bodily exertions, he left that great work
'
The Comparative Grammar of the South African Lan-
"
guages unfinished and what made his death most griev-
;

ously felt was the fact that he left behind him nobody who
was either capable or willing to continue the work so pro-
pitiously begun. Bleek was a student in the fullest sense of
the word he was always employed in collecting new facts
;

and data, always busy extending the already wide range of

his knowledge. He never was a teacher ; he never trained
any student he left no pupil or pupils to whom he had im-
;

parted any of the knowledge he had of the particular branch

of Philology upon which he was engaged. The natural result
of all this was, that since Bleek 's death little or nothing has
been done in Bantu Philology, at least not as far as indepen-
dent research is concerned.
It is true, however, that in the 32 years that have elapsed
since his death, we have obtained a far greater knowledge of
the Bantu languages than Bleek ever possessed. With the
opening up of the great centre of Africa there rose up before
our eyes tribe after tribe, speaking dialects of Bantu, of which
Bleek could never have dreamt. A large number of gram-
mars, dictionaries and vocabularies have passed through the
press in that third of a century ;' indeed, the accumulation
of new facts and data has grown to such an extent, that the
most enthusiastic student finds it impossible to master even
the very elementary principles of the more than 200 Bantu-
dialects at present spoken over the greater part of Africa.
As is always the case in a young science, the material, at
present at the disposal of the student of the Bantu-languages,
[G. 171907.] c
is
anything but homogeneous. There is, indeed, a good deal
of solid, useful material, carefully gathered by competent
men ;
there is not a little absolutely useless stuff thrown upon
tin pseudo-scientific market by men who were better ac-
quainted with the than with the science of language.
rifle
Above all, an immense amount of indifferent work
there is

collected by amateurs who meant well, and often did very

fair work, but who were unable to grasp true philological

principles on account of never having had any scientific train-

ing. Hence it is often the painful and laborious task of the
student to separate the good wheat from the chaff. What
often makes the work of the writers practically useless is their
unacquaintance with the elements of phonology. Instead of
following either Lepsius' system or that of the Missions (intro-
duced by the late Professor Max Miiller), several compilers of
vocabularies have followed a system of their own, and, with
unpractised ears, have put the most absurd sounds upon paper.
So we have, inter alia, a vocabulary of the Kamba-language,
in which some of the words end in r, an absolutely impossible

ending for any Bantu word. To this must be added the fact
that very often the spelling varies in writers of different
nationalities. Where, e.g., the English authors write um-
bundu, the French write oumboundou ; such things must not
be lost sight of when we make use of valuable contributions
in French, such as Junod's splendid monographies on the Ba-
Ronga tribe.
It would be
useless to deny that, for a considerable part
of our present knowledge of the various Bantu-languages, we
are to thank the Mission Societies and their workers. Wher-
ever these men have gone, they have rightly considered it
their firstduty to become acquainted with the language of
the people, to whom they were to preach the Gospel. As soon
as they had sufficient knowledge of the particular dialect they
came touch with, they began to translate portions of Scrip-
in
ture vocabularies then followed, afterwards shorter or larger
;

grammars, and, in many instances we find even most valuable

dictionaries, based upon a careful study of the language.
Koelle, Krapf, Steere, Livingstone, Boyce, Arbousset, Grout,
Colenso (not to mention many more) were the first to do good
and sound work. Among the many fine works of later times
from the hands of missionaries we may mention Brincker's
Otyi-Herero Dictionary, as .well as his grammar and dictionary
of Oshikuanjama Scott's Mang'anja Dictionary, Bentley's
;

Dictionary and Grammar of the Kongo language, and the

recently published splendid dictionary of the Zulu language
by the Rev. A. I. Bryant.
In Central Africa valuable work has been done by Sir H. H.
Johnston and his able assistants, who have made us acquainted
with the Masai language, which, though not belonging to the
Bantu group, has had no little influence upon several Bantu
languages. The Germans are also beginning to pay attention
to this Central Group of Bantu, which is of the utmost impor-
tance. Yet there is a large lacuna here, waiting to be filled up,
and good grammars and dictionaries of Luganda, Ki-Hima,
and the dialects of the other tribes surrounding lake Victoria
Nyanza, are urgently wanted.
The vast country constituting the independent Congo State
is practically unexplored from a linguistic point of view, al-

though Sims and others have made a good beginning.

In the West, from the mouth of the Congo to the borders
of Liberia, and in the great Hinterland behind it, fine work
has been done by the French, not only by the Roman Catholic
missionaries, but also by a number of French officers, many
of whom have been specially selected for this purpose by the
French Government.
In Northern Rhodesia Mr. A. C. Madan is at present doing
very fine preliminary work, asproved by is hand- his recent
books on the Senga and Wisa dialects. But much is required
to be done yet in this neighbourhood. A good grammar of
Mashona is required ;
or better, a series of works upon the
Mashona, such as the Rev. H. Junod has written upon the Ba-
Ronga. From Elliot's Dictionary of Tebele and Shuna it is
perfectly clear that the Mashonas speak a most remarkable
and interesting Bantu-dialect.
From the above it is apparent that the student of Bantu has
at present a large amount of material at his disposal, and is thus
enabled to undertake researches and reach conclusions, which
Dr. Bleek could never have dreamt of. Besides this, it should
be borne in mind that, since the days of Bleek, Philology has
made gigantic strides. Not only have we solved the Assyrian
problem to a remarkable completeness, but we have proved,
thanks to Lenormant's valuable labours, that in very ancient
times, about 4500 B.C., there lived at the mouths of the Tigris
and Euphrates a highly-cultivated nation, the Sumerians, who
spoke a language nearly related to .Finnish, that is a language
belonging to the so-called Ugro-Altaic group. This is a fact
which, as I hope to show in "this essay, is of the utmost impor-
tance to the student of the Bantu languages, a fact which will
completely revolutionise our present ideas regarding this par-
ticular province of Philology.
The Ugro-Altaic group itself has been very carefully studied
during the last forty years. Whatever the faults of the Gov-
ernment of Russia may be, want of interest in the study of the
languages of its subjects in Asia has never been one of tin m.
It has always granted ample support to students, and a mere

glance at the reports and the works of the splendid Academy

of Sciences at St. Petersburg is sufficient to convince anybody
of the correctness of this assertion. The best dictionaries of
Mongolian, Mandchu and Turki are in Russian, fastren, a
Finn by birth, has done more to make us acquainted with the
languages related to Finnish than perhaps any other scholar.
The Hungarians, whose language belongs to the same group,
have taken up the question in real earnest, and Yambery and
Hunfalvy have done most valuable work. The Germans and
Austrians have followed their lead, and at present we have a
considerable number of works in German upon the subject.
A small work by Dr. J. Grunzel, of Vienna, entitled " Entwurf :

einer vergleichenden Grammatik der altaischen Sprachcn

(1895), is perhaps one of the most concisely and clearly written
books upon a vast subject, and personally I am much beholden
to it, as through it my eyes were opened, and it made me see
the Bantu in a clear and unmistakeable light.
Bleek was deprived of these opportunities. Indeed, he
really never devoted himself to the question of the Origin of
the Bantu languages. His main work was to compare a num-
ber of Bantu dialects with each other he moved absolutely
;

within the Bantu circle. It is true that he seems to have had

even as his great friend and protector, our riever-to-be-forgotten
Governor, Sir George Grey, an idea that there was a strong
link between the languages of Africa and those of Australia,
New Zealand and Polynesia. That relation does exist, but not
in the sense usually applied to it. Bantu and Australian are
but the different radii from a common Proto-Turanian centre ;

hence their relationship. The fact has, however, given rise to

a good deal of misapprehension, and this was not a little
strengthened by the theory of the late Dr. Sclater, who placed
a now submerged continent, Lemuria, between Africa and
Australia. The Revd. A. I. Bryant has unfortunately been led
astray by this theory, and in the introduction to his Dictionary
of the Zulu language has ventured to assert that the Bantu
moved from Lemuria into Africa. This theory must be con-
sidered as untenable in the face of the fact that the greater
eastern half of Madagascar is inhabited by a nation which
speaks a pure Dravidian dialect, and which, in my humble
opinion, has as its fatherland the South-East coast of India.
Although cm the west coast of Madagascar a Bantu dialect is
spoken, it is fairly certain that the tribes which speak it were
originally either immigrants from Africa or, what I think more
likely, were placed there against their will by the earlier
Dravidian inhabitants.*
We have to look outside of Africa for the origin of the Bantu
race. In trying to solve the problem we are necessarily
thrown back upon the resources of Comparative Philology and,
* I do not consider
myself qualified to pass an opinion upon the Lemurian
theory, though I believe that there exists a good deal of zoological and
botanical evidence in favour of it. But if Lemuria ever did exist, I am con-
vinced that it became submerged before the human race entered upon its
Western migrations.
to a small extent, upon the somewhat dangerous principles of
Comparative Religion. All other resources are closed to us.
The Bantu knew no other art than the making of pottery and
the plaiting of mats ;
several of the tribes, but not all, were
acquainted with the art of melting iron ore and hammering
the hot iron into weapons or implements. It may be that
some of the tribes which arrived in Africa, at a later period,
were acquainted with the art of weaving a rough cloth, but the
evidence upon this point is not quite convincing. The Bantu
have left no buildings behind them, for the huts they lived in
were but frail fabrics. The antiquary will thus find little scope
for his science in Africa, south of Egypt and Libya. Still less
does the historian. Everything before 1500 A.D. is practically
an absolute blank as far as the history of the Bantu is concerned,
and the few data we have in Arabic writers about the Zanj do
not throw much light upon their history. It may be, as Dr.
G. M. Theal surmises, that in ancient Arabic unpublished
writings, we may find something more definite about the
Bantu along or near the East Coast of Africa, between Cape
Guardafui and Cape Delgado. But if we do, it should be re-
membered that the ancient Arabic authors were not a little
given to exaggeration, and were not altogether unacquainted
"
with the art of prevarication," and that hence any informa-
tion given by them has to be received with the greatest cau-
tion.

Philology must, therefore, be our guide, and in a case like

the present it is perhaps our surest guide under the circum-
stances. There have been conditions under which a race has
lost its original language, and has adopted the language of
another race as its own. So the French, originally a Teutonic
race, have adopted a Romance language in South Africa,
;

the Berg-Damaras, a pure Bantu race, speak Hottentot, the

language of their conquerors. On the West Coast of Central
Africa, there are tribeswhich were probably originally pure
Bantu, but which now speak a Negro dialect, and the con-
verse case has been found too. In Fulde and in Haussa we
have two languages made up of all kinds of elements Bantu,
Hamitic, Negro, Berber and Arabian. Yet there can be no
doubt but that originally both the Fulde and the Haussa
tribes are of true Bantu origin, though in ancient times they
may have borne another name, and though a large percentage
of the blood now flowing through their veins is of non-Bantu
origin.
But, in general, putting extraordinary circumstances aside,
a race maintains its language as one of its racial characteristics,
and even political union does not always interfere with this
general rule, as, for instance, is shown in Switzerland. On
the other hand, it must be clearly remembered that in a race
which has been subject to migrations, the language is apt to
undergo various changes. Such changes are partly due to
climatic influences, partly to change of customs, partly to the
fact that, even if the race keeps its blood pure, the language
will absorb words from other languages, so-called loan-words.

In Africa the Bantu became not the slaves of the indigenous

races, but their conquerors, and that evidently from an early
date.* But, notwithstanding that fact, the Bantu lan-
guages did absorb some part of the vocabulary of the languages
of other races. In West Bantu we can clearly trace the
influence of the true Negro languages in Central Bantu that
;

of the Nilotic and Masai languages in some southern Bantu

;

dialects we meet the clicks of the Hottentot, who though not

indigenous to Africa, evidently reached that continent long
before the Bantu did. In East Bantu, along the coast, there
is a strong Semitic element, principally due to Arabian influ-
ence in Mashona there is another Semitic element, which is.
;

however, of far older origin, and must, without the slightest

doubt, be referred to Assyrian.
Yet, broadly speaking, the Bantu languages have remained
remarkably pure under the circumstances, and in more than
70 per cent, of the words, we have no trouble in tracing them
to their original roots.
It is quite different with Bantu Phonology. Here climatic
influences and certain customs have created some of the
most remarkable changes. The tribes who live on highlands
have a completely different pronunciation from those who live
in the lowlands or along the coast. In some Bantu languages
we still find the old original uvular r or gh, e.g., in Kafir ;

while in others the gh has been lost and the more modern r
sound substituted, a fact which may be due to the influence of
original elements, but was probably not altogether uncon-
nected with climatic influences.
In some tribes, who (or whose ancestors) were in the habit
of knocking out their front teeth, we find labialisation of the
sibilantic sounds, i.e., a tendency to lisp. Thus we in haw
Zulu hi for s, and thl for an original ts or g. The custom of
some tribes of boring holes in their upper lip and inserting a
ring (the pelele) must have considerably interfered with the
pronunciation of certain sounds, and made it impossible for
them to sound either p, b, or m in the ordinary manner. Hence
we find these letters replaced by other, often very compli-
cated, sounds.

*
Though I believe such to have been the general fact, I should wish to
state here, that it is evident to me, that at some period or other, a section of
the Bantu, including inter alia the ancestors of the South Eastern Bantu
(Zulu, Kosa, Amabomba, etc.) was subjected by the great Masai tribe, and
that this fact has not been without its influence upon the language of these
Bantu tribes.
 This question of Bantu Phonology cannot be treated in
this essay, as the subject is too vastand far-reaching. I have
only referred to it in order to show its importance, and to
illustrate in how far the pronunciation, and hence the spelling
of Bantu in its various dialects is affected.
The questions to be answered in this essay are :

1. WHAT IS BANTU ?

2. WHENCE DID THE BANTU RACE ORIGINATE ?

It may seem very audacious on the part of the writer to

try and give the answer to these two questions, which are of
such immense importance not only to South Africa and to
Africa generally, but also to the Science of Philology. Yet
the writer's researches and studies during an unbroken period
of three years have gradually convinced him of the general
correctness of his answers, and for that reason he considers it
his duty to place the results of his work before the scientific
public as well as before general readers.
The answers to the above questions may be conveniently
gathered in the following theses :

I. The Bantu language belongs to that group of languages,

generally known as the Ugro- Altaic.
II. The fact that in the Bantu language there are two distinct
groups of words, one of which is far more archaic than
the other, entitles us to come to the conclusion that
there have been TWO Bantu invasions of Africa.
III. The first Bantu invasion of Africa commenced from
some part in or near Hindostan, and the language of
these first invaders is directly connected with the Non-
Aryan languages of India.
IV. The second Bantu invasion of Africa started from the
mouths of the Tigris and Euphrates, and probably
took place about the year 680 B.C.
V. The real original home of the Bantu race is the Peninsula
of Malacca, and the Pagan races at present found in
that Peninsula are ethnographically and linguistically
very nearly related to the present Bantu races of Africa.
And as a corollary I beg to add :

VL The Ugro- Altaic group of languages as well as the Bantu,

which forms part of that group, have arisen from a
mixture of Hamitic and Turanian elements.
It is now my dutyto 'adduce the proofs in support of the
above assertions. In doing so, I wish to remind my readers
that this essay is merely a preliminary study, and that hence
I must be as succinct and short as
possible. The more volu-
minous proofs and the more elaborate arguments in support
of my theory must stand over until I am enabled to put my
larger work upon the subject through the press. I am only
 8

at the very threshold of my labours, so to speak, and for that

reason I have limited myself in this essay to what may be
railed the lexicological part of the Bantu languages. I have

not touched upon grammatical formations, partly because

my researches on this matter have scarcely begun, partly
because otherwise this essay would become too bulky. I
have only very slightly touched upon the important question
of the Bantu The
object of this essay is, as said
prefixes.
before, merely to make
the philological world in particular,
and the reading public in general, acquainted with a dis-
covery, which the writer believes to be of considerable interest
and scientific value.

SECTION II.
"
Dr. Joseph Grunzel gives on pp. 4-8 of his Entwurf einer
"
vergleichenden Grammatik der Altaischen Sprachen (Wil-
helm Friedrich, Vienna, 1895) a review of the different lan-
guages which belong to the so-called Ugro- Altaic group,
which he divides as follows :

r (a) Samojedian.
I. Urahan or Ugnan Group b ) Finnish.
j j

f (a) Turki.
, , . .
^ ! (b) Mongolian.
II. Altaic Group . . . .
< / / T
)
'
(c) Tungusian.
(d) Japanese.
Each of the sub-divisions of the two groups is again split
up into various languages and dialects. Thus Lappish, :

Esthonian, Permian, Votish, as well as Finnish itself, belong to

the Finnish group, with several others while Uigur ish, Kir- ;

gish, Osmanli (Turkish) and Tataric, belong with many others

to the Turki division. The Mongolian sub-division contains,
inter alia, the East and West Mongolian, the Kalmuk, and
the Buretian ; Manchu belongs to the Tungusian group, while
Japanese forms a group in itself, but has various dialects,
slightly differing from each other.*
Grunzel has not included Accadian nor Sumerian in his
division, but Lenormant correctly classes these with the
Uralian or Ugrian Group, and in general they seem to contain
elements of both Samojedian and Finnish, and to be probably
an older branch. But Accadian, or rather Sumerian (for I
consider the two languages to have been originally distinct
from each other, and prefer using the latter name alone, for
the purposes of this Essay) had an immense advantage over
the other languages of the Uralian group, in the fact that it
was reduced to a system of writing, as early as 4500 B.C., and
*
Japanese' is generally divided into spoker, and written Japanese. These
differconsiderably from each other, due to the influence of Chinese upon the
spoken language.
formed what is called the Cuneiform Script. This system of
writing was afterwards adopted by the Semitic races which
settled in the plains between the Tigris and the Euphrates.
With that Sumerian language we are now fairly well acquainted,
though there are still a number of words or so-called ideograms
which puzzle the best Assyriologists. Still, we know quite
enough of the language and even of its grammar, to enable us
to use it for comparative purposes with Bantu.
The great distinctive philological phenomena in connection
with the Ugro-Altaic languages are the following :

A. All syllables are open and end in a vowel. Occasion-

ally a final syllable ends with n or ng. This is known as
nunation. In some Ugro-Altaic languages, e.g. in Turkish,
this rule would seem not to apply, but this exception is only
apparent and not real, and the final vowel has really been
dropped.
B. There exists a tendency to drop certain consonants be-
tween the vowels, which often results in so-called contraction.
For instance :

(1) Ugrian Okul = son.

Turkish o-ul = ,,

Kirgish ill = ,,

(2) Mongolian tak (for taku) = mountain.

Kirgish ta-u
Altaic ta = ,,

(3) Mongolian negu = to stray.

Kalmuck nou = ,,

Buretian nu = ,,

(4) Uigur ket (for ketu) = to dress.

Altaic ke-i =
Kaschgar ki = ,,

(5)1 Mongolian Kube-gun = child.

Tungusian kuna = ,,

Japanese kf> = ,,

(This latter example is a very clear instance of a mono-

syllabicword arising by pure dropping of consonants and con-
traction from a polysyllabic word.)
C. There is a remarkable change of consonants in several
instances. For example
k = ng = n

= z

= v = m
 10

D. To express adjectives (and, in a few cases, even substan-

use is often made of a repetition of syllables.
tives] For example :

Kirgish dzibdan-dzibdan = quick.

Altaic akkir-akkir = slow.
Manchu gilta-gilta
= shining.
Manchu bura-bara = dark.
Osnuinli javas-jav.i> = slow.
Mongolian tandur-mandur = fright.
and many more, as given by Grunzel I.e. :

E. The Ugro-Altaic languages are generally agglutinative in

character, but the system of agglutination often differs in the
various languages.
F. The Ugro-Altaic languages have no real degrees for com-
parison of adjectives. V
G. The cases are formed in Ugro-Altaic languages by affixes.
H. The Ugro-Altaic languages have a special negative con-
jugation of the verb, a form unknown in any other group of
languages.
These are some of the distinctive peculiarities of the Ugro-
Altaic languages, and if we carefully examine the Bantu lan-
guages, we shall find each of these characteristics back in the
latter linguistic group.
A. In Bantu open and end in a vowel.
all syllables are

We often find that a syllable apparently ends in ;;: or n,

the m being in most instances the representative of an original
n. In reality, however, the n or m belongs to the consonant
following it, and represents the nasalisation thereof, a feature
peculiar to the Bantu group, and of pure Turanian origin.
This nasalisation is not found in all Bantu languages in an
equal degree, nor is it easy to find any particular rule for this
characteristic.
Thus we have :

Kafir
Suaheli
Kafir
Kafir
Mpongwe
Suaheli
Kongo
 II

A remarkable instance of nasalisation is found in Herero

oka-kambe horse, which in Mang'anja is kavalo and in
di.

Mpongwe kavala, and is derived from the Portuguese cavallo =

horse.
B. There exists in Bantu languages a tendency to drop certain
consonants between the vowels, which often results in so-
called contraction.
Thus :

Kafir kulu=big becomes Suaheli kuu = big.

Zulu twala= to carry ,, ,, twaa= to carry.
Kaffer (isi)fuba
= chest (ki)/w= chest.
Mang'anja kwera = to climb ,, ,, kwea=to climb.
Mang'anja fika=to come ,, Mpongwe bia=tocome.
Suaheli pika = to cook ,, Kongo via = to cook.
Herero muna=to discover ,, Mpongwe miato discover.
Kafir (in) dlovu
= elephant ,, Secwana tlou = elephant.
Suaheli killa=every ,, Kamba kia= every.
Herero &>a=fear ,, Mpongwe fo'=fear.
This list could be increased twentyfold without any trouble,
but the above is sufficient to prove the existence of the prin-
ciple, which is certainly of no little importance in Bantu
philology.
C. The changes of the consonants in Bantu agree in the
main with those of Ugro- Altaic, but the former has besides
several other interchanges, which were common in the lan-
guages from which both Bantu and Ugro- Altaic have arisen,
but which appear to have been lost (except in a very few
instances) in the latter group.
The principal one of these interchanges is k=p = b=f=d,
an interchange which had its origin in ancient Turanian where
k constantly interchanges with p. 5 is the softer form of p,
and in its turn interchanges with d, while / originated from an
aspirated p=ph. These interchanges can be traced between
Sumerian and Assyrian, and generally between the Turanian
and Semitic languages.
In the Bantu languages we find them very regularly, as the
following few examples out of many, will show.
Mpongwe
'
!
7ee = also. \'\ ><
Kongo mpe = a\so.
Zulu (o)koko= ancestor. Suaheli babu = ancestor.
Zulu (um) bo nda = assembly. Mpongwe nkanda = assembly .

Zulu buta = to assemble. Kongo kutana = to assemble

Kafir kwa=at. Mang'anja pa=a.t.
Zulu kude = awa.y. Herero = away.
/>o
Kafir (u) donga =riverbank. Suaheli =
/wgw riverbank.
Kamba kianda=riverba.nk.
Zulu &w/=great. Mpongwe (m)polu= great.
Zulu bopa = to bind. Mpongwe kora = to bind.
Suaheli funga = to bind.
Mang'anja pa = by. Secwana &a = by
 1 2

Another change, which is not uncommon in the Dravidian

languages, viz., that between r and t is sometimes found in
Bantu. I have not been able to find many instances of this
remarkable interchange in Bantu, but met with some notice-
able instances in the Si-ronga, which has several old forms
not found in any other Bantu languages. So :

Si-Ronga rumela = to send is Shuna tumira to send.

,, pseru = our ,, Zulu etu = our.
randa = to love ,, Zulu tanda = to love.
tatana = father Secwana rara = father.
Herero tate = father.

Si-ronga mafura = fat ,, Zulu (a) mafuta = fat.

There are some minor interchanges between letters in
Bantu, which cannot be treated here, but to which attention
will be drawn when they are met with in this essay.
In general the interchanges which take place in Ugro-Altaic
find their counterparts in Bantu in exactly the same manner.
(a) k=ch=g=ng=n (m).
Examples.
Kafir kumbi= locust
(in) is Kamba n%ie=\ocust.
Mang'anj a chambu= medicine Suaheli
(u}ganga= medicine.
,,

Herero (otyi) huka= morning

,, Mang'anj a mawa= morning.
Mang'anja kamwa=mouth ,, Kamba moomo = mouth.
Kafir (in) kaba= navel ,, Herero o)ngna= navel.
And many more examples could be adduced
(b) ch = h*
Examples.
Suaheli chunga = to nourish is Herero hunga = to nourish.
Suaheli chuna = to peel ,, Secwana huna=to peel.
Suaheli changa=sa.nd ,, Herero (e)heke=sand.
(c) g=/.
Examples.
Suaheli gon/z#a=sick, is Mpongwe /aga=sick.
Mang'anja ganda = to slay ,, jona = to slay.
Kamba guy a = to steal jufa to steal.
Kamba gunwa = to drink ,, jonga=to drink.
,,

N.B. In Mpongwe the / is really the same as a y, and

simply another form of writing the same sound, just as we
write in English year and young, but in German jahr and jung.
In Mpongwe, however, the / is very often used as an initial
letter, whilst in most other Bantu languages it is, in such
position, either changed or hardened into g or nasalised into ny.
*In Suaheli ch is at present a sharp sibilantic sound like ch in English church.
Originally, however, this ch represents (except in a few instances) an old ki:.
and this accounts really for the weakening of the ch into h. Such weakening
from kh to h is very common in the Semitic languages, and kh in the other
Semitic languages, for instance, is generally rendered in Assyrian by h. (See
Sayce's Assyrian Grammar (1904), p. 62.) Suaheli has been strongly influ-
enced by Semitic.
 (d) / = r.

This is one of the most common changes in Bantu.

Mang'anja ny/#=hunger is Herero (o)ndyara= hunger.

Kafir (um) o&/t'= leader ,, (omu)hongere= leader.
Kongo tala = to look ,, ,, tar=tolook.
N.B. In Herero the r
(i.e., the ordinary English r) is more

frequent than in most of the other Bantu languages.

(e) t d. This is also quite common.

Kongo diato eat is Kamba tiye

= to eat.
Herero (otyi)tyitua=fa.ct ,, Kongo diambu = ia.ct.
Herero (un)^y#=to hope ,, Suaheli taka=io hope,
etc., etc.

There are a very large number of words in Bantu which

show these interchanges, but the following examples must
suffice :

Suaheli = lung. Mang'anja

Mpongwe (i)bobo= ,,

Kafir (u)mongo= marrow. Suaheli (u) bongo

= marrow.
Mang'anj a panga = to make. Secwana bopa=to make.
Suaheli fanya =
Kongo vanga
Suaheli />y=new. Mpongwe miala=new.
Zulu (isi)balo number. Secwana palo=numbcr.
Secwana bona= to obtain. Herero muna = to obtain.
Suaheli
Herero po'=or. Suaheli (a)w=or.
Kongo (o}vo= ,,

(g) *

Zulu ala to oppose. Suaheli zuia = to oppose.

Kongo zonzeka = to prepare. Herero rongera = to prepare.
Kafir (isi)zaiu
= proof. Herero (otyi)raise= proof.
Zulu Kongo /w&=tosink.
D. In Bantu the repetition of syllables and words to express
adjectives (and sometimes substantives] is a very common occur-
rence. For instance :

Mpongwe ogaza-gaza = active.

Herero kanda-kanda = anxious.
Kafir kohlakele = bad.
Zulu (isi)dika-dika = a corpse.
Secwana (\e)here-here
= cunning.
Herero ti-ti = few.
Kafir (\}holo-holo = hollow.
Mang'anja (c\\\)tunzi-tunzi an mage.
Kafir (u) kwe-kwc the itch.
Mang'anja nya-nya
 '4

Zulu (\)niki-niki
-
= a rag.
Zulu ratya-ratya = rough.
Kafir tsafa-tsafa = ,,

E. The general character

of Bantu is agglutinative, but this
varies in the different languages.
This fact is so well known that it does not require any proof
on my part.
I The Bantu languages have no degrees of comparison for
.

adjectives. Reference must be made to some other object, or

the matter must be expressed by using a verb implying super-
"
iority. Thus in Kongo the sentence, that one is the better
"
of those two chairs," is expressed by saying The chairs, :

"
these two, that one has excelled in goodness," and to say, he
" "
isthe tallest of all," the expression, he surpasses all in height
"
must be used. So in Herero I am greater than he," is Ami
:

"
omunene PU ye, which is literally I am great next to him,"
:

"
while "he is tallest of all is eye omunene tyinene komeho ya
"
avehe, literally he is great very above all."
:

G. In Bantu there are practically at present no cases, except

the Genitive, which is formed by a preposition or an infix.
There is, however, strong evidence of the existence of an old
affix na, by which a number of cases were formed. In Finnish,
where we find at present 15 cases, this na is still found in the
Essive case, but it is more than likely that Dr. M. Weske is
correct in maintaining that in ancient Ugro-Altaic most of
the cases were formed by affixing na or some other short
affix beginning with n*
In Kafir this na, in the form of ni (with occasional change
of the preceding vowel) still forms the locative, just as it still
"
does in some expressions in Finnish. So esityeni means in
"
the dish from the noun isitya=dish, and I am inclined to
think that in such adverbial expressions as kona (here, or there)
and kunye = together, we also find this ending, the latter repre-
senting the so-called Comitative case (=with) in Finnish.
The subject is too intricate to be treated here in full, but
certainly deserves the attention of Bantu scholars, as it throws
a good deal of light upon Bantu philology.
H. Bantu, like most Ugro-Altaic languages, has a special
negative conjugation of the verb, which may be considered
one of the most remarkable characteristics of this group.
In Finnish we have the form en syo I do not eat =
Turkish ;

has a negative form of the verb by infixing ma between the

stem and the affixes. So in Bantu we find a negative form of
the verb, which is constituted either by a prefix or an infix.
"
*See the interesting study of Dr. Michael Weskc, entitled Untersnchungen
"
zur vergleichenden Grammatik der Finnischen Sprachstammes (Leipzig,
" "
1873). Also C. W. E. Elliot : A Finnish Grammar (Clarendon Press,
1890}, p. 23.
 15

For instance :

Kafir ngi = la.m. angi = I am not.

Herero okusuta = to pay . okuhasuta = not to pay.
Suaheli si piga=\ have beaten, si ku piga =1 have not beaten.
Kongo evanga =they make. kevanga ko = they do not make.
Altogether there are very many points of similarity between
the construction of verbs in Ugro- Altaic and that in Bantu,
but these cannot be discussed here, without entering more
deeply into the subject than can be done in this sketch. It
may be mentioned, however, that in most of the Altaic lan-
guages, just as in Bantu, the root of the verb is usually found
in the second person singular of the Imperative.

SECTION III.

If the above remarks show a remarkable similarity between

the morphology of Bantu and that of the Ugro-Altaic languages,
the lexicological or dictionary part, shows still stronger affini-
ties. In most cases we find in the Ugro-Altaic languages the
same roots as in Bantu, sometimes in exactly the same form,
sometimes changed in accordance with the known rules of
phonology.
I purpose giving here a short comparative list of some of the

Ugro-Altaic languages and of some of the Bantu languages. /

As regards the Ugro-Altaic languages I am compelled a^
present to limit myself principally to \FiwwishJ (and its immedP Q <

ately related dialects), Turkish, Hungarian and Sumerian, and

Japanese, because, unfortunately, a really good Mongolian or
Manchu Dictionary is not at my disposal. In the present
instance I have limited myself to 100 words, taken at random
from a far larger list in my possession, but the similarity of the
root words are so strong, that even the merest tyro in philology
must be convinced that the two language groups are intimately
related. During the course of this essay several more ex-
amples will be given.
Ugro-Altaic. Bantu.
1. Magyar so-mo = the eye. Zulu (ili)so
= the eye.*
Ostiak sem = ,, Mang'anja (di)so
= ,,

2. Ugrian sul-do = cheap. Kafir (si)-sulo= cheap.

Esthonian halv= ,, Suaheli hafifu=
Japanese (ya)-sui= ,,

(s and h interchange in many dialects of Ugro-Altaic as well

as of Bantu.)
*For convenience's sake I have placed the Bantu prefixes between ( ), so
as to show more clearly the root-words. In Ugro-Altaic the last syllable or
letter is generally an affix.
 ;. sthonian sanaka =stat'f. Zulu (um) saca=staff.
Jakutish >awa N-cwana .. tsamma =
(w = m in most languages).

4. Finnish sa pas = boot. Mang'anja (n)sapato = shoe.

Syrjenian sapoy= ,, Iv-n^. (n)sampatu = boot.

5. Finnishsapo = dress. Secwana seaparo = dress.

(Compere Babylonian subatu= garment, which is evidently
derived from a Sumerian root.)
6. Finnish happo=quick. Herero haka-hana=quick.
Magyar sab = Mang'anja sadiza =to hurry
(b = d)
7. Lappish oive = brain. Herero (omu)-uvi=brain.
Finnish aivu =
Mordwinian ui =

8. Lappish naggi = to fasten. Mang'anja (ze)nga to bind.

Mag. nugas = to fetter. Suaheli na = to bind.
Sumerian nigin Kongo (ka)nga = to bind.
= to surround.
9. Finnish nalka= hunger. Mang'anja njala=hunger.
Esthonian nalga= ,, Herero (o)ndyara= ,.
Livonian nalga = Mpongwe njana =

10. Finnish keikka = round. Kafir (isi)giki= round.

,, kikkura=ring.
Japanese kyoku=crooked

11. Permian gogar = round. Mang'anja gango = ring.

=
Sirjenian gogar ring. Herero (o)goho = ring.

12. Finnish kona=stiff. Zulu (lu)kuni= stiff.

kena= ,, Mpongwe keta= .,
/

13. Finnish jauho=meal. Kafir (um)gobo=meal.

Livonian jouv = (g=j and b = v.)
Sumerian ku = Luganda (en)gano = meal.

14. Finnish tiid = to make. Herero tyita = to make.

Esthonian to = to work. Secwana diha =
Mordwinian Ronga (yen)tya= ,,

tyan = to make. Suaheli tenda =

Japanese tateru=to make

15. Finnish tata= father. Mang'anja tata = father

Magyar tata= .. Herero tate=
Japanese tete-oya= Kongo tata= ,,

Kongo tata-na =
16. Mongolian axala= leader. Mpongwe oga = a king.
Sumerian aga= leader. Luganda agalala = to sit in
master, lord. [state.
Secwana (mog)ogi= chief,
[leader.
(The root of this word is probably the Mongolian a\a elder
brother, which is found back in Suaheli (nd)ugu= elder
brother, and in Secwana (nn)ake= elder brother. This very
interesting word will be referred to again in a later part of
this essay.)

17. Japanese suke = to help. Zulu sekela = to help.

'

Isubu kalana=
(Isubu has clearly lost the verbal prefix.)
18. Mongolian bum-k = dark. Mang'anja dera=dark (b=d)
i

Tungusian buru = to darken Mpongwe (m)pira = dark.

[

Kongo bubu= darkness.

19. Altaic su=tolove. ,

'tierero suvera = to love.
Turkish sev-mek = to love. Kongo zola= ,,

Japanese suki = to love. i)su

= care, attention.
Zulu
Sumerian su= heart.
(The Herero huura = to love, is really from the same root,
but with the interchangeable h form, which is found in Finnish
hywdna = to love (= Herero suvera) and in Finnish huoli = care.)
20. Mongolian
Buretian
Japanese
Sumerian
 [8

25. Mongol. kundu = to adore. Kongo kunda = to adore.

Tungusian kundu = Oshindonga kunda = to salute.
adoration. Herero kumba = to pray.
Finnish kunnioittau
[=to adore.

26. Mongol kitu=to cut. Suaheli kata = to cut.

Sumerian xas to cut. Kongo kuto = sheath of a knife
=
Japanese kata-na = knife. Zulu kwata=to cut all round.
Tungusian koto = knife. Luganda (a)kambe=a knife.
Buretian kitogo = knife. Kikuye kutinnia = to cut.
Finnish (lei)kata = to cut. Bobangi kete = to cut.

27. Sumerian xur = to bind. Mpongwe kora = to bind.

Lappish kar-et=to bind. Secwana golega =
Sirjenian kor-to = to bind. (k=g; r=l).
Buretian kulenam = Aduma (e)kota =
[to bind.

28. Sumerian xir = cry. Kamba gure = to cry.

Finnish kir-jun = Zulu kala= to scream
Herero kua=
Oshindonga kuga =
\

29. Sumerian kul=seed. Kongo (n)gulu

= to sow.
Livonian kull=tosow. Herero kuna = to sow.
Bobangi (mia)kela = to sow.

30. Sumerian kumk = finger- Suaheli (u)kucha= fingernail.

[nail. Kamba (n)gwa= ,,

Ostiak kunic= Haussa kumba =

Finnish kynsi= ,,

31. Sumerian aka = tomake, Zulu aka = to make, build.

[build. Secwana aga=
Japanese ku= labour. Suaheliaka=
Kamba (ku)aka=
Luganda kola = to work.
Isubu bola = to. make.
(The Isubu word is a very clear instance of the interchange
between k and b in Bantu.)

32. Sumerian ka-ga = to name, Kafir 'i)gama=to name.

to call. Secwana kua = tocall.
Finnish kir-kua = to call. Mpongwe kamba = to speak.
Kongo kamba = to speak.

(The real root of these expressions is rt=mouth.)

33- Sumerian ka=door Zulu (isi) caba = door.
kado= * Kafir
Japanese (old) (u)cango= ,,

Turkish ka-pi= ,; Mang'anja komo=

34. Sumerian kin-dea = a Herero (omu)kinduia = a

messenger. messenger.

35. Sumerian xan(hi)=afish. Herero (e)hundyu=a fish.

36. Sumerian gal=great. Zulu kulu= great.

Kafir kulu =
Manganja kulu= ,,

Secwana golo= ,,

Mpongwe (m)polo=
(p
= k)

37. Sumerian gim = to produce Zulu ki-pa = to produce.

Kongo yima= (g
= y).

38. Sumerian lu = sheep. Kamba (i)londa= sheep.

Zulu (uma)lusi= shepherd.
Bongo romba= sheep (r=l).

39. Sumerian zun=s abundant. Kafir zonke = abundant.

40. Sumerian kuma = to be Zulu pumula = to rest.

[quiet. Mang'anj a puma = to rest (p

=
[k).
41. Sumerian xi = fish. Herero (o)hi=fish.
Kamba (i)kuya = ,,

Oshindonga (o)xi= ,,

Wisa (in)swi= (ch

[=sw)
Oshikuanjama (o) shi=fish(ch
[=sh)
42. Sumerian sit = to protect. Zulu sit-elesa = to protect.

43. Sumerian an-kal = to look Mang'anja an'gana = to look.

44. Sumerian subbu = to flee. Kafir (uku) saba = toflee.

45. Sumerian tag = to ask. Suaheli taka=to ask (k=g).

46. Sumerian tila = tolive. Herero tura = to live.

47. Sumerian paga= hunger. Zulu (i) pangu= hunger.

*The word kado by not used in Japanese, but it appears in the
itself is
expression kado-guchi = a door-way. The root of this word is evidently
identical with ha mouth, opening.
 20

.;8. Sumerian gal = to be. Mang'anja kala= to be (k=g).

Kongo kala = to hi-.

49. Sumerian kim=as. Miaheli kama=as.

50. Sumerian kuku = to dress. Zulu g(qoka)= to dress.

51. Sumerian mul = abyss. Secwana moleto = abyss.

52. Sumerian mud-da = blood. Secwana madi blood.

53. Sumerian be = blood. Herero (om)bundu = blood.

Zulu (ubu)bende= blood.

54. Sumerian kal-an= strong. Mpongwe (n)gulu= strength.

kal-ga = strength. Kongo (n)gola
=
Mang'anja kal-am = strong.

55. Sumerian ba = to give, to Zulu and Kafir pa = to give.

[bestow. Secwana (a)ba = ..

Herero & Mpongwe

pa = ,. .,

Suaheli po= ,, ,,

Luganda wa= ,, (b = w).

i

56. Sumerian dur = to dwell. Herero tura = to dwell.

57. Sumerian kigar = dwelling. Kaffir (i)kaya = dwelling (g =

Or).
58. Sumerian ux=flea. Mang'anja ukuku=flea.

59. Sumerian ma=to call. Zulu memesa = to call.

(in form really a causative).

60. Sumerian sa = tocall. Mang'anja (e)sa = to call.

Finnish sa-noa = ,, Herero (i)sana =
Suaheli (i)ta= ,, (s
= t).

61. Sumerian pa=nose. Mang'anja (m)pumo=nose.

Sumerian ka = , ,
Suaheli kua = . .

Secwana (n)ko =
(Here we have a clear interchange of p and k not only in
Bantu but also in Sumerian, which is of great importance.)

62. Sumerian (Gi)zug = reed. Mang'anja (d)zogo = reed.

63. Sumerian xabir=locust. Zulu (in) kambi= locust.

64. Sumerian gu= to destroy. Zulu (i)ngozi=dnru

ID (o)kuta= ,. death.
 21

65. Sumerian a-kal= current. Zulu (isi) kukulu = strong cur-

[rent*
66. Sumerian bar=brother. Manganja (m) bale = brother.

67. Sumerian bar = side. Manganja (m)bali=side.

(the same word as 66.)

68. Sumerian sa-gal = food. Sec wan a se-ya=food (g=y).

69. Sumerian mul=on. Secwana mo=on.

70. Sumerian im=self. Herero im=self.

71. Finnish ja = to cease. Suaheli (i)sha

= to cease.
Kafir-Zulu yeka =
(N.B. In the first case the /, originally a sibilant, was
changed into its corresponding sibilant sh ; in the second case,
as very often happened, it lost its sibilant sound and became y.)

72. Finnish laka = to cease. Mang'anja leka = to cease.

73. Finnish tuoli = a chair. Kafir isi(tula)

= chair.
Secw. se(tulo)
= chair, f

74. Finnish kaula=chest. Herero (oru) kora= chest.

Mpongwe (e) gara= ,,

75. Finnish powi= chest. Zulu-Kafir fuba = chest (f =p ;

b=w).

76. Finnish walikoita = to Herero vareka = to choose (v

choose. =w ; r=l).

77. Finnish kuweta = to climb Zulu-Kafir kwela = to climb.

Suaheli kwea =

78. Finnish kappa=mantle. Zulu-Kafir (in)gubo= mantle.

Secwana kobo =
*In the Sumerian word the a belongs to the root, for a = water and kal =
strong. In Zulu the first k of kukulu is probably affixed, or the prefix .was
rather ku, so that we would have ku-akulu = kukulu. Later on the fact that
tin word really had a prefix already was lost sight of, and a later prefix I'M'
added.
fit is quite possible that the Kafir and Secwana words are mere adaptations
of llu- Dutdi word stoel = chair, and that on the other hand the Finnish word
is also a Teutonic loan-word, as many other Finnish words are. The Finnish
may, however, be directly derived from the Turanian rout Ilia to raise, from
which probably our throne (with the usual inserted r) is derived. The Mpou-
gwe (,s a )t(iu<ia and the Ron^a (tlii)tama, which may be related to the Finnish.
are certainly not derived from the Dutch. I believe that the chances are that
the Kafir and Secwana words are of Dutch origin.
 22

79 Finnish watteus = clothes. Kongo (mo) watu= clothes.

Kafir (izi)vatu =
80. Finnish tulla = to come. Secwana tla = to come.

81. Finnish saalia = to com- Zulu sola = to complain.

[plain. Zulu (i)silola
= a complaint.

82. Finnish sakkoon = to con-, Secwana sekisa=to condemn.

[demn. Suaheli hukumu =
*
,,

83. Finnish wamma = damage. Kafir (uk)wona= damage.

Suaheli vunja =

84. Finnish hata = danger. Herero o-kuta= danger.

Suaheli hatari =
(See No. 64, where the probably older Sumerian form is

given.)
85. Sumerian mer-lam = day . Zulu (u)muhla=day.
Finnish wa-lo= Kongo lumbu = ,,

86. Finnish sampaa= button. Kongo sumbo = button.

87. Finnish pukea = to bury. Herero paka = to bury.

88. Finnish mytty =bundle. I Secwana riata=bundle.

(ri=ny.)

89. Finnish (ky)psya

= to burn '

Mang'anja psya=to burn.

90. Finnish sywa = deep. Zulu suve = deep.

91. Finnish synkka=deep. J Kongo (a)xinga=deep.

92. Finnish kokko=peak. Zulu (isi)gonga=peak.

(N.B. In these two last examples the kk in Finnish chan-
ges into Bantu ng, which is in perfect harmony with the phon-
ology of the latter group.)
93. Finnish kunta
= family. Herero (otyi)kutu= family.

94. Esthonian kaza=wife. Herero (omu)kazendu wife.

Kafir (in)kazana = ,,
Kafir kazi= female.
(The etymology of Herero omukazendu is as follows : omu
= article + prefix for human beings kaza ;
= woman or female ;

ndu = person.)
Although the Suaheli hukumu would agree letter for letter with Finnish
sakkoon,it may be that the former word is a direct adaptation of the Arabic
hakam, from which language many words have been taken over in Suaheli.
 23

95- Finnish kisa

= to dance. Suaheli chesa= to dance.

96. Ost-iak sana=fine. Secwana se-sane = fine.

97. Finnish saulii=care. Kafir i-xala=care.

98. Sirjenian si=sound. Zulu (um)sinda=sound,

99. Livonian ibus=hair. Kafir (u)baya=hair.

abbon = beard.

100. Sirjenian si = a hair. Mang'anj a tsi-tsi= hair (plural)

Permian si=

Although there may be in the above list a few words, the con-
nection of which might appear doubtful, the vast majority
certainly tend to show how intimate the relations are between
the Ugro-Altaic languages and the Bantu languages.
It will be seen that in Sumerian, as in a few other languages
of the Ugro-Altaic group, there seems to be a tendency to end
the words and syllables in consonants instead of keeping them
open. But this tendency is possibly only apparent, for we
are only able to read Sumerian by the light of Cuneiform script,
and it is, in my opinion, still a question whether, in pronounc-
ing the words, the Sumerian did not add a vowel after"the last
consonant.
It will also be noticed that a large number of Bantu words
are directly traceable to Sumerian, with but very slight changes,
and hence the question arises, whether we are on that account
entitled to consider the Bantu, or, at all events, a considerable
portion of them, as direct descendants from a Sumerian stock.
This question will be treated in this essay at a later stage, but
here we have first to consider another and more immediate
question.
The fact that in the above list there is such a marked simi-
larity between the Bantu roots and the Ugro-Altaic roots, must
not lead us to the conclusion that Bantu is derived from Ugro-
Altaic. This would be a serious ethnological mistake, for the
differences between the two races, from a physical and ethno-
logical point of view, are too wide for such an assumption.
The Finns, for instance, and their nearest relations, the Sir-
jenians, have an undoubted strain of Aryan blood in their
veins, and Aryan influence is also traceable in their language. In
the Mongolian, Manchu, Jakutian, Tungusian and Japanese
members of the Ugro-Altaic ethnological group there is a strong
strain of what is generally called the Mongolian or Tataric ele-

ment, an element which is at present not easily defined, but

which would seem to be connected with the influence of an
ancient Malayan race.
 The Bantu have no strain of Aryan blood, as far as we can
tind out, nor do they show the marked characteristics of the
Mongolian group, such as for instance the greater .or less
obliquity of the eyes, though this was, to some extent possessed
by the races which in the very oldest times inhabited the coun-
try near the mouths of the Tigris and Euphrates.
The proper view to take, therefore, of the question in hand
is to consider that both the Ugro-Altaic race and the Bantu
race sprang from one common stock, centuries and centuries ago.
When they parted,. from that common stock cannot be definitely
fixed, but considering the fact that the Sumerians are known
to have inhabited the country near the mouths of the Tigris
and Euphrates as early as 5000 B.C., we must come to the con-
clusion that Bantu and Ugro-Altaic had then already entered
each upon their own path of development.
In a matter like this historical ground fails us altogether.
It is true that we may be able to trace certain myths, but after
all the only sound footing we have is in the department of

Philology, and we are absolutely limited in our researches to

that department, though able, here and there, to call the
somewhat uncertain science of Anthropology to our aid.
In doing so, we are at present compelled to take a jump, and
to skip, for the nonce, certain intermediate stages which will
be treated of afterwards. I must therefore ask my readers to

accompany me to the Malacca Peninsula, where I believe we

will find the origin of the Bantu.

SECTION IV.

In August, 1906, there appeared in England a book which,

in a remarkable degree, combined that practical view of
matters, so characteristic of the Anglo-Saxon, with the sound
one is accustomed to meet with in German
scientific scholarship
and French works and which (alas !) has been so often
found wanting in even the best English works. The
'

title of this work is The Pagan Races of the Malay

Peninsula," and the authors are Messrs. W. W. Skeat
and C. 0. Blagden* The former is already well known
to the general English public as a man intimately
acquainted with the Malay language and Malay magic. Mr.
C. O. Blagden has been known to the scientific world as a
valuable contributor to the Journals of the Royal Asiatic
" "
Society, especially to the Strait Settlements branch
thereof, and his co-operation with Mr. Skeat in this valuable
work considerably enhances the value thereof.
*The publishers are Messrs. McMillan and Co., London, and the work is a
credit to this already renowned firm. The works (two volumes, at 2 is. each)
are splendidly illustrated, and printed very nearly without a single mistake.
 25

It is impossible for me to give even a short resume of this

work in these pages, and I must refer readers to the work
my
itself. I can only state here the bare facts.
In general the Pagan inhabitants of the Malacca Peninsula
can be grouped under three headings, viz. :

A. The Semang.
B. The Sakai.
C. The Jakun.
Though, as the authors have very clearly shown, there are
large portions of the Malacca Peninsula in which the three races
have more or less blended, the characteristics of each race
can yet be easily traced in several tribes, which have kept
themselves fairly pure. These characteristics may shortly
be defined as follows.*
A. Semang Type. Height of men about 1,491 mm.; of
women 1,408 mm.; skull-index brachyo-cephalic (or bullet-
headed) to mesaticephalic skin of a dark copper or rather
;
"
chocolate-brown colour, passing into a shiny black "; hair
(which is generally shaved off) woolly, like that of the Negro
and the Papuan forehead low and rounded nose remarkably
; ;
"
broad and flat or spreading "; cheeks full, but with the
cheek-bones not very prominent eyes round, wide open, and
;

straight (i.e., not oblique like those of the Mongolian races) ;

chin feebly developed (i.e., rounded off, and frequently almost

unmarked) mouth variable, but rather large as a rule, the
;

lips,which are also variable, being generally well formed, but

" "
sometimes turned outside or everted beard, none to ;

speak of, as a rule, but when found, thin and straggling, or,
occasionally woolly like the hair itself, f

B. Sakai Type. In height the Sakai are, I think, without

doubt, a slightly taller race than the Semang or Negritos.
The shape of their head, on the other hand, is in marked con-
trast to that of the Negritos, as they belong in type to the
dolichocephalic, or long-headed races. Their skin-colour
varies more remarkably than that of any of these tribes,
being in some extremely dark-brown, in others a remarkably
light yellowish-brown, much lighter than that of the Malays.
Their hair, too, is long, black, and wavy in character, some-
times with a slight reddish tinge in reflected light. Their
forehead is flat, and projects remarkably over the root of the
nose, which latter is, as a rule, somewhat fine and small, and
often slightly tilted at the tip. The cheek-bones are very
*For particulars and for the fine portraits of the various races I must again
refer the reader to the work itself.
tThis description, as well as those of the Sakai and Jakun, have been taken
verbatim from Skeat and Blagden's above work, vol. I., pp. 34-36. In the
appendix to the work, pp. 573-602, the reader will find very valuable further
anthropological data, which I could not take up in this sketch, but which
contain much information from the hands of Professor Virchow.
 26

broad, especially when considered in relation to the rest of

the features. Their eyes are a very dark brown, small, hori-
zontal, and often half closed, as different as can be from those
of the Semang. Their chin is long and somewhat sharp and
pointed. Their mouth is of small size, with lower lip full,
loose, and often conspicuously projecting. Their beard is,
as a rule almost non-existent, but a few individuals occur
who are fairly well covered with hair.
C. Jakun Type. In height the Jakun appear to be, if
anything, a little taller than the Sakai (in which case they
would be the tallest of the three aboriginal races). Their
head is brachycephalic or "bullet-shaped.'.' Their skin is .

generally of a dark coppery colour, not unlike that of the

.Malays, but with a tendency to darker shades, which are,
doubtless, due partly to exposure and partly to their manner
"
of living. Their hair is long and straight, or smooth," and
of a dark bluish-black tint, such as is invariably found in the
hair of Mongolian races. Their forehead is usually well
developed. Their nose is, as a rule, thick, flattish, and short,
with wide open nostrils, though it must be noted here that a
more developed type occurs. Their cheekbones are very high
and well marked, like those of the Mongolian type. The face,
as a rule, is inclined to be flattish. Their eyes are dark brown,
of moderate size, and rarely with some slight tendency to
obliquity. They have, as a rule, a strong chin and somewhat
square jaws. Their mouth, as a rule, is large and broad,
though frequently moderate and with well-formed lips. Their
beard is of the scantiest.
Thus far the authors of the " Pagan Races of the Malay
Peninsula."
In the appendix to Vol. I. the following data are given :

Semang. Cephalic index 78.9 81.1 :

74.4 80.6 81.6; ; ; ; ;

85-
Sakai. Cephalic index 77.4 :
; 76.1 ;
73.8.
Jakun. Cephalic index 76.7 :
; 83.3.
Capacity :

Semang. 1245 c.c. 1425 c.c.

;

Sakai. 1370 c.c.

Jakun. 1032 c.c. ;

1190 c.c. ; 1230 c.c.

As I
superficial know-
do not lay claim to any but the most
ledge of the science of Anthropology, I shall not discuss these
figures, but I beg to express an opinion that it would be well
ifsome South African scientist, who is acquainted with Anthro-
pology and skull measurements, would go to the trouble of
comparing with each other these figures with :

(a) The known figures of the non-Aryan races of India.

(6) The known figures of the Hamitic races, i.e.. Galla,
Somali and Abyssinians.
 (c) The known figures of the Hottentot, the Bushman,
and the principal Bantu tribes.
(d) The known figures of the true African negro.
(e) The known figures of the principal Ugro-Altaic tribes
and races.
Iventure to think that a careful examination and com-
parison of the anthropological data of these races would lead
to very important results.
Leaving such researches to men who are better qualified
than the writer, I must now return to the discussion of my
own department, that of languages. I have drawn up a very
careful list of Bantu words and of words of the three languages
of the Malay Peninsula, and I confess that the results were
somewhat startling to me. As I wish to show how I came
to these results I beg to give here a list of some words. On
the left side will be found the Malacca words, on the right side
the Bantu words, and here and there I have given some illus-
trations from other Ugro-Altaic languages. The list is by no
means complete, but it is quite large enough for practical
purposes, and certainly tends to show what the relations are
between the races of Malacca and the Bantu races, and how
far their lexicological affinities extend.

SECTION V.

Malacca. Bantu.

1. Sakai pela = to abandon. Bobangi kila = to abandon

(p
= k).
2. Semang ;a = to abandon, <

Zulu (shi)ya= to abandon.*

Kafir (&)yeka= ,,

3. Semang pe = above. Zulu pe-(zulu) = above in the

heaven.
Wisa pa= above.
4. Sakai puru = abscess. Bobangi mpdta= abscess (r=t)

5. Semang */& = afraid. Luganda tya = to be afraid.

Mpongwe tia= afraid.
Shuna tya = to be afraid.
Haussa tuizer a = afraid.
6. Semang te-koh = after- Ronga (an) thaku = afterwards,
[wards.
"
"The ski in Zulu is naturally no prefix here, "but probably means a thing,"
so that the word is a compound one, meaning to abandon a thing."
 28

7 Sakai />w/a=again. Umbundu (olu)wali= again.

(p=b=w).
Oshikuama va/=again.
Aduma bua = again .

8. Semang ga-ga= strong, Kongo (a)-ngo-lo strong.

(reduplication of root ga.) Wisa (a)-ko-sa= strong.
Luganda gu-mir= strong.
(N.B. mir = man.}
Oshindonga (zi)-gw.
Compare Sumerian god
[strong.

9. Semang (ya)-sol
= to stroke Secwana shola = to stroke.
Tebele (e)sula
=
(Japanese sasura= )

io. Sakai posh=storm. Zulu (isi)pepo= storm.

Kafir (isi)pango= storm.
Wisa (im)/><?o=storm.
(Sumerian papa= storm.)
n. Semang begiyu= storm. Luganda (ki}buyaga= storm.
(omu)yaga = storm,
(with metathesis of y and g.)

12. Sakai helu= storm. Mpongwe (o)gula= storm.

hal-hiU = (Sumerian tm-hul = storm.)
(with reduplication.)

13. Sakai dudao=stick. Mang'anj a (n)dodo = stick.

Kamba (ri)deta
= stick.

14. Sakai lu-dal= stick. Luganda (o)luga= stick.

Zulu (u)luti= ,,

15. Semang balu= stone. Zulu (u)m&ek= stone.

Manganja mwala =
Dualla (i)dali= ., (d
= b)
1 6. Semang* (ya) nyan = Zulu niana= to stand
[to stand. (ny = m).
Kamba (u)ngema= to stand.
(ny = ng).
17. Sakai jin-juk = to stand. Kongo jujila
= to stand.
(with metathesis ?)

18. Sakai pit

= to squeeze. Secwana pit-la
= to squeeze.
Boban^i pioto
=
19. Semang pia = to squeeze. Wisa fina = to squeeze (p =f )
.

20. Sakai kenan to squeeze. Zulu ka-ma = to squeeze.

Manganja kanikiza= ,,

21. Sakai baru = -& plain. Secwana (\o)-bala=a. plain

22. Semang z0w = a dog. Herero mbua = a dog

(w = mb).
Secwana mbwa=& dog.
(and in several other lan-
guages.)
23. Sakai cha(cho) =dog. Kikuya chui = dog.
chau= ,, Zulu (i)govu= ,, (ch=g).

Semang chu= ,, Mang'anja garu= ,, (ch=g).

Kamba solu= ,, (ch=s).
(Compare Sumerian (ur)chu =
dog).
24. Semang nyang = dog. Zulu (i)/a=dog.
(Compare Japanese (i)nu =
dog).
25. Semang 'mpa = to dream. Zulu pupo = to dream.

26. Sakai nvo = to drink. Herero nua = to drink.

Secwana nwa ,,

Mang,anja mwa= ,,

Luganda nywa =
(Japanese nomu= ,, )*

27. Sakai ngun = to drink, Kamba gunwa = to drink.

(another form of No. 26. Aduma egnua=

28. Jakun jo-oi

= to drink. Mpongwe jongo= to drink.
(Syrjenian juny = to drink.)

29. Sakai kara=to dwell. Herero kara = to dwell.

Kongo kala =
Snaheli ka-a = ,,

Shuna gara = , ,

30. Semang ya to dwell. Secwana (a)ga = to dwell

(g=y).

Zulu (a)ka = to dwell.

The old form of the Sakai is actually found back in the Dualla nyo = to
drink.
3i. Sakai ba, be, bi, bo, bu*= Zulu = father.
father. Kafir ubawo =
(See Skeat and Blagden op. cit.

II., page 598.)

32. Semang ta = father. Herein tate father.

Sakai ta-ta = Mang'anja tat a ,,

Dual! a titc=my father.

Kongo ta-ta = father.
Isubu te-ta =
And many more languages,
both in Bantu and in Ugro-
Altaic. Secwana ram and
Mpongwe rere are probably
the same forms, where the t

has been changed into r.

Luganda kitanga, kitawe, etc.,

probably belongs to this
group, though the prefix ki
seems inexplicable here.

33. Sakai mahitnfat (adj.) Zulu cuncifutai'Ai ( n )-t

Mang'anja mafuta= (n).
Secwana mahura= (n).
Luganda amasavu- ,
(n)
(h=s).
Ronga t (n).

34. Semang w/*=fat (adj.). Kongo amaf*=sfat(adj.).

meche =

35. Semang (en)cA=: fat (adj.) Shuna kora = to be fat.

Sakai be-cho = fat (adj.) Herero e-kara id.\.
(adj.).
Bobangi bakatala=to be fat.

36. Sakai = fat (adj.) Luganda -gcvu = iat (adj.)

[Compare with 33-35.
Japanese koeta=ia.t (adj.)
Manchu hetu= ,,
Sumerian (mi)-chu= ,,
(n)
Jakutish sia (for jin)

*These rootwords, which include the whole scale of vowels, are of great
importance, and tend to show that at any rate in these and related groups
"
vowels cannot be considered as fixed elements of the language," but that
they are changeable as circumstances demand. Such would be in absolute
agreement with the Vowel-harmony of the Ugro-Altaic languages, of which
traces are found in Bantu.
fThese and the examples given under No. ^4 show that the lact that
(innifuta is now in Bantu considered as a plural noun, is simply the result of
false analogy, based upon the supposed prefix ama.
 31

37- Sakai jaras= finger. Mang'anja chala= finger,

Luganda (en) galo= finger.

38. Semang chand-ras Suaheli chanda= finger.

fingernail.
39. Semang ^'
Ronga (li)
-tiko = finger.
(In several old languages
"
the words for arm,"
" "
hand," finger," are
confused or intimately
connected.)
40. Sakai ayam=i<y\\\. Kongo eyembe = fowl.
Luganda enyonyi= ,, (?)

41. Sakai dena=iow\. Herero ondera =iowl.

(Compare Japanese tori
= fowl.)
42. Sakai puk: pok: Zulu inkuku = fowl .

pup: keok: Kafir inkuku = ,,

= fowl. Mang'anja nkuku= ,,

N.B. The form keok is Secwana kok= ,,

found in only one dia- Luganda enkoko= ,,

lect, but it shows the Suaheli kuku=

tendency to interchange Wisa nkoko =
P and k.

43. Sakai seng = forest. Zulu (i)singa= forest.

Secwana sekgwa = ,,

44. Jakun hu-tan= forest. Herero (otyi)hua

= forest.
Oshidonga (oshi)c/mfl
= ,,

Wisa (i)chonde= ,,

Mpongwe (i)g a==

(Compare Finnish kalra
forest.)
45- Semang maku = egg. Manganj a maika = egg.
Isubu moko= ,,

Kamba itumbi = egg.

46. Sakai tab = egg. Ronga tanda , ,

(Compare Japanese tamago

= egg.)
47. Sakai zai-s= blood. Luganda (om\\)sai
= blood.

48. Semang yap= blood. Zulu (i)gazi

= blood.
Kafir (i)gazi
= blood.

49. Sakai hawa=nir. Suaheli hewa=u\r.

(Compare Turkish hewa= air.)
50. Semang Luganda (o) muka breath
Zulu (u)moya=inr.
Mang'anja mpweyaair.
Secwana moea=nir.

51. Sakai /r/^=arrow. Kafir (u)to/o= arrow.

52. Semang halag = lizard. Oshindonga ekalala= lizard.

53. Sakai huwah = to love. Herero sttvmz = to love (s=h).

,, huura =
(Comp. Japanese horeru

54. Sakai = to love. Oshindonga chola = to love.

(Probably another form Oshikuanjama hola = to love.

of 53-) Luganda (a)ga/a = to love.

55. Sakai pad, /><'

= new. Suaheli Pya = ne\\ .

Wisa pya=ne\\.
Herero />r
= new.
Mang'anja (tso)/>am? =nc\\ .

56. Semang Luganda (e)^*>o=night (h=k)

= night.
57. Semang tu-woi ; tu-wi Herero (ou)tnku = night.
= night.
58. Sakai to = pig. Mang'anja
Ronga
Zulu (ingu)/i*6=pig.*
59. Sakai bis=pig. Luganda (em)fo>j'
= pig.
Herero

60. Sakai kumo=pig. Wisa (in)smfa=pig.

Mang'anja
Kafir
(h=k).
61. Semang pe = to cook. Zulu peka = to cook .

Semang sema=
-
62. all. Suaheli ima all.

63. Sakai kile= angry. Wisa (u)kali

= anger.
Bobangi ;z /c c/c/r = anger.
Ivamba killalu =an^i v.
Suaheli (\\}kali
= anger.
Luganda (e) kirn urn Danger.
"
*In Sakai gau is also pig," and it is quite possible that the above three
Bantu words are a compound of two repeated words meaning the same, a
linguistic phenomenon not uncommon in Eastern language. Kamba has for
" "
pig ngue, which is clearly related to gau.
64. Sakai jeboli=a.rm. Suaheli

65. Sakai (a)&7/ Mpongwe (o}mbu = ashes.

Luganda evu= ,,

(b
= v).

Isubu = ashes. 4

66. Sakai charu=bed. Secwana

67. Sakai bok = to bind. Zulu bopa = to bind (k

= p) .

Secwana boha = (k = h).

68. Sakai (che)kat=to bind. Aduma ckota = to bind.

Bobangi koto =

69. Sakai kage = to bite. Aduma (e)kaga=to bind.

70. Sakai lemun=tootih. Zulu liuna=to bite.

lemuin =to bite. Kafir luma =
Luganda lnma= ,,

Secwana loma =
Oshindonga lumana ,,

Luganda ku = at.
Wisa ku= ,,

Umbundo ko=
Secwana kwa= ,,

72. Sakai ba-kul = basket. Bobangi (e}boko

= basket.
Kongo (m)bango =

7,;. Sakai Zulu (i)gowowgo=bat.

74. Sakai ti = to be. Mang'anja ndi = io be

Herero (t=r).
Wisa // =

75. Sakai (en)^f=breast. Adumi /?///^=chest.

chest. Bobangi ntolo= ,,

Kongo tulu = , ,

Oshindonga (on)//o= ,,

76. Sakai bit = iVmale breast. Zulu (i)6e/= breast.

Kafir (ama)6^/<?.
= breast (plural)
Luganda (e)bere = breast.
Bobanbi libelc = breast.
Dualla (di)be= ..

[G.i;. 1907.]
 34

77- Sakai bah = brother. Mang'anja (m)balc = brother.

(m)bungi =
j

Kongo ,,

78. Sakai kolo = elder brother. Zulu (um)kuluwc

= elder brother.
Oshikuanjama (omu)kulu
= elder brother.
Aduma (mu)kulu
= elder brother.
79. Sakai chengru = to call. Suaheli changa=to call.

Mang'anja chemera= ,,

Herero (i)satiga
=

80. Sakai kap = to catch. Secwana a = to catch.

81. Sakai kat = io catch. Zulu (qa)a/fl = to catch.

Mpongwe kota =
Wisa (i)kata= ,,

Oshikuanjama kuata= ,,

Luganda kwata =
82.
 35

In the above list I have only taken up a certain number of

words, which agree both in form and meaning, but if related

meanings and somewhat divergent forms were taken up in it,
the list could easily be brought to include 400 or 500 examples.
It has been said that language in itself does not prove any-

thing, but in cases where such complete harmony exists as in

the above list, I think the statement requires some qualifica-
tion.
But it could be shown that a good deal of the customs of
the Bantu races may be traced in origin to the Semang and the
"
Sakai. The former build " beehive structures, like our
Southern Bantu the latter build the rectangular oblong huts
;

found in Central and East Africa. Many of the marriage

"
customs between the races agree, and the spirit and ancestor-
worship," are found in the Malacca Peninsula "
as well as in
Africa. Even the painting or " smearing the body with
white paint or clay is used in certain religious or quasi-religious
ceremonies by the Semang and Sakai, just as is done at present
at the circumcision rites in several Bantu tribes.
Yet it would be a mistake to suppose that the Semang and
Sakai ever came from the Malacca Penirisula to Africa. On
the contrary, I am convinced that many centuries passed before
the ancestors of the present Semang and Sakai, who left the
Malacca Peninsula thousands of years ago, reached Africa.
But this important subject requires some paragraphs of its own.

SECTION VI.

Messrs. Skeat and Blagden's book contains in Vol. II., pp.

379-472, a very valuable introduction to the study of the lan-
guages of the Malay Peninsula. They rightly class these lan-
guages under the so-called Mon- Annum family, a name given
originally by Logan, but often called on the Continent the
Mon-khmer family. Both names are not very fortunate, as
they really denote an artificial mixture which, in fact, does not
exist.*
That the language of Annam is probably a purer language
than the languages of the Malacca Peninsula, I am convinced ;

but I do not think that this entitles us, as both the above-
iii' ntioned authors and that great authority Schmidt do, to
come to the conclusion that the Malacca Peninsula was in-
vaded by a Mon- Annam race from the North. On the contrary,
I am of opinion that the opposite has taken place, and that the
*Ainun)2 or Annamese is the language spoken in Cochin-China, Annam, and
Tongking. Mon is the same as the Talieng language, spoken in Pegu. Khmer
is tin-usual philological name for the language of Cambodja. Dialects con-
nected, and closely connected with these.^are, however, spoken in India.
Malayan Peninsula was originally invaded from the South, at a
time when probably the Indian Archipelago and Australia
formed one immense continent.
In any case, as Messrs. Skeat and Blagden have taken con-
siderable pains to prove, it is perfectly clear that in the Malay
Peninsula we have two very clear types, as well as a third one,
which is at present somewhat doubtful.
First, we have the Semang or Negroid type.
Secondly, we have the Dravidian type, shown in the Sakai.
Thirdly, we have the Jakun type, which has remained the
least pure, but which shows in many points resemblance to the
Mongolian type.* In how far this "
Mongolian type, can be
considered as an original type is a vexed question." From a
linguistic point of view Mongolian as a language shows any-
thing but originality, and is undoubtedly a Misch-Sprache (as
the Germans call it) containing botli Turanian and Dravidian
elements.
In my opinion we have in the Sakai the pure Dravidian
element, or better the so-called Hamitic element. Anyone who
looks at the faces and the build of the Sakai as illustrated in
the work of Messrs. Skeat and Blagden, must be struck by their
resemblance to the Gallas, the Somali, and the Southern Abys-
sinians ; and if one studies the languages of the Gallas and
the Somalis (for Abyssinian has been too much influenced by
Semitic to allow a clear comparison) he must be immensely
struck by the similarity of the roots of these languages and the
roots of Sakai. f
The Semang is the Negroid, pure and simple. Unfortunately
the term Negroid is a very badly chosen one, and its relation
to the term Negro is liable to misconstruction, as if the former
was a kind of graft upon the latter, a supposition which is
absolutely at variance with the facts. I should propose to
call the Semang a Proto-Turanian, i.e., the main element in
that immense race, which we now call the Turanians, but
which at present contains Dravidian and perhaps even Aryan
elements. The term Turanian, indeed, is often given to all
the non-Aryan languages of Asia, and seems to have been
used by Max Miiller in that sense. There is, however, a very
great divergence between these languages, though Max Miiller
was undoubtedly right when he assumed "a genealogical
"There can scarcely be any doubt that thejaktin type, as at present existing
in the Malacca Peninsula, has been strongly influenced by the Malayan ele-
ment. Unfortunately nobody knows to-day what the Malay really is. Per-
sonally I am convinced that he is not indigenous to this part of Asia, and I am
very much inclined to believe that he was of Semitic origin, and that he "did
not settle in these parts much before ,500 B.C. In all probability he migrated
i

from Southern Arabia, and is really a Sabacan.

| Although I carefully went into this matter, and have actually drawn up
comparative lists, this subject cannot be treated in this essay, but must remain
over to a future opportunity, if such should offer itself. Vita brevis ars lonqa
must apply here.
 37

relationship of Turanian speech." It is, in my opinion,

all
most an early period, portions of the Semang
likely that, at
or Sakai race went far north in Asia and Europe, but that,
through the occurrence of the second glacial period, they were
driven again south to more equatorial regions. Their long
absence from the parent stock, the different environment and
the influence of the colder climates, must, have had consider-
able results, not only physical, but also linguistic and perhaps;

this explains the divergence between the present Mon-Khmer

group of languages and that generally called the Tibetan-
Burman group. The people speaking these latter dialects
certainly came from the North, as the most ancient legends
show, but this coming may have been in reality a returning.
The Khonds in India have a decided recollection of once
having lived in the far north, beyond the Desert of Gobi.
That man existed upon earth with the Mammoth, the Cave-
Lion, the hairy Rhinoceros, and the other animals which trod
the northern earth between the first and the second glacial
period, is a fact accepted at present by most geologists and
anthropologists, although some difference of opinion exists
about the placing of this period, and some scientists are
inclined to accept only one glacial period.

The reason why I 'say that the great Turanian linguistic

group contains at present even Aryan elements, is on account
of the fact that in a large number of the Northern Altaic
languages (as distinguished from the Southern Altaic, or Bantu]
there are undoubted traces of Aryan in some of the races
;

which speak those languages there is even a strong Aryan

element.
In short, I may state that I should be inclined to defend,
both upon ethnological and philological grounds, the follow-
ing propositions.*

I. At a very early stage in the history of mankind the

greater part of Further India was inhabited by a Negroid or
Proto-Turanian race, which may have been aboriginals of this
part, or may have moved thither from what is now the Indian
Archipelago. At present we have not sufficient facts to
liable us to determine the latter question.

II. At some time or other in gray antiquity i.e., many

thousand years before Christ, and perhaps even in an earlier

geological period of the earth this part of Further India was

*It may be stated here that these propositions are by no means new or
"t'li-dhi^. Onatrrfa^fs, /V .Minimi and other learned men have held more or
less the same views, though these have been violently o>m bated
by other
M -ion lists. The latest discoveries, to which I shall refer later on, do, however,
show that Otiatrefages especially had hold of the riqht end of the question, but
.'hat his views \\anl careful testing.
 J8

invaded by another race of mankind, which we may call the

Hamitic*
This later race caused a migration of the Negroid race, only
a very small number of individuals remaining in Further
India, and these are the ancestors of the present Semang in the
Malacca Peninsula. The smallness of their number, and the
fact that they were, no doubt, hunted from place to place by
the newcomers, prevented the Semang from either increasing
at an ordinary rate, or from making any material or intellec-
tual progress. To-day the Semang belongs, with his relations,
the Andamanians and Nicobarians, to the lowest races of man.
In the Malay Peninsula several of the wildest tribes are still
nomads, who do not stay longer than 3 or 4 days at one place,
and the fact that this custom is ascribed to a command of the
Deity, points to the conclusion that once upon a time it was
absolutely necessary for their existence, that their enemies
should not become acquainted with their hiding places. (Skeat
and Blagden, I., p. 173, note}.^
The Negroid race that left Further India after the invasion
of the Hamitic race, wended its way westwards, and increasing
in number, continually moved in sections to the west, so that
itonce upon a time formed the population of, at all events, the
southern part of Europe. This has now been conclusively
"
proved by the discovery of the so-called Man of Grimaldi,"
who was found in Italy a short time ago by the Prince of
Monaco. The skull of this man is of pure Negroid type, and
very nearly approaches that of the present Semang.\
But this Negroid race also invaded Africa, and here he
changed into that sub-race which we now know as the Hottentot
Race.
I am quite aware that this is a somewhat startling assertion,
but there are a few proofs, which simply put the question be-
yond doubt. In the first place it is well known that the
Hottentots are a very old race in Africa, and that in the very
earliest time of historical Egypt they inhabited the land of

*I call this another race because at the time it invaded Further India it had
clearly different characteristics from the Negroid races. But this fact does
not really militate against the principle of I'nogenesis, because it is quite pos-
sible that, at some remote time, the two races sprang from one source, but that
they were compelled by natural circumstances and environment to follow
different lines of development, which ultimately resulted in such variations,
that they practically formed two races.
fin other races nomadic habits were developed by scarcity of food (e.g., in
the case of the Bushman in South Africa) in later cases, by the necessity of
;

tinding pasture for flocks (as is still done by several tribes in Central Asia).
None of these causes could have operated in the Malay Peninsula, where both
fruit trees, edible roots, and small game are. even to-day, very plentiful.
Hence the agency must have been human here.
JFor particulars of this very important discovery I must refer my readers
to an article by the renowned Dr. Verneau in L' Anthropoloi>ie of May, 1906,
Vol. XVII., pts. 3-4, where the matter has been fully discussed. According to
Dr. Verneau this negroid type is actually still traceable in some of the present
inhabitants of Italv !
 39

Punt. In 1533 B.C. an expedition (by no means the first)

was sent by Queen Hatshepset to Punt, evidently the present
Somaliland, and of this expedition we have a full account and
several bas-reliefs, which will be found in Dr. E. A. W. Budge's
"
History of Egypt," vol. IV., pp. 5-11. The illustration on
page 7 in this book shows the Hottentot with all his peculiar-
ities,even to the steatopygia.*
"
This must settle the fact that the Hottentot is a very old
"
inhabitant of Africa. But there is remarkable linguistic
evidence to connect the Hottentot with the old Negrito of
Further India. It is well known that the Hottentots called
themselves Khoi-Khoin, which is a plural form, translated as
Men or (wrongly) " men of men," the reduplication being
simply the ancient form of the plural, found in many lan-
guages, even in Sumerian. Now in some Semang dialects koin
is a man, and in Nicobarese koinh, or koin, is the usual name

for man or male.

Although I have gone to considerable trouble in the matter,

I have been unable to find any other
languages besides the
above, in which the word koin or khoin is used in the sense of
"
man," although related words are found in other languages
of the Mon-Kmer group and this certainly shows that there
;

must once upon a time have existed either a relation or an

identity between the Hottentot and the Negroid of Further
India.
But if one further carefully compares the language of the
Semang and the language of the Hottentot, there is much in
favour of the conclusion that once upon a time these languages
were identical or nearly so. Naturally Hottentot has changed
and has absorbed new elements in the 70 or 80 (or probably
even more) centuries, since the Hottentots left the parent
stock and the Semang and the Nicobarese have changed,
;

though in a different direction. Not being yet in possession

of any good work on Nicobarese or Andamese, I have not been
able to devote much study to this interesting subject, but
hope to do so soon, and will then publish my conclusions in a
short form.f But the following few words are certainly re-
markable in their close resemblance :

'^.Hottentot. Sein-in^.

Kliarob=bed. ,
Karob =bed.
khora to stretch. |
cherok =to stretch.

*Skeat and Blagden in their aforequoted work, Vol. I., p. 578, mention a
cas< <>f
steatopygia in a Scmaii" woman. It is a pity that this matter has not
been treated more fully by the authors.
t-Vn article on this question was published by me in the .J/V/r,/i/ Monthly
of June, 1907, written since this essay was put in type.
"
+ For Hottentot I have principally made use of G. H. Schils' Dictionnaire
"
etymologique de la langue des Namas (Louvain, 1895).
 Hottentot.

khui =t<> arise. kviu =to ar

sa = to cha =to go.
tsa =blaok. -va = black.
a =to make a noi- a-eli =to roar.
a-rib (=a + ri) =dog. fi-eh =dog.
H =to drink. auuh =to drink.
ao =to throw. fi-uin =to throw.
cha =to burn. chfim =to light a lire.

chob =cheek. kaba =cheek.

chob = beard. i
sabau = beard.
da = to 1*0. dali =to go.
do =red. to =red.
di =to play. '
de =to play.
ebe =much. ebeh =much.
ga =to speak. cha =to speak.
ga =hard. (te) ga =hard.
gei-rab = jackal. cha-lau = jackal
(ch=g; r=l).
ha (mo) = when? i ha (nadeh) =when?

Even the clicks in Hottentot can be traced back in Semaug.

So if we represent the dental click by we have the Hottentot
|

|
aob= snake agreeing with the Semang ekob, which was
probably originally jkob or gkob ; Hottentot az>*6 = rain |

is the Nicobar ami = rain, originally gmi ; Hottentot aro = |

to increase, find its correspondent form in Nicobar karu = to

increase, which was probably ykaru in early times. As
already said, I have not yet been able to give the subject
that attention necessary to form a definite opinion, but from
the little I have seen, I believe that the study of these two
languages would lead to astonishing results.

It is well known that the Hottentots have some idea of a

Divinity whom they call Tsuiguan or Tsuni-t>oan. Dr. T.

Halm has written a work upon this, which, however, contains
most improbable theses, especially the idea that Tsitni-ooam
"
was a Solar God. The fact is that Tsuni-goam is the Thunder
God." His name still appears among the Ao-Nagas in India
as Tsun-gretn. This is clearly the Semang so=
spirit, and
the Semang kaii or kare = thunder, which latter name in
itself constitutes a God of a Semang tribe. (Compare the
Semang form

*Masudi calls the God of the Zanj. Maklandjalou, which probably meant in
"
ancient Bantu the Spirit of the Lightning." The Southern Bantu believe
" "
that the Great Chief has control over the lightning, and they never mourn
over those who are killed by lightning, as they were " sent for " by the Great
Chief.
 At present this is all I can say in support of the theory, that
the Semang who landed first in Africa, became there that race
which we know as Hottentots.*
But the whole of the Negroid or Proto-Turanian race, which
had left the Malacca Peninsula and Further India, did not
wend its way westwards. A considerable number settled in
the Indian Peninsula, now known to us as Hindostan.
Messrs. Skeat and Blagden have in a little map on page 442
of Vol. II. of their above-quoted work shown that the so-called
Mon-Annam Family spread to India, and have rightly classed
the Naga Territory as one of the places where languages
related to the Semang and Sakai are spoken. But it would
seem that the authors have put too narrow a limit upon the
distribution of these languages.
An
author of great repute, Mr. B. H. Hodgson, has shown
that there exists an intimate relation between the languages
spoken by the aborigines of North-Eastern India, and again
between those languages on the one hand, and the Dravidian
and Tibetan languages on the other hand.f An examination
of the vocabularies of more than eighty Indian Non-Aryan
languages, givenby Hodgson, shows this relation very clearly.
Another
great authority upon Anthropology, Professor
Virchow, came to the conclusion that there existed even some
ultimate relationship between the Veddas of Ceylon, the
Tamil of South India and the Sakai, and though Virchow
arrived at this conclusion upon purely anthropological grounds,
it is certain that the results of linguistic researches carry him

out.
We must therefore come to the conclusion that a con-
siderable portion of the original Negroid or Proto-Turanian
race settled in India, and that these Semang spread over a
ion>iderable part of that Peninsula. At a later period they
were followed by the Hamitic race, the Sakai, who were
perhaps then already mixed with a Jakun or Proto-Mongolian
element. In India the Semang element became the Kolarian,

*I am leaving the real Negro tribes of Africa outside this essay, simply be-
cause otherwise my subject gets too vast to be treated in this essay. But I
believe that the deeper we study the question, the more we must come to the
conclusion that the lines between Negro and Bantu are far too sharply drawn
at present, and that in reality itis very difficult to find the Borderland between
these two. Naturally their languages have had an absolutely different
development, just as the races had a completely different history.
valuable work has been collected in two volumes, entitled
" fHodgson's
.Miscellaneous Essays relating to Indian subjects," which form part of
Triibner's Oriental Series. In some matter Hodgson perhaps goes too far,
l)ii t there can
scarcely be any doubt that in the main his views are correct,
and deserve more notice than has been taken of them till now. Although I
have by other works checked most of his vocabularies, I must here state that
without his labours as a guide, I could scarcely have brought mine to a
successful issue. Another valuable little work is Forbes' "Comparative
^Grammar of the Languages of Further India" (Allen & Co. London, 1881).
:
 4-'

and the Sakai-Jakun developed into the true Hraridian. But

the- races became mixed in
many instances, and tnl
which contained both element-. The most important of
these tribes were undoubtedly whom we find the
f
i

remnants in that part of the population of India now known

as the Nagas, and who principally inhabit that hilly country
of the Central Provinces, known as Xagfutr. To th.-> A
I shall have to refer somewhat later at considerable length,
for there cannot be the slightest doubt, but the ancient A
arc the direct ancestors of the first Bantu invaders <>/ South Africa,
I shall now adduce direct linguistic evidence in support of
this assertion, and in doing so I shall not only make UM- "1
the principal Naga-languages in their present form, but also
of many of the other Non-Aryan languages of India. l-'oi

in the ancient times when the Bantu departed from India

the differences between the dialects of the various aboriginal
tribes inhabiting India were not as clear-cut as at present,
and it is evident that elements not now found among
the Naga-languages, but in the more or less related lan-
guages, must have formed part of the parent stem, and
were thus introduced into ancient Bantu. Practically,
the list that I am drawing up in the following pages, is the
result of a comparison of about sixty African languages with
more than eighty Indian dialects. It has, I need scarcely
say, entailed an immense amount of work, but I considered
it advisable, under the circumstances, to travel over as wide
a range as possible, so as to be certain of the results. On the
other hand the limits of this essay have compelled me to keep
the list within certain bounds. Hence I only give here twelve
words, but these are surely enough for practical purposes.
In drawing up these lists I have, as far as possible, traced
the words back to their Semang or Sakai originals. The fact
that this could not be done in all cases, is principally due to
want of material. The only list of Sakai and Semang words
at my disposal was that found as an appendix to Skeat and
"
Blagden's Pagan Races of the Malay Peninsula," which,
though fairly full, is by no means complete. Unfortunately,
I was unable to obtain either a Andamese or a Nicobarese
dictionary. As regards the Indian languages, I have princi-
pally made use of Hodgson's Comparative Vocabularies, and
of a small number of other Vocabularies and Dictionaries,
which I was able to obtain with the limited means at my dis-
posal.
Where I considered explanations or remarks necessary I

have done so in a separate column, in which I have also

shown the related words in North- Altaic.
43
 44
~
r
 45

w f ~
:

333-.S33.H
"3 ^
,
rt __,
si

g D 0) en
-- r: o u c
O ) J3 4>

II :
"

._
^ -^ ^:

S O>
{(j _;
47
 1
~r
:

:.3,o
~ba

3-3
^5 u
.2 a
'

2
j s
. 73
u

u
rt .- !?
p o
BO 5 9
c e a

kfc
o

- M
-i; rt rt ? '? c rt -0 S
3 3 3 >,3 3 X>>ScS P.g
O.CX c s c c c c o> c c E 5 fc
 49
5-SJJsi5T5'^2 II S nj
 X
CU
'

c3

- rt
3
M .
O
in
_-, "
3
1>S
X O -H
r;
11 " C
' ~
,

r _= .-2

*, "e "c * - i ^'TT

 53
C :
_
x
'

: .- i = =
w ^ 4> 3
,

l>
5 eo f> 5 o e
c >
^ajp.mrt'~'-n ~>c
e5 =s

. '-

"* - -
; aj
rt CJ >>~ o - -
* _?! C -
X: rf 'C > C

>.

s?
si
R
.' T
[3

^ 2. .-<---- - 35

- :-
=

rt
rt 5f-

^ 11

oc
C
W
 8

~ ^
3 .i
CL - c cj

H|J M
"" ia
P
g 2 c-s o 8,1
O O O C
> u> J5
g e ^
j
if ii ii ii Ii"
! I
C~
.1 "S.

*
5 ^s
5*0 >"
<
J3 I

^c o meri

-oS S o
o C
G C ao

3
rt
g 3
re

tTa G c"

3
i '35
S S

I
a
1
rt

c3 -I
11

o o o _ -1-3 rt
g J5
ass||s2
ooo o 2^ -
re C g ^
C g C C
Q d S t cd
g g o 5 .is o g rt rt ^J bc^e ^ g
3 3 C C 5 = ~ J< ^ O O rt .-e oc S

(4
ri i
Q QJ

it 5
>>_ rt
3 S S

M *a<?>

&o

H
 SECTION VII.

Thus far I have established -prima facie evidence of the fact

that originally the Bantu came from Further India, but that
the two races, the Sakai and Semang, having amalgamated
to a certain extent, there arose a new race, the Ugro- Altaic or
Turano-Hamitic (as it really should be called) and that the
Bantu form the southern branch of that race. That is, at all
events, the clear tendency of the linguistic evidence brought
forward by me.
There is other evidence in favour of this theory, and fairly
strong evidence, but I must leave the consideration thereof
to a later stage of this work. Not to disturb the philological
context of this essay, I must now first discuss the question of
the second Bantu invasion, which, as I said, I consider to have
come from the mouths of the Tigris and Euphrates.
I have in the first list, Nos.
27-70, given a considerable num-
ber of Sumerian words, which seem to be immediately con-
nected with Bantu expressions having the same meaning. In
the comparative list of Bantu, Indian and Malacca languages,
it will be, however, seen that Sumerian, as an
Ugro-Altaic lan-
guage, is derived from the same original stock as Bantu. It

might thus be argued that the Sumerian and the Bantu words
are similar, merely on account of having a common origin.
But there are certain facts which show that the connection
between Sumerian and Bantu must have been more close and
more direct, than would have been the case if only influences
of common origin had been at work.
History teaches us that the various Sumerian and Accadian
communities were conquered at a fairty early date by a Semitic
race, which is known as the Babylonians. This Semitic race
adopted, to a large extent, the civilisation of the conquered,
which was certainly higher than their own, and, among other
things, took over the Sumerian form of writing known
as the Cuneiform or Wedge-writing. This latter fact
seems to have been of considerable influence upon the Semitic
language of the Babylonians, as well as upon the Sumerian
language itself, which latter took over Semitic words in its
vocabulary.* In the course of time the old Babylonian Empire
came under the influence of the later Assyrian kingdom, which
was also of Semitic origin, and gradually there seems to have
collected near the mouths of the Tigris and Euphrates a some-
what mixed population, known in Assyrian history .is the
"
men of the sea lands." These people probably spoke a
Sumerian dialect, strongly interspersed with Semitic, not only
Babylonian, but also Southern Arabic.
"That this is a tact is well known t<> .ill student.^ oi \>-,\n.in .tn.l Snim-n.m.
but the principle has certainly been i mir<l too t. ii m /. l>. /'
M M.
itcn.il>
"
for a Sumerian Lexicon (Leipzig. i<x><>). ThU i> due to tin- t.u that 1'rince
t

has not grasped the idea that both in Sunu-nan ami in Semi tic \\c have a
strong Turanian element.
 5"

There very Bantu scholar knows, a very strong

Semitic element in several of the Bantu languages, and a
strong Semitic physical element, which cannot be of purely
Arabian origin, is very noticeable in some of the Bantu tribes
living in the neighbourhood of the River Niger in West Africa.
From a philological point of view, the Semitic influence upon
Bantu can partly be traced to Arabic, and this is particularly
noticeable in Suaheli. This Arabic element is, to a very large
extent, of later origin, and is probably due to the fact that since
the 8th century after Christ large Arabian colonies and king-
doms were established along the East Coast of Africa, from
Cape Guardafui to Cape Delgado. These Arabian kingdoms
remained in existence till far in the I7th century, and even to
this day, this Arabian element makes itself felt through the
commerce which Arabians carry on not only along the coast of
Africa, but also in the very centre of the Continent.
There is, however, another and older Semitic element in
some Bantu languages, which cannot be referred to Arabic
and which, I am convinced, is of Assyrian origin. The most
striking instance of this is found in the Shuna language, spoken
by the Mashonas* or, as they call themselves, the Makalanga.
A very fair Shuna language has been pub-
dictionary of this
lished by" Mr. W. A. London Missionary Society,
Elliot, of the
"
entitled Dictionary of the Tebele and Shuna Languages
(London, 1897), of which I have made use.
The following short comparative list will show how strong a
relation apparently exists between the Assyrian and the Shuna
language, and in some instances the relationship is so remark-
able as to make a direct Assyrian influence, as it were, a sine
qua non.
Shuna. Assyrian.

i. pishana =a severe pain, pishannu =to cause trouble.

2. dzudzura = to paint. zurah = to shine,

"
(as causative, to make shin-
ing.")
3. sima =to plant. simnm =a plant.

*The name Mashona has been objected to by several scientists, who prefer
to call the tribe the Makalanga. This latter name is supposed to mean " the
People of the Sun," Ma = people, ka=ot, langa (for ilana) =sun. I strongly
doubt whether this is not a case of Bantu popular etymology. As far as I am
aware, the preposition of in Bantu does not enter into any tribal name of the
Bantu. The Makalana are probably an offshoot of another tribe found
more north, and called the Makaranga (rl), and this makes me inclined to
"
think that the real meaning is the people of Karan," which might have quite
another explanation. Why the term Mashona should be objected to, I can-
"
not quite see. The expression means the people who hiss." a name which
was probably given them by some other Bantu tribe, on account of the large
amount of sh and other sibilants in the Shuna language, a factor which
certainly distinguishes Shuna from other Bantu languages.
 Shuna. Assyrian.
4. msimba = power. sibu = to grasp firmly.

5. bisu = to produce. basu =to be.

"
(as causative, to produce.")

6. zwara = to propagate, zan. = to propagate.

7. roba =to punish. ra'abu = to be angry.

8.
idanga =a cattle kraal, dunnu =a bed (really
"
a place to rest in ").

9. shinga =to labour. sakanu = to work, make.

10. bona = to look. baru = to see.

(N.B. Semitic r generall}- changes in Bantu into n, as most
Bantu languages have not the true r.)
11. penga = to be mad. pinga =to be mad?*

12. igona = plank. kannu = plank ?t

13. igondoro =goatram. gadu = ram.

*

14. nzhira =road. girru = road.

(The Assyrian guttural is here softened in Shuna int
sibilant, a common occurrence.)
15. berenga =to read. berti = sight.
(Tebele bala = to read.) baru = to see.*
16. pira = to sacrifice. pirketi = a kind of sacrifice.

17. ishangu =shoe. semi = shoe.

"
*Delitzsch give> in hi*
Assyrian Handworterbuch." p. ;:. the word pinga :

without the translation, but it seems to me from the context of IV. K. S, that
"
the translation to be. mad," would lit in very well. If this be correct, we
have here an instance of the Shuna language enabling u.* to inul the meaning
of an Assyrian word, and I believe we might meet with more case> <>t a Minilar
nature.
" "
fDelitzsch does not translate this word, but the meaning plank wnuld
seem to fit in, though in one extract quoted by him it might have the rel.ited
meaning of "sheet (of "iron)." The Sumerian is-fatt-M-MA would certainly
support "the translation plank." Compare i^-^n-.fiii^n^iii n beam (J.i
kosen = beam," Sirjenian gozna -beam ).

*These words, whit li, .is tar u -un ,i\\ UK arc n->t touii'l in aiu other Bantu
I

language, must be very old, as the later Bantu lost the art ot rea-lmg prob.ibly
soon after his arrival in Alnr.i. I'.ut in Sumir and Habyloma ncarlx
8V*Ty "in-
could read, as we know from Hilpivtht- l-.\|iloration> m HibU-lamh."
'

l-'m
this reason I venture to connect the Bantu word* with tin- Assyn .

baru. I may stat<- that it seems to me th.'t e\.-n in --.im text* bam
" '
" "
may have the meaning ofto read." 'In ni; and n-.nlmg
are nearly related, ami we often find mention ina-i -;-tiMi-> pi.-
" " "
such a manner that the people could >ee them ('. .. read them ").
18. gu-ipa = sin. egu = to sin.

19. dakwane = small, dakku = small.

20. nzimbha = tsetse llv. z.umbu fly.

21. imanda = anger. mammu = anger.

22. dabira = to answer. gabru =s>to answer.

(N.B. The change from g to d is certainly very remarkable

in this case.)

23. musana =back. seru =back.

24. ba =to be. basu = to be.

25. naku = to be beautiful, nasku = to be beautiful.

26. ishiri = bird. ishshuru =bird.

27. kondamu =to bow. kadadu = to bow.

28. mpeto = border. patu = border.

29. mflwana (pL bana) child, banu = child.

(N.B. The vShuna singular is a contraction of mun-icana,
where the w is the regular representative of the /;.)

30. buya =to come. ;

ba'u =to come.

31. tebera =to come after, tibu = to come after.

32. idama =a command. adu = a command.

33. shandura =to change. shanu = to change.

34. buraya = to destroy. balu = to destroy.

35. zwimba =to desire. shabu = to desire.

36. buruka =to descend. burn = descendant.

37. imauri =a deaf man. amiru =deaf.

38. sweta = to draw. shatu = to draw.

:
^
39. simiga =to fix. samu = to fix.

40. erera to flow. gararu = to flow.

(The has
initial g
dropped in Bantu, and the Shuna has
reverted to the older Indian form, as is clear from the
really
Dravidian ant = river).
 59

41. paradziga =to fly parashu =to fly.

(of sparks).

42. nganu = garden boun- ;

gannatu = garden boundary.
[dary.

43. ishaba = gazelle.

= to go.

= happiness.
=hole.

= to hunt.
= to lock.

= to lock.

=love (n).

= to part.

= \vant (n).

= wind.
= wood.

=god.

(Perhaps this isone of the strongest proofs concerning the

relationship of Shuna and Assyrian. If the word was the only
Semitic word in the first-named language, one might feel in-
clined to ascribe it to Arabic influence, but in this case this
can scarcely be done, even admitting that, at a certain time
in their history, the Mashonas came into close contact with the
Arabians, which is highly probable. But except in a few
tribes who have turned Mahomedan, the word Allah, for dod.
is nowhere found in Bantu dialects, and even Sualieli. which
shows very marked influences of Arabic in its vocabulary, has
retained the word Mnniiti for the Deity. Under such circum-
stances, and taking into consideration tin- other evideno - re-
garding the relation of the two languages. am convinced that
I

Mu-ali must be referred to Assyrian origin).

 6o

Delitzsch is of opinion that the Assyrian word is really an

adaptation of the third personal pronoun = he, but I am not
so sure of this, and both words might be referred to Semang
edjah= chief, the Assyrian having adopted the word from the
Sumerian. But in any case. the Shuna seems to be the only
Bantu language which has taken the word over, unless Dualla
"
sanga = father, master," must be referred to the same root,
which is very doubtful.

57. ^huna msafuri copper. Assyrian siparru= copper.

The Shuna form is here really the older, and the history of
the word is so remarkable and interesting that I shall give here
" "
a short account of it. The original word for copper was
"
chu-phar, a Turanian word, meaning beyond the sea," or
"
from the other side of the sea," from Turanian chu = water,
sea, and phar = beyond. This name was probably given to it
by the Turanian tribes of Western Asia, because they obtained
their copper from India across the sea. The Sumerians called
"
it ud-ka-bar, i.e., the metal from beyond the sea," ka-bar
being the equivalent of the original chu-phar, with this differ-
ence that in ancient Turanian the ch was soft, and the word
was pronounced su-p-har, the ph being a real aspirated p. In
this way two words for copper arose, one with the soft ch and
the other with the hard ch or k. The Semites followed the
soft pronunciation, and hence we have in Assyrian siparru, in
Arabian asfar (brass), etc. and this pronunciation was carried
;

over by the Sumero-Babylonians to the Bantu, when the

language of the former had already absorbed considerable
Semitic elements.
But the ancient Sumerians (before corning into contact with
the Semites) went over to the island of Cyprus, under the
leadership of their great conqueror, Sargon I., who ruled about
3700 B.C. In all probability they formed a settlement on the
island, a fact which is patent from the old Sumerian and
Babylonian cylinders and seals found on that island. They
found rich copper mines on the island, and called it kabar or
kit- par, and hence the Greeks, when they became acquainted
" "
with the island, called it kupross, whence our name copper
for the metal in question. So it happens that to-day we call
" "
it copper and the Shuna call it msafura, a slightly changed
form of sipparu, and up till now there probably was neither
7

an Englishman nor a Mashona who was aware that in reality

they are using the same word under different aspects of
phonology.
To prove how correct this explanation is I may point to
"
Arabian nuhas and Hebrew nehast, both meaning copper."
Both these words are derived from the Indian Nagas, the name
of the people who brought the copper from India to Western
Asia, and who, as I shall show, were the ancient Ugro- Altaic
 6i

traders,who lived near the mouth of the Indus. This is shown

by the Ethiopian nahas = copper, and as in that language g
and h interchange, there is every probability that the old word
for copper was nagas.
There may be some people who are sure to stamp this kind
of thing Philological Romance, but these should not forget
"
that often facts are stranger than fiction."
Some of the ancient Assyrian words were taken over in
Bantu with metathesis of syllables. So :

Shuna. Assyrian.
tjero = ghost. ru-chu = ghost.
tizha = to run away. satu = to run away.
ma-kusero= skin scrapings. karasu=to skin,
di-muga = to be sorry. agamu = to be sorry.
There is, besides this apparently direct connection between
many words in Shuna and Assyrian, a good deal more evidence
of Assyrian influence upon Bantu, especially in that branch
of Bantu which we call the Bechuana. For instance, there
was a chief among the Barolong called Tau = the lion,
and we have also the tribe of the Ba-taung (=the
people of the place of the lion). But Tau is nothing
else but the Assyrian tau = to eat, and we have thus here the
"
lion called the eater," a name he also had in Semitic, as is
clear from Judges 15: 14. "Out of the eater (i.e., the lion)
came forth meat." (See Ftirst, Hebrew Dictionary, sub vocc
akal = to eat.)
So the name of the South African supposed desert known
to us as the Kalihari, is nothing else but the somewhat changed
"
compound of Assyrian kalu = all," and harbu (Sumerian
aria)
='' desert," and may thus suitably be translated as
""
great desert."
In the other Southern Bantu tribes there arc undoubtedly
several words connected with Assyrian. Zulu baba = to burn,
seems to agree with Assyrian kababii = to burn, and Zulu zi =
kraal, with Assyrian zirru = ience. But there are /w< words
in Southern Bantu which are without the slightest doubt of
Sumerian and Assyrian origin.
The first is name given by the Zulus and
unkulukulu, the
Kafirs to a being, which was apparently a high deity, but of
which the Bantu have not any more rlear e. meeptinn. Uishop
,1

Tulluway has in vain tried to get at the bottom

"
of thi> matter.
but he could not get further than the idea of the great-i^n at-
one," or "the old-old one," which tin- Bantu them-' -1\<
present possess of this mysterious bcin.u, and who they even
believe to have been a man once upon a time. Authorities
like Callaway and Max Miiller have therefore eome to tin-
"
conclusion that we have here to do with a case of amv-t<>!

worship." But if the Zulu were able to remember his ane--ir, ( !

home, he would have no difficulty in connecting Unkulukulu
with the old Sumerian An-gal-gal, the Great God of Heaven,
with which it agrees letter for letter. Gal in Sumerian means
"
great," and is exactly the same word as Zulu ktdu, which
in some of the Bantu dialects, e.g., Secwana is goto. The
repetition merely gives emphasis to the adjective. An in
"
Sumerian meant the God of Heaven," and in Zulu would
"
become un, the u being, so to say, the " leading vowel in
Southern Bantu and often taking the place of an original
a. By such a very simple process of Comparative Philology
we are able to settle a matter which has considerably troubled
missionaries as well as Bantu scholars, including the great
Dr. Bleek himself.
The second word I wish to refer to here, is the well-known
ama-pakati. In Zulu and Kafir this word denotes the coun-
cillors of the chief, and the word is generally explained as
"
meaning the middle-men," from the adverb ^>fl&flfr'= middle.
These ama-pakati are, in a certain sense, the sub-chiefs of
the districts in which they reside, and there represent the
great chief. It is their duty to watch over the behaviour
of the people, and to hear minor cases to collect the fines
;

imposed upon evildoers and they are also responsible for

;

"
the due payments of the presents," to which the tribal
" "
chief is entitled, and as these presents are really a mere
" "
euphonism for taxes they may be said to be the chief tax-
collectors.
In Assyrian, paliatu (pachatu) meant a district or division
of the kingdom, and the governors or satraps who ruled these
divisions in the name of the king were called the Amel-pachati .

which is, letter for letter, the same as the Bantu ama-pakati .*
Delitzsch derives this word from the verb pichu, which means
"
to impose taxes, to rule," and this is probably correct.
It is therefore likely that the adverb pakati =
middle, is derived
from the noun, and not vice-versa, and this is more probable,
because in Bantu most adverbial expressions are really special
forms of nouns.
Considering therefore the presence of a distinct Assyrian.
or rather Babylonian element in several Bantu dialects, we
are entitled to come to the conclusion that the Sumerian-
Babylonian influence on Bantu must have proceeded directly
from the mouths of the Tigris and Euphrates, and that by
means of an immigration from that part of Asia.f

"There exists in Assyrian a parallel form with k, viz.. pakadu. with a primary
" "
meaning to protect," and a secondary meaning of to rule, to govern."
" "
fl am using the phrase the mouths of the Tigris and Euphrates purposely,
because in ancient times the two rivers had their separate embouchures, and
did not join, at as present, in the Schatt-el-Arab. The waters of the Persian
Gulf are constantly receding and it has been calculated that in 6,000 years the
land at the mouth of these rivers has gained more than 50 miles upon the sea.
1
 63

SECTION VIII.

We shall now consider some other evidences regarding the

"
Origin of the Bantu, for, as has been said the relations
between races of mankind cannot be determined by philology
alone." This is a truism which anthropologists are continually
impressing upon philologists, and in a measure they are
correct in doing so. Anthropology, however, would no doubt
attempt to settle the question by the shape, capacity, etc.,
of the skull, a question I am not going to enter upon, because
I know too little
upon the subject. But it is to be hoped
that some competent authority will soon take this matter up,
upon the lines suggested by me in a former*part of this essay.
Comparative religion is another science which demands
attention, when we wish to inquire into the origin of races,
and it is
undoubtedly of immense importance.
Unfortunately
the Bantu's religious ideas seem rather vague. We have
already mentioned that Bishop Callaway's attempts to get
at the core of the Zulu's idea of a God, met with but scant
success, and attempts made by other missionaries among
various Bantu tribes have not given much better results.
"
It has often been said that the Bantu have Ancestor-
worship," but I doubt whether this is correct in the main.
That he worships the ancestors of his chiefs, or rather the
" "
spirits of departed chiefs, is undoubtedly true, but I

question whether a Zulu or Kosa would worship the spirit of

his great-grandfather. It is true that when a certain snake
enters a hut, he is left in possession and not harmed in any
" "
manner, because it may contain the spirit of So-and-So,
probably one who lived formerly on or near the spot "
where
"
the hut stands. But, as I am not aware that the spirits
are ever considered to be embodied in any other animal than
snakes,* I am inclined to call this Serpent-worship. This
certainly would agree with the Indian origin of the Bantu,
a matter I shall refer to somewhat later.
" "
The various Bantu names for God are very dangerous
to use, except in certain clear instances. In some dictionaries
the Bantu word for God looks suspiciously like a word coined
for the purposes of religion and Bible-translation by the mis-
sionaries. Such is, for instance, the Luganda word Katonda,
"
which means the Creator," from the verb tonda=to make, to
en-ate." But some Bantu words for the idea God are certainly
capable of explanation. The Secwana Moditno. for instance,
reminds us strongly of the .Wvria'n Dimmcm. in Sunn-riaii
" "
*The case oi the Siboko, or animal worshiped >r .l.uuv.l t<>. > omnium
a> thi^ i. cvulcniU .1 i<-lu .i
among the Bechuana tribes, does not apply lu r.
ancient Tribal <>r family T< item /SHI. winch has m reality nothing t<- ilo with

religion.
 "4

An-ki-a, the God Heaven and Earth. Mulungu. the Mang'-

of
anja expression, most
is probably connected with the old Bantu
langa =high, with which the Zulu ilanga = sun, is also connected.
"
I at first, indeed, felt inclined to consider the word as a per-
"
sonification of the Sun, but came back from it, through the
fact that, as far as I am aware, there is no trace of Sun-worship
in any Bantu tribe, and the name for the Sun in most Bantu
tribes seems to be connected with an old Turanian root ju or
"
shu, meaning warm." In Oruhima it is izuba, in Suahelr
jua, in Herero e-yuva, in Secwana tsatsi, in Mang'anja
dzuwa, in Luganda juba, etc.
"
Among all Bantu we
find a great respect for
tribes spirits,"
which are generally inimical to mankind, and this form of be-
lief, with all its attendant institutions, such as medicine-men
and witch doctors, is still found in the Sakai of Malacca. It is
in reality the same as that Shamanism found all over Asia, in
places not touched by Brahmaism, Mahomedanism, or Chris-
tianity. It was the old Bon religion of Tibet, and it still plays
a not unimportant part in Northern Buddhism or Lamaism.*
"
Among a large number of Bantu tribes the Spirits under
the Water," called in Zulu and Kafir the Izizi, play an impor-
tant part. There can be no doubt but this is a very old Ugro-
Altaic belief.
" Among the ancient Sumerians the great God is
Hea, the God of the Sea," who may, perhaps, be the same as
the much-revered Ahti of the Finns. Indeed, among the Finns
the water-deities and water-spirits play a very important part,
"
and their Wesi-Husi, by which they denote the evil spirits of
the waters," are not only etymologically connected with the
Zulu Izizi, but bear also the same character. The Southern
Bantu's idea seems to be that these Izizi are really evil spirits,
and that is proved by the fact that sacrifices of an ox are made
to them, in cases of drowning, to induce them to release the
person taken by them. (See the well-known tale of the girl,
"
given by Dr. Theal in his Kafir-Folklore.")
The Comparative Folklore of the Sakai, Semang, Indian
tribes and Bantu cannot be treated herein, as little as a com-
parison of the social customs. Each of these subjects contains
enough material to write a large-sized book on, and cannot be
discussed within the limits of this essay. But it is necessary
to mention here certain facts of historical nature.
In the first volume of his edition of Herodotus, p. 650, Raw-
linson states that in ancient times a Cushite or Ethiopian race
extended itself along the shores of the Southern Ocean from

Abyssinia to India. It started from the Indus along the sea-

coast through Beluchistarr and Kerman, which latter (as Raw-
linson thinks) were the true countries of the Asiatic Ethiopians.

*The Semang. on the other hand, adore a Thunder-God (Kari). and of this
seem to be considerable traces among the Hottentot as
religion there well as
among several Bantu tribes.
 65

That the Asiatic Ethiopians of Herodotus did not start there,

but in the Peninsula of Malacca, or in some country south of it,
I have tried to show, and I have also
attempted to show that
the Semang and after them the Sakai moved to India. There
is, as Caldwell has shown, proof that the present Dravidian

races (which are nearest the Sakai and may have been caused
by. a fusion of Jakun and Sakai) lived in ancient times in
Northern India, when Southern India was inhabited by what
is called a Kolarian race, but which was probably a Semang

race, and by an old diminutive race of which the Veddahs of

Ceylon are the remains.
But Semang and Sakai mixed, and perhaps both even mixed
with the Jakun, and a new race arose. What that race called
itself we do not know, but when the Aryan entered India he
called these people the Nagas, or the Snakes. How this name
arose is very difficult to tell. It may be that this race had

developed a Serpent-worship, the traces of which are still found

among several aboriginal tribes of the Malayan Peninsula. It
may be that these people called themselves with some word
resembling Naga, and which the Aryans took over and made
Naga of, because these people proved their worst enemies.*
The Tibetans called these people the Klu (pronounced lu\,
and this probably because the Nagas, with whom they were
acquainted, were living along the shores of the Indus and its
tributaries, for 'the Klu (as well as the Naga) were supposed to
"
live in the rivers, and in Tibetan klun is a river or stream."!
At the time of the invasion of India the Nagas seem to have
principally lived along the Indus, and at the mouth of this
river. The researches of Sir Thomas Holdrich and of Surgeon-
"
Major C. F. Oldham (see his The Sun and the Serpent,"
London, 1904) have shown that in very early times the race
in the neighbourhood of the mouths of the Indus was not only

highly civilised, but that it was a great sea-faring race,

which founded several colonies on the West Coast of India,
even as far as Malabar and Ceylon. Cunningham (quoted in
"
Sir H. M. Elliot's Races of the North-Western Provinces of
India," vol. I., p. 113) connects these Nagas with the Takkas
in the Punjaub, and thus places them between the Jlu'luni ;md
the Indus, and Beames (who edited Elliot's work) places th<-m
even as far north as Kashmir.
"
"That seems to be the opinion of Lefmann in his Geschichte cles alien
Indiens," page 366 (Berlin, 1800). In Atnuntus,' or Mn. which, as already
said, is very nearly related to Semanq, ngu'o't is man. and the Kaqas may li.ivc.
as several nations have done, simply called themselves nii-n. \<H;<I in tin-
" "
meaning of snake is certainly not Aryan in ..nyin. Inn is tound in Siamese.
and in certain non-Aryan languages in India. The true \r\an li-r -nakr is
Sarpa.
India, and in
fThe name klun is still that of one of the non-Aryan tril-i-s >t

Burmese, Sak and other languages, lu (which is tin- modern tonn ot kl'.i. and
means In Sum. nan
agrees with the pronunciation of the Tibetan word)
m.in.
too lit means man, and the compound a*za-lu-ln is used m the sense of mankind.

[G. 171907-] '

 66

There are certain expressions in Tibetan regarding these

Nagas or Lu, which strongly seem to bear out Cunningham's and
Beames' ideas. The Tibetans consider the Lu as a kind of
demigods, having human heads and the body of a serpent.*
They were believed to be the guardians of great treasures
underground, a myth which clearly refers to their being mer-
chants and traders, who brought treasures from the low coun-
try to the high plateau of Tibet. This is, in my opinion,
clearly proved by an "
old Tibetan Festival, called the Lu-theb,
which is denned as the coming upwards of the Lu from their
retreats in summer," while another expression Lu-dog, means
"
the retiring of the Lu to their abodes in the nether regions,"
and this was fixed at the beginning of winter. Now this clearly
applies to the traders along the banks of the Indus, who each
summer (the only time when the high mountain passes could
be crossed) went to Ladakh and neighbouring countries with
their wares, and returned home before the winter had covered
the passes with snow.
The name or syllable lu appears in the designations of a large
number of nations and places in Western Asia, and in my
opinion this proves how far this race, i.e., the Ugro- Altaic race,
once spread. In the Bible we find the Ludi as the name for a
race in Asia Minor, where we also have the Graecisized forms
of Lydia and Lykia.^
We further have the* district of Luristan, in Persia, which
means really " the dwelling of the mountain Lu," ri being a
common Turanian name for mountain.' ' '

This brings us to the word Luganda in Africa. U-ganda is

the name of the country at present the people are called
;

Bagunda (singular Mogunda), and the language is called

Luganda. But the prefix lu, as the prefix for a language in
Bantu, is absolutely unknown, and in accordance with Bantu
etymology, the language should be called Stganda or Seganda.
Under such circumstances I am convinced that we have to
do here with a confusion of ideas and expressions, of which
there are many other examples in the Bantu dialects. I am

of opinion that the original name of the country was Luganda.

If so, the etymology of the word is perfectly clear, viz., from
Lu (the name of the old Nagas) and gan, the very common
"
word in Turanian and Altaic for garden," but also meaning
* Actual
representations of such Nazas are found in Indian sculpture. Is
it not possible that the Nagas, sailing as they did in boats and ships (which
glided over the water as a serpent on the land) were therefore represented with
bodies of a serpent ? In the Sumerian story of Oannes, who came to the
mouths of the Tigris and Euphrates in a ship, we find the same connection of
the head of a man and the body of a fish.
fBoth Lydia and Lykia can easily be explained. Lydia was Ln-de = the
Lu-community from Tibetan (Turanian) de-race, tribe, community. Lykia
would be a compound of Lu and the Sumerian (i.e., Ugro-Altaic) Az =land,and
"
thus mean the land of the Lu."
 67

" " "

land." In Sumerian gan is a field," and gana is a gar-
den." The word is, on that account, also found in Semitic.
Thus Gan-Edin, the garden of Paradise, really means " the
Garden in the Plain," Assyrian Gan-Edinnu In the Luganda
language we find the same root (with k=g) in eki-kan-de, which
"
means a deserted garden."
There is another word of importance in the Luganda language
in connection with this matter, and that is olulimi= language.
This is evidently a combination of o (the article) lu= people
(the Lu) and emc = voice. (Compare Mongolian ama, and in
other Altaic languages imi = voice.) The second / is merely
euphonic, in accordance with a false concord.*
Now it is an undoubted fact that in Luganda we have a very
old Bantu language, which has clearer affinities with the Man-
chu branch of Ugro-Altaic than probably any other known
Bantu dialect. It is also a fact, proved by the traditions of
several Bantu tribes, that the regions near Lake Victoria
Nyanza was the ancient and central home of the Bantu.
But how did the first Bantu get there ? This is a question
which at present it is very difficult to answer, but after having
taken all facts into consideration, I am inclined to think that
the Bantu came to Africa by sea and not overland. Tin-
Hottentot may have come to Africa as a Semang overland,
and he was probably followed by Sakai tribes who became the
ancestors of the Gallas, Somali and other Hamitic tribes of
North Africa. But the earliest Bantu must have come much
later. At all events it certainly is remarkable that on the
Egyptian monuments of early date we find no picture of a
Bantu, nor do we find any description of Bantu tribes, f
In the second volume of his Herodotus, Rawlinson mentions'
an invasion into Africa of Asiatic Ethiopians about the year*
1300 B.C. Unfortunately I have not been able to trace hi?
authority, but upon the supposition that with Asiatic Ethiopia
he means the country which the Greeks called Gedrosia and we
Beluchistan (which is clearly the Eastern Ethiopia of Hero-
dotus) I am inclined to believe that we have here a trace of the
,

first Bantu invasion.

conquest of the northern
It is fairly certain that after the

part of India by the Aryans, that part of the population which

we have denoted as Klu or Nagas was dispersed. A section
of them sought refuge in the jungle regions, now known as
Nagpur, a hilly and nearly inaccessible region, where even to-
day we find a number of tribes which bear the name of Nagas. "
"
*It is remarkable that the Luganda n;imr t<>r a gardener omtam-- appar-
ently the same root, as it also is omu-limi. But here tin- Ihni is quite another
word, and is related to the Zulu lima = t cultivate, a \\nnl that is touiid in a
number of Bantu dialects'.

fDr. G. M. Theal. however, has informed me that m the

British Mu.seum he
has seen some relief work, which clearly represents Bantu men and \\.unen.
As, however, he does not know lioin uhat period the representation dated,
this leaves the matter as far as it was.
i j
 68

and whose languages bear marked relations to the Bantu

dialects. Others sought refuge in the hills south of the
Himalaya, from Nepaul to Assam. Again others, and pro-
bably those who were living in the Indus valley, crossed the
mountains to the west and settled in Beluchistan, where a
part of them continued their seafaring life, and another part
became nomadic cattle-farmers. These were the ancestors of
the present Brahui, whose name is probably derived from the
old Turanian (Tibetan) ba-hjo = herdsman with the often
,

inserted r. In Tibetan ba-glan-spyod is the name of a country,

west of India, so-called because cows feed on the land, and form
the main wealth of the inhabitants. This I take to mean the
northern portion of Beluchistan.
The Brahui language is considered by competent authorities
as belonging to the Dravidian language group, but it certainly
has many Ugro-Altaic expressions in it. So has the Baluchi
language, and so has Sindhi, the language spoken along the
lower course of the Indus. Sindhi has, like Baluchi, a strong
Aryan element, and the former language is rightly considered
as belonging to the Aryan group, though it certainly contains
elements which are non-Aryan. In ancient times the number
of these non-Aryan elements was undoubtedly larger than at
present. Now it certainly is remarkable that several old
Bantu words find their equivalents in these three languages, as
the following examples will show :

(1) Bantu ngombe, the general name for the ox, is a pure
non-Aryan word, found in Khamti and Laos respectively as
ngo and ngoa, and in Kuswar, Pakhya, Dadhi and Denwar as
gai. In Sindhi it is guan.*
(2) Luganda" (one of the oldest of the Bantu languages) has
"
for bullock endawo. Oshikuanjama and Suaheli have for
male animals ndume. In Sindhi we find the related dana =
bullock.
(3) Mashona gara = to inhabit, clearly agrees with Sindhi
ghar = a house, which found in several of the non-Aryan lan-
is

guages of India. Dadhi and Denwar have ghar Kuswar has ;

ghara.
(4) Mashona bhati
= plank, which appears in Kongo as cbaya
and ebandu, and in Luganda in olu-bawo is clearly the Sindhi
patt plank.
(5)The Zulu-Kafir amasi = sour milk is clearly the Baluchi
niass=curd butter, or better, milk mixed with buttermilk.
"
(6) In Brahui the name for Sorghum rulgarc,
our kafir-
corn," is juar, and this name is fairly common throughout
India for the plant. Does not this agree with juala, the Zulu-
Kafir name of the beer, made specially from kafir-corn ?

*.\lthough at present found in India in several non- Aryan languages, it is

not unlikely that the root go (gont or gor) has been loaned from the Aryans,
who probably introduced the domesticated ox into India.
 (7) Mashona run-zhi = a. needle (where run is a prefix) is con-
nected with Sindhi sui = a. needle. In Permian si is " thread,"
and the Finnish sui has the same meaning. We have thus here
the very same substitution which we find in some of the Bantu
dialects. So Zulu has usun%ulu for needle, while Kafir has
usinga for thread.
Many more examples might be given. So Pushtu, or the
language of the Afghans, which contains many old Ugro-
Altaic words among its mixed vocabulary, has khaza for wo-
man, which agrees with the Bantu well-known kazi.
On these grounds I feel inclined to believe that the first in-
vasion of Bantu in Africa came from some country near the
mouth of the Indus, or a little to the west of it, and this would
agree with the account given by Rawlinson. These probably
landed in the present Somali-land (the ancient Punt] and
driving the old Hottentot population before them, made their
way to the neighbourhood of Lake Victoria Nyanza, where a
portion of them settled, while another portion straightway
worked their way to the west, where they became the ances-
tors of the Dualla, Mpongwe and other Bantu tribes of West
Africa.
Naturally this is a theory, and such, I am afraid, it must al-
ways remain but at all events it is a theory with some evidence
;

at the back of it, and as such it has a right to existence and to

be carefully considered.
Of the second invasion of Bantu in South Africa we have,
however, more historical and trustworthy data. I have al-
ready shown that on account of the ancient Babylonian and
Sumerian elements in some of the Bantu dialects, this invasion
must have come from the mouths of the Tigris and Euphrates.
At the mouths of these rivers many of the old Sumerians
had taken refuge, and their number had undoubtedly been
strengthened by many Semites, who for some reason or other
had considered it advisable to leave their original habita-
" "
tions. Nominally subject, these Men of the Sea Coasts
were in reality independent of Babylonia, though tin-re
never seemed to be any trouble between them and the authori-
ties. Matters changed, however, when, about the year 900
B.C., Babylonia became a dependency of Assyria, and the so-
called King of Babylonia was really an Assyrian vassal. Many
Babylonians, dissatisfied with the new regime, must have gone
to the Coast, and strengthened the independent community
which gradually was formed there. This community at l.i-t
became so strong, that it became an eminent source of danger
to the Assyrian Empire, and it would seem that the Kin^- <>t
Assyria began taking steps to stop the continual incrr.isr of
"
power of the Men of the Sea Coast." The latter, thereupon,
rose in open rebellion about 720 H.I . .mil fmnnl a capable and
daring leader in Merodach-Baladan, tin- same person mentioned
 7

in II. Kings, 20. For some time the rebels had it all their own
way, because the attention of the Assyrian King was drawn
to other, more important, matters. They conquered Babylon
and seemed to have penetrated to the very borders of Assyria
itself. Then, however, Assyria determined to subdue the foe.
For a considerable time a keen struggle went on between
Merodach-Baladan and his sons on the one side, and the Kings
of Assyria on the other. But though the Men of the Sea Coasts
were assisted by the Elamites and the tribes west of the mouth
of the Tigris, the power of Assyria at length prevailed. The
rebels were completely vanquished a huge slaughter took
;

place, and those who did not fall by the sword left the country
and sought refuge in the islands of the Persian Gulf. But
Esarhaddon, the King of Assyria, equipped a large fleet and,
attacking the rebels in the islands, succeeded in driving them
even out of this refuge.
It is more than likely that the men who were thus driven out
of their country, were aware that in or near Somaliland, on the
neighbouring coast of Africa, there were living races who were
nearly related to them. The Men of the Sea Coast were great
mariners and traders, and if we remember that the great centre
of trade in those days was the island of Socotra, it certainly
is not surmising too much, if we believe that the Sumerians
knew who lived in the country opposite Socotra. For that
reason I believe that these fugitives, of whom nothing more is
heard in Assyrian history, found a haven in Somaliland, and
proceeded to join the Bantu tribes near the lakes. Assyrian
chronicles place the flight of the Men of the Sea Coast in
680 B.C., so that this may also be considered the date when
this second Bantu invasion of Africa took place.
It would seem that in the majority of cases the two elements
coalesced, and this need not astonish us. The Sumerians
originally came from the mouth of the Indus, as is clear from
the Babylonian legend of the Deluge and the story of Gilga-
mesh, and hence there must have been a time when their fore-
fathers and the forefathers of the first Bantu invaders were
living together. The fact that there always has been, since
the very earliest times of history, a considerable trade between
the mouths of the Indus and the mouths of the Tigris and
Euphrates, must have kept old tribal traditions alive, which
would otherwise have sunk in oblivion.
It may be easily understood that the remnant of the Men of
the Sea Coast which reached Africa, was but a poor remainder
of the once powerful community. Their leaders, their priests,
and their influential men had all fallen bv the sword, and those
who ultimately survived, must have belonged to what we might
"
style the lower classes of society." Hence the fact that no-
thing of the old Babylonian or Sumerian civilisation came over
to Africa. Once in a strange country, shut out from their old
 71

home, the sea, the newcomers completely changed the tenor

of their life, and became agriculturists and cattle-raisers, like
the first Bantu had become. Gradually the distinction be-
tween the two elements wore away. But not altogether. The
Babylonian element was strong and died hard, and the old
Semitic linguistic factor was never altogether eliminated. It
is impossible to-day, indeed, to draw a sharp line of demarca-
tion between the various Bantu tribes, nor can we exactly tell
which tribes crime from the first invaders, and which tribes
trace their origin to the later invasion. But, as a general rule,
we may lay that in those tribes, in whose language we
down
find but single traces of Semitic (e.g., in Dualla, Mpongwe,
Isubu, Luganda and Herero), we may see the descendants of
the first Bantu invaders of Africa. But where, like in most of
the Bakwena languages, and especially in the language of the
Mashona, we find a very strong Semitic (not Arabian) element,
we may fairly conclude that the origin of those tribes can be
traced to the later invasion. But it should never be lost from
view, that the Bantu has become a very mixed nation, and that
at present we have but the excrudescences of the original ele-
ments, and not these elements themselves.

SECTION IX.

Those of my readers who are acquainted with the peculiar-

ities ofthe Bantu languages may perhaps say that the fore-
going paragraphs are all very nice and true, but that they
do not throw any light upon two very important matters,
namely, the Prefixes of the Bantu languages and the Concord
of the Bantu sentence.
I readily admit that these questions are of the highest
importance, and that, unless I can give a satisfactory solution
of them, my foregoing theories will not carry much weight.
For that reason I shall discuss these questions here, even at
the risk of making this essay considerably longer than I origin-
ally intended.
I am quite aware that my solutions of these problems run

completely counter to the at present generally accepted theories

of Bantu philology, but that is not my fault. Bleek, Kolbe
and, in fact, all other Bantu scholars seem to have considered
that the prefixes were the peculiar and characteristic stump
of the Bantu languages, that they originated within Bantu
itself, and that, indeed, Bantu had a philology peculiarly
its

own, widely differing from that of the other language groups.

Hence they originated a system of classifiers, and to every
singular prefix a plural prefix was assigned.
At first sight it
would seem that the spirit of the Bantu languages is really
 72

in accord with the principles thus laid down. So in Kafir the

m or first-class nouns form their plural in ba, and the or m
second-class nouns form their plural in mi. Various attempts
have been made to show that words signifying a certain class
of objects, are compelled by linguistic laws to have a special
prefix, and Bleek as well as Kolbe have displayed a great deal of
original acumen and ingenuity in attempting to show what the
form of these classifying prefixes were. It is very doubtful
in how far these attempts have been successful, and, personally,
I feel compelled to state that these attempts have done
more to complicate Bantu than anything else, while they have
tended to place Bantu philology upon an absolutely false basis.
For the idea up till now entertained that Bantu has a system
of philology peculiarly its own, is beside the mark if it were
;

it would, from a scientific point of view, imply that the Bantu

race, as a collection of human beings, had an evolution of its

own, independent from and unconnected with other human
races. In other words, there must, then, once have existed a
specific anthropoid ape, from which the Bantu race evolved.
There is nothing which entitles us to such a supposition, and
every bit of true Bantu philology pleads against that view.
for all the general rules of philology, and all the principles of
phonology, which apply to other Turanian languages are also
applicable to the Bantu languages.*
The real facts are these :

The oldest languages of the world made use of a system of

prefixes, infixes and suffixes. This has, in the case of the Mon-
Khmer language (which is one of the parent languages of
Bantu, and probably represents the Semang in its purest form),
been shown by Schmidt, as quoted by Skeat and Blagden (in
their aforementioned work, vol. II., p. 447). Thus in Khmer
we have :

kat = to cut
khnat = measure
kunat = piece
thkat = pain
tamkat = suffering
skat = to cut off
sangkat = division
pangkat = to divide
where we find that by the addition of various prefixes and
infixes different meanings are given to the root kat. .

*This naturally implies that I do not accept the so-called pronominal

theory in Bantu. Indeed, I consider the idea that the prefixes were derived
from the pronouns, to be historically impossible. Pronouns are late
developments in the history of language, and are generally shortened forms
of original personal nouns. That some original prefixes were afterwards used
as pronouns, is an assertion I could agree to the reverse supposition I can-
;

not subscribe to. Prefixes and determinatives belong to the oldest forms of
languages.
 73

In Semang and Sakai we find a similar system, though the

matter does not seem to be thoroughly investigated. (See
Skeat and Blagden, op. cit., II., p. 774.) We have in Semang
apparently a number of prefixes, such as ke, pe, ta, ba, etc., and
in Sakai also a number such as he, che, ke, te, etc. In how far
these are classifiers is a matter which has not yet been examined,
but in Annamese, which is a pure Mon language, there exist a
large number of classifiers, such as bank (for round things), bo
"
(for things which form a totality "), cay (for trees, plants,
and things made of wood), mieng (for soft things), so'i (for
ropes, threads, etc.), dam plantations and gardens), etc.
(for
When the Semang and the Sakai went to India many of the
dialects seem to have lost this peculiar power of formation,
and to have adopted rather a system of affixes. Yet, some of
these prefixes can be traced in a few languages, and especially
in the Naga languages. So in Angami Naga we find te/uh = a.
dog where
;
te is a prefix, which is
dropped the moment the
word is compounded with another, f.i. fuhpfd = z. male dog;

fuhkru = a. bitch. In mithu = cow, mi is a prefix in thenu = ;

goat, the is a prefix. In Ao Naga, ta is a common prefix in

adjectives, and ku and mang are verbal prefixes, etc. Several
instances could be drawn from other non-Aryan languages of
India to show that this prefix and classifier-system formerly
held sway in them, but has been lost. Thus :

" "
Chepang yuk = monkey is Brahmu pa-yuk. So the "
root
pe = io speak, is in Kami ta-pe, while "to be silent is in
Brahmu ma-pe, where ma is the prefixed negation. Khamti
wan = sun becomes in Laos kang-wan. So the root nt=day
" "
or sun becomes in Bododi-ni, in Kumi ka-ni, in Singpho
si-ni, in Lepeha sak-ni, in Kami ma-ni the root
; chu water
is in Gyarung ti-chi, and so on.
In the Tibetan language we evidently had originally a com-
plete system of prefixes or classifiers, which have now worn
down to the so-called prefix-letters which, though written, are
not pronounced any more, and have nearly all been lost in
Lepeha, which is nearly related to Tibetan.
It would seem, thus, that in the Ugro-Altaic languages there
arose a tendency to get rid of the prefixes, and to use affixes
instead, and this naturally did away with the principle of
classifiers. In Japanese we- find, however, a trace of the
prefix-principle, in the fact that the plural of several
wonl> i^
still formed by prefixing some noun or expression of multitude,

a system which is regularly followed still in Annamese, and of

which examples will be given later.
It is a remarkable fact that some of the f>rc/i\<^ in Bantu
are used in some of the other Ugro-Altaic language- .1-- ,///u<\s.
Thus the prefix ku (or uku) which in most Bantu languages is

the prefix of the infinitive of the verb (or. to he more correct,

the noun-form of the verb) is found in Mongolian and related
 74

languages, as an affix with the same meaning. In Mongolian

"
it is known as the infinitive noun ending. So ire-ku is the
"
coming" ja-bu-chu (chu = ku) is the going." Turki has the
;

same affix ku or gu, though in most cases the initial consonant

is dropped. In Jakutish we have kor = to see, and kor-u (for
"
original kor-ku) for the seeing, sight."
So we have in several Bantu languages a prefix In, which
often forms nouns from verbal roots. In Mongolian the affix
I has the same effect, and a strengthened form lik in Turki is

the same.
It may seem to be rather peculiar to maintain that, within
the limits of one group of language, there should be a change
from prefixes to affixes. Arid yet the phenomenon is really a
very common one among three very nearly related languages,
of the Aryan group, viz., German, Dutch and English, languages
so nearly related, that they might be called Teutonic dialects.
Yet it is a very easy matter to show that, what are prefixes in
Dutch and German, are- affixes in English, and no less easy is it
to show that the prefixes are older.
(a) In Dutch and German we have a prefix be, which has
been nearly* completely lost in English, and the place of which
in English is taken by a preposition after the verb, which from
a true philological point of view is really an affix.

Dutch fo-loopen is English to walk over.

fo-leggen ,, ,, to place in.
to fly at.
German &e-sprechen ,, to talk over.
or about.
fo-dauern ,, ,, to be sorry about.
,, fo-denken ,, to think over.

(&) German aws-gehen is English to go out.

,, to seek out.
aws-blasen ,, ,, to blow out.
Dutch mY-halen ,, ,, to take out.
uit-geven ,, ,, to give out.

(c) German aw/-stehen is English to stand up.

,, nieder-sitzen to sit

and many similar examples.

examples under (b) and (c) the prefixes

It is true that in the
in German and Dutch are separable, and can be put behind the
verb in certain flectional forms, but this really does not affect
the question, and it certainly cannot be done with the examples

*I say nearly because we still find this prefix in the English words bedaub,
besmear, bespeak, beware, etc. As the Dutch and German prefix be is rendered
by various prepositions, so the prefix be in these English words has really
different meanings.
 75

under (a). The fact that in English the prepositional affix is

written separately from the root-word, does not affect the
question either, for this is a mere spelling and not a structural
question. In speaking there is nothing to show that talk-over
is structurally different from be-sprechen, except the fact that

the one word has a prefix and the other an affix. In fact, in
some words, such as the noun lookout, which is the Dutch
uitkijk, the affix in English is actually written as one with the
root, and there are several more expressions of a similar kind.
Everyone acquainted with the rudiments of Teutonic phil-
ology knows that the prefix construction is the older, and is
found in the immediate parent of German, Dutch and English,
namely in Gothic, which is very rich in prefixes.
Of the Ugro- Altaic languages, outside of India, the oldest
representative we are acquainted with, is undoubtedly Su-
merian.
Now in Sumerian we have certainly prefixes, and prob-
ably, too, a system of determinatives or classifiers, which were
taken over to a considerable extent by Assyrians.
As regards the prefixes, we have first the series a, e, i, u.
Prince, in his Sumerian Dictionary, vol. I., p. xvii., says that
"
these prefixes have an abstract signification," a term which is
somewhat difficult to understand. I am inclined to believe
that in reality these prefixes were the demonstrative article
a = this, used as the article the or a. We find the same pheno-
menon in Lhota Naga, where every word, if not otherwise
defined, takes a prefix o. In Annamese we also find an article,
and the fact that in Bantu we have this same prefix a, e, i, u,
with evidently the force of the article, makes me think that the
Sumerian prefix had the same effect.
ki as prefix in Sumerian denotes place ;

lu denotes the agent

sa denoted, as Prince maintains, abstract relationship.
=
nam ( face) is an abstract element, usually denoting a state
or condition.

Of the determinatives in Sumerian we really know very little.

except their Assyrian meanings. Before male proper names a
sign was put, which is read as dis or gis in Sumerian before the
;

names of tribes and professions, a si^n which reads iiu-/i<

man in Assyrian, and which in Sumerian would read in full
mu-lu, but as determinative probably read short, as mid or inn.
Before trees and wooden objects a determinative was placed
which in Sumerian was generally read as is, but somrtinu > as
mu ;
its Babylonian form was isi.

In discussing the Bantu prefixes and their origin, shall I

have to refer often to these Sumerian prefix^ and determina-

tives, which are undoubtedly closely connected with the old
Bantu prefixes. In doing so I shall follow, for convenience
 76

sake, the generally accepted system of Bleek, as explained in

"
his Zulu-prefixes, on page 161 of his Comparative Grammar
of South African languages." These are as follows^ (without
the article] \

Singular. Plural.
i. mu. 2. ba.
3. mu. 4. mi.
5. li. 6. ama.
7. isi. 8. izi.

9. in. 10. izin.

ii. lu. 12. izi ; izin.
*

14. bu.
15. ku.

There is only this difference between my arrangement and

that of Bleek, that the latter considers the a of the 6th prefix
and the i of the 8th, gth, loth and I2th prefixes as the article.
I shall explain, somewhat later, why I differ from him in this
view.
I. The ist prefix mu.
It is generally consideredby all Bantu scholars that this is
the special prefix used to denote persons, and there can be no
doubt that this is correct. It is undoubtedly the Sumerian
determinative which was rendered in Assyrian as amelu. Mu
is a very old name for man, or human being, and appears as

ma, me, mi, mo, and mu in a large variety of languages, f

The root form of this word is found in the pure Bantu form
mu-tu,met with in Aduma, a language which displays some re-
markably pure forms the form is also found in Kinika. Mutu
;

or any of its other forms is always used in Bantu to denote a

man, i.e. a male, though in the plural it also includes women,
and has then the general sense of people, a development which
"
is evidently of late date. The root ma as meaning human
"
being is already found in Sakai, and in Semang (with a prefix
ha] as ha-me, and (with an affix) as me-nik.
In* the Indian languages the word does not appear in its pure
Bantu form,J but in Assyrian we have mutu in the sense of
"
husband," and in Tibetan we find a related form mthu in
"
the meaning of virile power." There can, however, be no
doubt of the origin of the word. In Semang tau (tu in some
*The 1 3th prefix of Bleek is the diminutive ka, which is really an infix, and
which I cannot consider here.
"
tSee for particulars of this B. H. Hodgson's
"
Miscellaneous essays relating
to Indian subjects (Triibner's Oriental Essays), Vol. II., pp. 59-61;
{In one Indian language Mru, man is mru, and this agrees with the Shuna
=
wfuw man, and the Kikuya murumo. The Indian Mru probably stands for
an original mtu, as the change from t to r is by no means uncommon in either
the Indian languages or the Bantu dialects.
 77

dialects) means "male," and with the prefix ww = human

being, mutu would thus mean a " male human being," which is
exactly its real meaning in Bantu.
In several Bantu dialects the word has, however, undergone
remarkable changes. In Zulu it is u-mu-ntu, as usually written.
This spelling is, however, really wrong, and should be u-mun-tu.
The u is the article, mun is the nunated form of mu, that is, a
form with an infixed n, which is a very common phenomenon
in Bantu tu has remained unchanged.
; In Kafir the word is
umntii, with elision of the vowel. In the Herero we have
o-mun-du, perhaps formed under Sumerian influence, where
" " "
du is son or male," as well as tur, which latter is simply
a strengthened form of tu. Secwana mo-thu, and Serotse mo-
nu are deviations, the latter being perhaps connected with the
very common non-Aryan Indian form mana. In Mang'anja
we find the form mu-ntu, as Scott writes it. Scott wants to
make out that this word consists of the prefix mu and ntu =
matter, thing, and that the word would thus really mean the
"
man-thing." I am afraid this view is indefensible. The
plural of theMang'anja word is Antu (for Bantu or Wantu] and
this would not agree with Scott's idea, but shows that Mang'-
anja evidently has worn-down forms, and as such I can only
consider the Mang'anja form the same as Zulu u-mun-tu. with
nunation, but elision of the article.*
Upon the principle of this rootform mu-tu is probably based
the first prefix, which is used for personal designations. In
many dialects its combination with the article u (or its varia-
tions), namely, umu, is shortened into um, especially in Zulu.
II. The second prefix of Bleek is the prefix ba, which, i- used
to form the plural of the words of the first class.
There has been a good deal of unnecessary speculation about
the origin of this and the other plural prefixes in Bantu, due
to the fact that the philological position of the Bantu languages
was not understood, and that a special system of philology
was thus built up by the first Bantu scholars, such as for in-
"
stances the Pronominal System of Kolbe in his Language
"
study based upon Bantu (London, 1888), a work upon which
evidently a good deal of time and great ingenuity has been
spent, but which is really useless.
In all the old M
on- Khmer languages the plurals were formed
in two manners (a) by repetition of the word, a form >till
:

found in Sumerian, and not infrequently met with even in

Bantu ; (b) by prefixing to the singular some word or >\ liable

*I admit that there is some difficulty about the M.un;'.ini.i l.irm. m \

Manganja chi-ntu ^a tiling. '.<- any (limn without litr. Then- n-itamU
"
old Bantu word ntu = anything that has a shape, or is i>cu<-j>til>l<- In tin-
be (omul tor this \\onl.
eye," and a very clear Turanian origin can Hut th-
composition of chi-ntu
is not vrt quite clear to inc. and I am not prrj..
" "
present to agree with Scott that mn-ntu expresses a living tlnni; and Jn-
"
ntu a dead thing."
 78

implying multiplicity. Thus in Annamese the idea men can

still be expressed in these two forms: ngu'o'i ngu'o'i=men
(by repetition of the singular ngu'o'i) or by nhung ngu'o'i,
really meaning all men. In Semang and Sakai the same
systems are followed. In several of the Indian languages of
this group, we find the same, with this difference, that here
has already taken place the change from prefixes to affixes.
Thus in Lhota Naga the nouns of multitude (oten, oyo, elam,
etc.) are put behind the nouns, which they qualify. In Ao Naga
" "
prungla is used in the same manner to form plurals.

Now, in Annamese we find the prefix bo used to express a

totality, i.e. a composite body of homogeneous parts, e.g. a
crowd of men. In Japanese this word in the form of ban,
" "
meaning at present ten thousand," but formerly any large
number," is still used as a -plural prefix to many words, especi-
ally such as have been introduced from the Chinese language.
Thus nin (originally Chinese) means man, ban-nin people, the
men. Mots is thing, bam- mots = things.* This plural prefix
ba or ban is clearly found in the language of Malacca, where
we have in Semang ba-lo =many, in Sakai fo'-ga = many,
and e-bah = 3\l ;
in Jakun we find ban-yak =many.
The root of this plural prefix is evidently the old verb ba =
to divide, found in Sumerian as well as in Bantu (Zulu aba,
Luganda ga-ba). The most ancient form of the verb was ka
(compare the above cited Mon-Khmer kat=to cut), and it
still has that old form in Semang, and in several Bantu lan-

guages (Kongo kaya, Bobangi kakola, Zulu (si)-ka = to cut,

etc.), but as already shown, there was always a tendency in
several Ugro- Altaic languages to change k into b or p.-\
This ba or ban appears in a number of Bantu words, all ex-
pressing multitude. A few may be quoted here :

Secwana bantsi = many.

,, ban we = some.
Bobangi bala = to count.
Zulu isibalo = a number.
,, balela = to numerate.
Luganda ba-la = to produce.
Dualla ben-ga = to increase.
,, a-ba = to divide.
" "
Now, it is sound expressing division or
this ba, this usual
" "
plurality which underlies the Bantu prefix of the plural ba.
For instance, the singular of the ancient word for man was
"
mutu, and the plural men," was formed by the prefix ba, so
*The fact that the of the original ban changes in the latter case to m (by
assimilation) is strong evidence in favour of the supposition that the of ban
is merely a case of nunation, and that the original word was really ba, which

agrees with Annamese bo and Sumerian ba.

fThe p form is still found in Finnish paljon ma.ny, which is exactly the
form of Semang balo, with a slight phonological change.
 79

that we first had ba-mu-tu. But the m

had after the labial b
a natural tendency to change to n, especially in quick pro-
nunciation, or to express the process more clearly, the second
vowel u probably fell out first (a very common thing, e.g. in
the Indian languages) and we had thus bamtu. But will, m
before t, always assimilate to the hard n, and thus we get
Bantu or, with the article, a-bantu.
When once mu had become the singular prefix to certain
words upon the mu-tu idea, it was but natural that, upon the
-nine analogy the ba prefix of ba-ntu should be used to express
the plural form of this class of words.*
What strengthens my
explanation, as given above, is the
fact that in Assyrian, which, as we know, took over several
expressions from Sumerian, we find the word banutu in the
expression mar-banuti (plural, or perhaps genitive), which is
" "
generally translated as the sons of the free," or the sons of
the noblemen," but which, considering its ideograph Tur-Kak,
"
I should feel inclined to consider as meaning the men of the
people," i.e. the lower free classes, f in opposition to the official
and priestly classes, a designation found exactly in the same
manner in English history, and in fact in the history of all
Kuropean States during the early middle ages, and which even
in Rome we find in homo liber as opposed to the liberti and the
servi. In this expression I certainly consider the ba a plural
prefix, and I may in support thereof point to the Assyrian
ba-chulati = soldiers. If this surmise is correct we have the
remarkable fact that Bleek in giving the name of Bantu to
"
what had been before called the Kafir races of Africa," has
unconsciously revived one of the actual old names" of the race.
If not, then such a coincidence of nomenclature is beyond the
ken of science."
III. and IV. The consideration of Bleek's third and fourth
prefixes, viz.,singular ma, plural mi cannot conveniently be
treated here, but must stand over until we have treated the
other prefixes.
V., VI. and XI. I shall now first discuss Bleek's fifth pre-
fix as found in Zulu, and in order not to have unnecessary
li,

repetition, I shall include in this discussion the eleventh prefix,

which in Zulu is lu.

"This naturally implies that the present system of prefix,-* or r/

developed in Bantu itself, after it parted limn the parent lani;uai;r>. I'his
1

certainly would not deny, but my contention

is that the ft in, if!,-
underlying
these prefixes or classifiers, and the original forms of these rla-^itiers must be
deduced from the more anment languages. The mistake, thus tar made, has
been to consider the principle as well as the form as of pure Bantu origin.
jThe expression is taken from the Behistitn inscription <>t KIULJ Danu-. and
at such a late period there had taken place no d<>\iln onsidrnd>le changes in
<

the original meanings of expressions, and in Darius' time mar-banuti may

" "
have been the sons of the noblemen," or patricians."
 8o

I strongly doubt whether, originally, this li and In were

really prefixes. In several Bantu languages they do not exist
at all. According to Bleek they are in Kikambo i(y) or u,
which is equivalent to saying that they do not exist, for the
i or u here is the mere old Bantu article. In Shambala the
fifth prefix has disappeared altogether, for if Bleek and Seidel
maintain that the fifth prefix is Shambala in n or m, they are
completely wrong, for the n and m here simply represent
nasalisation, and this is clear from the very fact that this
class has in Shambala no actual plural, but the plural is the
same as the singular.* In Herero the fifth prefix is lost, for
the e for that class is simply the article o in a weakened form,
"
and Brinker has correctly noticed this in his Lehrbuch der
"
Oshikuanjama (Berlin, 1891, page 88, II part), though he is
wrong in supposing that the old form was eli. But Herero has
for the eleventh prefix ru (for lu), and Oshindonga and Oshi-
kuanjama have lu, and the same is the case in Nano and Kongo ;

Mpongwe has no prefix lu. Dualla has for the fifth prefix
a very unusual and probably old Turanian syllable di, dj or
d, which cannot be treated here, while Isubu has the same.
It may thus fairly be concluded, that as this lu is not a general
Bantu prefix, like most of the other prefixes, it has a foreign
origin, and that I indeed believe to be the case.
There does, however, exist in some Ugro- Altaic languages
an affix /, which may be the remains of an original fuller form,

represented in Turki as lik. In Mongolian this affix / has, for

all practical purposes, the force of an article. Thus nom-la = to
learn, while nom-lal is the learning; mede is "to know,"
" " "
and medel is the. knowing or knowledge," and it would
thus seem that this / forms nouns from verbal roots, and in
Jakutish the / does the same.
In Arabic and in several of the so-called Hamitic languages
in the North-East of Africa, which have taken over several
elements of the old South Arabian languages (Himyaritic,
Minaean, and Sabaean) we find an article composed of a vowel
and an affixed and sometimes prefixed /. In Classical Arabic
we have el ; in Masai we have ol (masculine plural, t/),and in
Bari we find lo (probably due to metathesis). f
There can be doubt, but at a very early period the
little
Bantu invaders came into contact with these Hamitic
of Africa
races, which inhabited in those times not only Abyssinia and
the present Somali and Galla-land, but also the regions of the
Nile, north of Lake Albert Nyanza, and who may have pene-
trated even as far as the present British East Africa. At a
later stage the Bantu certainly came in close touch with the

This is accounted for partly by the fact that Shambala is one of those
Bantu languages which has lost the ordinary vowel-article.
fl have not been able, thus far, to trace the origin of these forms.
Masai, which have always been a warlike race. If we com-
tribes as well as some of
pare, e.g., the weapons of the Zulu
their customs, with those of the Masai, one must come to the
conclusion, that not only were some of the Bantu tribes in very
close touch with this race, but that they may even have been
subjected by it.*
The Ugro-Altaic Bantu had its definite article, which was
originally a, but which, in accordance with the ancient vocalic
harmony (of which we still find part preserved in Bobangi) could ,

become e i, o or u. When the Bantu came into close contact

,

with the Nilotic, Hamitic and Masai tribes their language was,
to some extent, affected, and this would be especially the case
with such Bantu tribes as were, for a longer or shorter time,
subjected by those races. So it happened that in some of the
Bantu languages the original article underwent a change, and
added to its vowel the lo or lu of the Hamitic languages, and
thus we have the ilu and ulu of Zulu-Kafir, and the oru (r=l)
and olu of Herero, Oshindonga and Oshikuanjama. But
these latter languages retained the fifth prefix as its original
article, as did Nana and Kongo. The Mpongwe, a Bantu
tribe, which evidently migrated to the west at a very early
period, never came sufficiently under Hamitic influence to
suffer any change and hence it kept its pure i and o.
in its article,
Tt is remarkable that, thoughin Luganda we have the
fifth prefix as e, 'this e is a so-called explosive one, which
tends to show that a letter, probably / has fallen out. So e'bere
(as written in Luganda) may stand for an older el-bere, and
this is the more likely because the pronominal concord of this
class is li. In Bobangi, again, li is only the pronominal con-
cord in the plural, while e is the concord in the singular.
There are several linguistic phenomena in connection with
these prefixes, which tend to show the correctness of this view.
So we have in Bantu a root zwe or zwi, which originally meant
" "
voice," and afterwards country," a root, which is of un-
deniable Ugro-Altaic origin. In Zulu zwi, meaning ." voice,"
"
lias the prefix i, while zwe, denoting country," has the prefix
Hi. This certainly proves that the prefix Hi cannot be a
classifier, as has been maintained by several authors.
Another reason in favour of my, theory is, that the plural of
the fifth prefix is generally atna, that is, that it takes one of the
general plural forms, and this is the same with the U \vnth
prefix, which usually has as plural the general plural form izin,
zim or izi, which is also the plural of the ninth class.
* The
Masai, I venture to suggest, are probably the descendants .>! those
Asmach mentioned in Herodotus II.. 30, as being Egyptian soldiers who
deserted and .mn;i .ited to Ethiopia. Herodotus places this emulation in tin-
nine ol Kim; 1's.immetii-hus, but it probably 'took place much earlier. In
"
ancient Egyptian Masu actually means a soldier," and in the M.ixn language
there are several words of undeniable Egyptian origin, e.g. kef>er=a\>
dito = girl and others.
 82

VII. and VIII. I shall now first discuss the seventh prefix of
Bleek, isi, which forms its plural in the eighth prefix izi.
Bleek considers the i here as the article, but I think lie is
wrong here. We have here, as I shall show, an absolutely
pure classifier, and the i of the article is probably contracted
with the classifier isi, hence we have really isi.
There can scarcely be any doubt but that this is a very old
Ugro-Altaic prefix and classifier. In Sumerian we have the
prefix is (ish) for all objects connected with trees or plants, or
for anything made of wood or in general we may call,
fibre ;

it the prefix of the- Issu in Babylonian means

vegetable world.
plant, tree, and its usual ideograph is a sign, which is usually
read is, but seems to be read occasionally mu. Of the origin
of this word there can be no doubt, as we find already in Sakai
"
isi in the meaning of wood," a root again met in the non-
Aryan languages of India, as in Dhimal sing, Pahdi si-ma,
Nachhereng "sa'a, Angami Naga si, Sibsagor Miri ising, etc.,
all meaning tree."*
In Sumerian we find this prefix in a large number of words,
directly or indirectly connected with the vegetable kingdom,
such as :

is-gusur
= a beam.
is-na = a bed (made of wood).
is-su-dis = a doorbolt (made of wood).
is-pan = a bow.
is-ma-nu = a cage.
is-gu-za
== a chair.
is-mar = a chariot.
is-ter = a forest, a tree.
is-sa-tur = a net (made of fibre).
is-sar = an orchard.
is-ir = a rope (made of fibre),
and a large number more.
In the Zulu and Kafir language we have, in a similar manner,
a large riumber of words, connected with the vegetable world,
which also have the prefix isi. Thus :

isi-biba = (a herbal) antidote for snake-bites,

izi-kali (pi.) = weapons (originally of wood, just like the
Sumerian is-ku weapons, with which,
indeed, the Zulu word is closely con-
nected) .

isi-dabene = wild banana,

isi-tya
= basin (of wood),
isi-kapa = blade of grass,
isi-nkwa = bread,
isi-dhliso = poison (vegetable).
*Under such circumstances it is very likely that the Bantu prefix is not
derived from Assyrian or Sumerian, but directly from the parent-language.
 isi-gabe young pumkin.
isi-konkane wooden pin.
isi-hlenga raft.

isi-rongo screen (of mats).

. isi-tuniz shade (of bushes).
isi-bonda stake.
isi-guba hollow vessel (made from a gourd).
isi-nga a clump of thorn-trees.
isi-pato handle (of wood).
isi-lando a needle (of thorn).*
and a host of others.

But *s in Sumerian also meant earth, and it seems than the

two words is were confused at an early period, a confusion
which became strongly developed in Bantu. Hence is, even
in Sumerian, is occasionally used as a prefix to things con-
nected with the earth, e.g. is-hu = \vorm (really "earth-ver-
min"), is-du = to pull down ( = to throw on the earth). In
Bantu there are a large amount of such examples, e.g. in Zulu :

isi-bomvu = red earth.

isi-buda = red ochre.
isi-sindi = sod.
isi-gagadu = hard soil.
isi-daka = black soil.
isi-sekelo = foundation.
isi-za = a plot of ground.
isi-kala = a chasm.
isi-dindi
'

= a clod of earth (really a bye-form of

isi-sindi) .

isi-wa = a crag,
isi-makade = a fixture,
isi-duma = a mound,
"
isi-hambi = a traveller (literally one who goes
over the ground ").
etc., etc.

It should be remembered that very often new words and

ideas are formed from old ones, which latter were originally
connected with plants or the ground, though the new ideas
have no such affinity. In such cases, the new words would
retain the prefix isi, in accordance with the old association of
"
ideas. Thus in Sumerian is-mi means protection." which
has apparently nothing to do with plants or the ground. But
" " "
the word originally meant shade or, literally, the dark
" " "
of the tree," and as shade was a protection against the
hot rays of the sun," it actually, by analogy, obtained that

*It is, however, more than likely that tins is not really an isi prefix, but that
the syllable si belongs to the root, and that the word is really i-*
 84

meaning.* So the Sumerian is-hu = a bridegroom, has in it-

self nothing in common with a tree or plant, but originally the
"
word meant the preparer of the hut," and the hut being
made of vegetable material, the word obtained the prefix is.
Not unlikely the same psychological process took place in
Bantu, where we find numerous words, beginning with isi,
which do not at present have any apparent connection with
either the earth or the vegetable kingdom. In many cases in
Zulu and Kafir isi stands as an apparent prefix for words taken
over from English and Dutch beginning with sh or sch. The"
sound sk the Bantu cannot pronounce, so he inserts the vowel
i between the s and the k, and as the words obtain the article

i, there originates an apparent prefix isi. In Bantu grammars

and dictionaries these words are really wrongly placed with
the prefix isi, though it would seem that the instinct of the
language considers them to belong to that class, on account
of the mere sound. So we have the Dutch schaap becoming
isikapu, schip=isikepe, zens=isizenze (this may be a pure
prefix), sikkel=isikeli, and several others might be mentioned.
In some cases the article i has nunation (adding of n) Thus .

"
the Zulu insikane = sedge, reed," ought to be isi-kane from
Assyrian kanu=reed or sedge. Zulu insimu (which in Kafir
has still been further deteriorated into intsimi}= field, is really
"
isi-mu, meaning a land covered with vegetation," from isi
and w#=land. The nature of this nunation being misunder-
stood, these words are now considered as belonging to the
ninth prefix class.
The eighth prefix izi, izin, izim, I believe to have been simply
caused by the usual re-duplication for plural forms, so that we
had the series isisi=issi = izi. I am strengthened in this view
by the fact that in the Shuna language, we actually find the
form zhinzhi in the meaning of many. Once being adopted as
the plural of an important class of nouns, this form gradually
grew into what may be called a general plural form, and hence
we find it as plural prefix for the tenth and twelfth classes, a
clear proof of the fact that the Bantu have long forgotten what
was the original meaning of these prefixes.
VI. Now let us consider Bleek's sixth prefix ama which forms
the plural of the fifth class, and also, in some cases the plural
of some words of the first class, and is especially used in form-
ing the names of tribes or nations, such as, e.g., the Amazulu,
the Amakosa, Amaswazi, et.c.f
I consider that the first origin of this plural prefix is in its
use as a tribal designation, and that originally, indeed, the
"
*We, in English, often use the word shade in the meaning of protection,"
and even the word umbrella, through the same association of ideas.
fThis tribal prefix Ama is only used in some Bantu languages a verv large
;

number (e.g., the Bechuana tribes) use Ba as the plural tribal prefix, and this
Ba is the same as the second prefix found in Bantu.
 "
prefix meant people, nation." There is to be traced a whole
"
series of thisword or expression ama, in the meaning of peo-
ple." In Semang we find for people hame, in Sakai mai and
mah, in Lhota Naga we have yam (where the y is inserted) ;

in Mandchu we have amaga= descendants, posterity; in

Turkish we have amme=people, and in Sumerian amar = pos-
terity. In Tibetan and Lepcha, in which very few words begin
with a vowel, we have the related forms mi = mankind, and
mwto = nation, words found in several Indian non-Aryan lan-
guages. That in ancient West Asia the term was known as
"nation," is clear from such names as the Amorites = the (

people of the mountain), the Amazones (which"

has nothing to
do with Greek Zone, but simply means the fighting people ")
and others, and it is probably a prefix in the Assyrian amelu,
"
which must originally have been a plural form, meaning man-
kind."*
For that reason I believe that ama was first used as a prefix
to nations, as it still is in Sakai, and that hence we find its
oldest form in Amazulu, Ama Xosa, etc., which really mean
" "
the people of Zulu," the people of Xosa," etc.
When once this ama was used for " people " or " nation,"
"
i.e., a collection of men, it was used in analogous cases, and
so we find ama-doda=men (pi. of indoda) ama-pakati = the ;

councillors ; =poor men ama-kwenkwe =

ama-hlwenpu (Kafir) ;

boys. Gradually its range stretched, and it was used even for
collections of women, e.g. ama-nkazana (Kafir) = women ama- ;

kosi-kazi (Kafir) = chiefs' wives. f At last, as in the case of

izi, the prefix ama became a general prefix for the plural.
This must have taken place already at a time when the Bantu
had not yet departed from their original Ugro- Altaic stem, for
"
in Japanese we find amata in the meaning of many," and in
"
Sumerian ama has the meaning of multitude."
I have already mentioned the fact that one has to be careful

to distinguish real prefixes from apparent prefixes in Bantu,

and it may be useful to illustrate this in connection with <?>;/</.
It is generally considered by Bantu scholars, and practically

by the Bantu themselves (because they use plural concords

"
in
connection with the words) that amanzi, meaning water,"
amasi = fermented milk, and amafuta = i<\(. an- real plurals.
Yet this is absolutely wrong, and the words which are to-day
plurals, were original singulars, and
have only come to be
considered plurals through false analogy.

*In how amn i. uinerted with uniu mother. i.mnot here di

far this .-. 1

hut ! considerquite likely that the latter is rr, illy tin- -ootword.
it

K.ilir ninti-nv<i>ii;f -the people oi old, nm.i has

fin the very remark. il>lr it-

lull lorce of "people." I'lu- ,-t doubtful it may mean

vmuJ.>i;y <>l nvnn^i- is ;

the same as M
tinman in the people <>! tin- uaters hut it might br that ;

nyangc ;i< tuallv stands fur .Yuo'. and that we have thus here one of the olcL
ii, lines lor the I'.antu.
 86

I have already shown that in amasi, the ma belongs to the

root of the word, which is clearly from the Beluchi mass sour
milk.
As regards water, there are two. old roots to represent this
element. The one is mi, mu or ma found already in Semang
and Sakai, and changed into some Turanian dialects into wa
(m=w), which originally meant the "water from heaven,"
namely rain. The other is ti, changed into "
most Turanian
dialects into si, chi or chu, which meant the water of rivers,"
i.e. the water on earth. In Sumerian we find the root a = water*
(which is probably =wa), and another related form mu, and
we also find the form a-me, which is generally supposed to be
a plural form, but which is possibly a conjunction of two

synonomous Such conjunctions of synonomous roots

roots.*
are not uncommon in ancient Eastern languages, and Hodgson,
op. cit., vol. II., p. 69, gives the Ugrian we-zi= water, as ex-
ample. Now Finnish has wesi, Hungarian has viz, and Japan-
ese has mu-zi. Bantu man-zi is formed upon exactly the same
principle. The n here is the ordinary nunation, as is clear
from the Makua mazi, Sesuto metsi (where ts=z), and Jao
me-si, while Kinika has mazi. The usual article a was placed
in Zulu and Kafir before it, and so we have amanzi. In course
of time ama was wrongly taken for the plural prefix, and hence
Bantu scholars have been giving some of the most absurd
etymologies for the word.
Amafuta is in exactly the same position, and is also a com-
pound of two roots, meaning the same thing. In Sakai mu =
fat ;
in Sakai pa-o = fat. In several non- Aryan languages
mo-to (where to is an adjectival affix) = fat, but in the same
group pi and phum is also fat. In Khyeng ma is fat in Shan ;

it is phyee. In Finnish we have woi (for moi) = butter; in

Magyar =
vaj butter.
It is. therefore, clear that ma-futa (the ending ta is found in
some of the Naga languages, e.g. in peleta =
ia.t} is a compound
of the two In Aduma fat =evongo, where we have
root- words.
only one root, agreeing with Magyar vaj.
XV. The fifteenth prefix of Bleek ku (with the article uku)
is a purely Ugro-Altaic prefix. It is weU known that in Bantu
it forms infinitives and gerunds. That is exactly what the
Mongolian affix ku or xu does, for which reason it is known in
"
that language as the Infinitive form." In Turki and in
Jakutish we have several forms. In Manchu ku as affix de-
"
notes only nomina
actionis."
It is, inopinion, most likely that all these endings are
my "
really different forms of an original uku = to make, to do,"
which is found in Sumerian as aka or ag (variant kak), in
*Prince in his Sumerian lexicon ascribes the reading mu to Semitic influ-
ence, but this is clearly wrong. I must confess that Prince's work is really a

very disappointing book, and is based upon completely wrong principles.

 7

Baluchi as kan-aga, in Brahui as kar. In Baluchi aga is the-

ending of all verbs and expresses the action. In Semang kai =
to make and in Sakai ka is actually found as a verbal infinitive
;

prefix, e.g. kabeh

= to build (root beh compare Annamese biia ;

= to make) kabuk = to bind (root buk) ; katut=burn (root

;

tut) katu = to pour (root tu)

; kajon=to give (root jno), and ;

a large number more, which the reader can easily find in the
Vocabulary at the end of Skeat and Blagden's aforequoted
work. It may be stated that in Sumerian this prefix does not
seem to be found, but that it is still found as an affix in some
" "
words. Thut ka-aga = to eat," is really to work with the
"
mouth," from &#=mouth, and aga=to work; mi-ga = to
"
set (of the Sun)," is really to make dark," from mi = dark and
aga = to make.
In Kafir uku is the regular prefix for the infinitive in Zulu, ;

and in nearly all Bantu languages it has the same force. In

Zulu (like in Manchu) it is often used to form verbal substan-.
tives thus uku-mela = opposition, from the verb mela = to op-
;

pose. In Kafir this is not so common, though we do find

examples, such as uku-fa= death, from fa to die.
Under such circumstances it is but natural that there is no
plural form for this prefix.
XIV. The fourteenth prefix of Bleek is bu, with the article
ubu.
must confess that this prefix is connected with greater
I

than any other, and that principally because

difficulties

throughout the Bantu languages this prefix, as a rule, is used

"
to form so-called abstract nouns."
In Turki and Mongolian we have an affix bur, which forms
nouns for verbal stems, and most-of these nouns seem to be
abstract. Thus, in Jakutish we have from the verb tolno = to
pay, the noun tolo-bur= payment, while in Mongolian we have
from the root tail = to declare, tail-bur = declaration. In the
" " "
Turki languages bar means existence and to exist." In
=
Japanese ba & state, condition. In Sumerian we have
bar = body, form.*
" "'
The conception abstract ideas must have been a fairly
of
late idea in the development of the human mind. In the Naga
languages very few abstract ideas "are found, a fact remarked
upon by Mrs. E. W. Clark in her Ao Naga Grammar," p. 5.
Snmerian too, possesses few abstracts nouns, and such as
there are seem to be preceded by the prefix nam =fate, position.
"
So, nam-lu-gal-la is used in the meaning of Majesty," but
"
literally it means the position of a kin^." There is thus some
n for supposing that in Sumerian these abstract concep-

*This root is most lil..-lv ronm-cto! \\itli S.ik.n /;a-/o = facc, and l>uku = to
The Semani;
>

ve. torin is niiilni. :md this r<x>t is found in Sium-riaii and

other A It aii- manages 1. as a parallel lorin. In Japanese tnukai = to face; in
"
SunitTinn )>t(i-rli<tr is front, apjuMrance."'
 88

tions arose only after the Sumerian had been conquered by the
Semites, whose power of abstraction was far larger. Now" it
"
would seem that this prefix bu, as a sign of abstraction is

not present in some of the older Bantu languages, or that it

was borrowed from the later dialects. In Nano, Angola,
Kongo, Mpongwe, it does not seem to" exist in "Dualla most ;

of the words beginning with bu are concrete objects, in

which bu really belongs to the stem of the noun (e.g. in boso =
face, and 60/0= ship), and in such words as bosangi = cleanli-
ness, and bubi=sm, the prefix may have been borrowed with
the word from other Bantu tribes.
In Luganda the bu class of nouns comprises a large amount
" " "
of concrete," as well as abstract nouns, and this is easily
accounted for by the fact that Luganda, though originally tin-
headquarters of the first Bantu invaders, became also the
headquarters of the second Bantu invaders, and was thus liable
to adopt several Sumerian- Baby Ionian words in its later
vocabulary.
Hence I am inclined to think that the prefix bu is directly
connected with the Babylonian banu=to form, and more par-
ticularly with the noun derived herefrom, bunu, which means
" "
outward appearance, form," and especially face." In fact,
in the latter meaning the root is found in a very generally used
Bantu word, which in Zulu is ubuso, a word which is, through
false analogy, supposed to have the prefix bu, but which is
" "
really u-buso.* The prefix would thus express the form or
" "
outward appearance of matters, and would lend itself per-
" " "
fectly"to describe what we call abstract nouns by giving a
form in language to things which have no form in nature.
" "
So ubu-ntu = mankind would be the general form of man ;

"
ubu-de = length, would be the form of being long."
As in all other prefixes, a considerable amount of mis-
"
understanding has arisen about this prefix, due to false

analogy." Thus utywoala =beer, is given in the new Kafir

Grammar of the Rev. J. McLaren as a palatised form of an
"
original ubu-ala, while Bleek in his Comparative Grammaj,"
p. 154, seems to consider that the u stands for an original ubu,
as this prefix appears in the Zulu form ubu-tywala. It is prob-
able that the newer form of the word is ubutywala, as would
seem from Secwana bo-yaloa. The root of the word is tywala,
which is the regular Zulu form of the old Baluchi juara = kafir-
corn (Sorghum vulgar e] under which name this plant is known
,

over the greater part of India. Ubutywala (for original ubu-

"
juara) would thus correctly represent that which is made
from juara," and this is strong evidence in favour of my ex-
planation of the original meaning of the prefix bu. On the
*It need scarcely be said that this buso = face is not derived from Babylonian,
but from the ancient Ugro- Altaic language, as is proved by Sakai balo = ia.ce.
Sirienish ban = face, front ;Votian ban = iace.
 89

other hand, u-tyani= pasture in Kafir is not, as McLaren

thinks, a form of ubu-ani, but stands for an original u-sani
(ty=s), from the Sumerian se-in-nu = forage, grain, straw,
and has only the ordinary article u as a prefix.
The more I study Bantu, the more do I become convinced
that our ideas of the Bantu prefixes require thorough and care-
ful revision, and that from a philological point of view we can-
not implicitly trust to the actual conditions under which the
Bantu themselves apply these prefixes. It is clear that the
Bantu have absolutely forgotten the origin and effect of these
prefixes, and, being led astray by false analogy and popular
" "
etymology, have made a fair mess of their own language,
even more so than the Englishman does, when he considers
such words as alms and riches as true plurals.
I may, e.g., quote here a very pretty example with reference
to the supposed prefix si or se denoting languages in Bantu.
Mosuto is one man of the Sutu tribe Ba-suto is the plural, and
;

Se-suto is the language of that tribe. Mo-shuana is one man

of the Shuana (now often written cwana) tribe Be-shuana is
;

the plural Sc-shuana is the language.

;
So Grout has I-si-zulu
for the Zulu language. Bantu grammarians represent, there-
fore, that for the single individual we use the first prefix, for
the plural the second, and for the language the seventh prefix.
This statement shows, with all due respect, how crude our no-
tions of Bantu philology are as yet.

The se or siused in front of the tribal name, to denote the

language, isno prefix at all, but is the old Ugro- Altaic word for
language, and is found in nearly all existing languages of that
group.
In Semang we have chu = to talk, and sua= voice. Sir-

jenish has si = word,and sua = to speak. Jakutish has

voice,
sej
= voice Manchu has se-mbi (where mbi is simply a verbal
;

affix) =to speak Turkish has sw2=word

;
Japanese has ji =
;

word, and se-tsu = ta\k (noun). Several non-Aryan languages

of India have che-wa = to speak, as well as jc, su-ang and su-
it in the same
meaning, which are all from the same root.
Finnish has srt-na=word; Magyar has so=sound, voice,
speech; Sumerian has 2g=speech, and evidently a si form,
from which the Assyrian si-ku = speech and si-ku-ru = to speak,
were derived.
From the above it is clear that the si or se in Bantu is evi-
dently the word for "language." In certain Bantu groups
(Suaheli, Kamba, etc.) we have the prefix A-/, and this prefix is
strong proof for our theory. For while si or ,sr is derived from
the Semang chu and sna. ki i^ derived trom the Sakai k-ui = lan-
=
guage, speech, \vhirh root j>, ,iK<> t.>im<l in Japanese knchi to
demand, and kotoba = language, in Mandchu knmitn \\\\\<\c ;
 9o

In Finnish pu-hua = to speak, and probably in Talien (one of

the Indian languages) han-kai = to speak, and in Sumerian ku*
IX. The ninth prefix in is in reality no prefix at all, but
simply the article in its i form with nasalisation or better nuna-
tion, and that same nunation also appears in the plural or tenth
prefix izin, which is the eighth prefix.
III. and IV. Lastly, there remain the third and fourth pre-
fixes to be discussed, viz., mu and mi.
It seems to me that the words which compose the third
class are a very heterogeneous mixture. In many of the words
-of this class, the mu is exactly the same as the mu of the first
class. Umu-zi= village, is nothing else but the article u, the
prefix w=man, and the root z&= dwelling, found in Sakai
si-ar and si-au=hut, house, dwelling, in some Indian dialects
and in several Ugro-Altaic languages, the real original root
being sa = spot, place. Umuti = tree (after
"
which word this class
has been erroneously termed the " tree class by some writers)
has no prefix mu at all. The mu here forms part of the root
of the word, arid represents the old Turanian mi (mai, mo,
mu) =fire, found in a very large number of non- Aryan
languages. Trees were considered as "things to burn," and
" "
the two ideas of " fire and " tree are intimately connected.
In some Indian languages (e.g. Southal, Singpo Kol, Bhumi
and others) sengel =fire while in Dhimal, Chepang, Cchi-
;

"
tangya, and many others, sing or sang is tree," and in Zulu
we find the latter root in isinga = a. forest. In the same way
Somali mi = forest is connected with mi = fire, and Mang'anj a moti
= fire, with umuti = tree. In Burj etian we find modo = tree, and
in Manchu the same word is moo, while Annamese, one of the

purest languages, has muk =tree. From the latter

Mon-Khmer
word perfectly clear that the mu in umuti forms a part of
it is
the root of the word. In words like umoya there is no prefix
mu, and the mo belongs to the stem of the root-word (see
above), and the u is the real prefix, or rather article. In Kafir
um-lambo =umulambo) = river, we have the prefix mw=man
(

and the root lam = way, road, and a river was thus defined as
"a road for man," a definition actually found not only in
Semang, but in several Indian languages, where the ideas
"road" and "river" are apparently confused, or very
intimately connected.
*This ku appears, inter alia, in eme-ku. This eme-ku with its counterpart
erne-sal, which together form the two languages of old Sumir has puzzled all
Assyriologists so far, and Prince, in his Sumerian Lexicon, pp. 13-15. has given
explanations which are very far-fetched. If Sumerian students had gone
somewhat further afield, they would have come to the very evident conclusion
"
that eme-ku is the spoken, ordinary language," the language of daily life,
while erne-sal is the " higher, written language, used in inscriptions and gov-
ernment decrees. Sal is connected with Semang snr-at = to write, Ao-Naga
zilu = to write, Angami-Naga thu (th = z) = to write Turkish zabtet
; ; Japanese
'

suri to print (" to rub ink on a stone ") ; Manchu sele,iyembi to publish ;

Japanese sho suru = to write Buretian zitranam = to make strokes.

;
 In um as a prefix for rivers in both Kafir and Zulu (Um-.
zimvubu, Umvolosi, etc.) the um has absolutely nothing to do
with the prefix mu, but this um is the shortened and metathised
word for ancient Semang munun = a. " ford," still found in Bur-
jetian umbanap=to wade, and, perhaps, with Sumerian mu-un,
"
which Prince, op. cit., p. 244, explains as what binds together
the land."
I believe that,
" fairly well the half of class 3 consists Jof
false analogies," and to do justice to the subject one should

carefully examine every word in this class, which cannot be

done here.
The plural form, or fourth prefix imi, I consider to be a mere
"
changed" form of ama, in accordance with the concord fbf
vowels so common in old Bantu.
In conclusion, I beg to reiterate my opinion that the prefixes
in Bantu require very careful examination, and that the rules
and principles now generally accepted should be carefully
tested. Careless work on the part of authors has certainly
complicated matters to a large extent, and independent re-
search is absolutely required.* The old theories, I am afraid,
will have to be gradually discarded, and we must look at Bantu,
its prefixes, its inflections, and its very
highly developed forms
in a different light than we have done thus far.

SECTION X.

Having given in the former paragraph what I consider the

natural and correct view of the Bantu prefixes, and the only
one in harmony with the historical development of this lan-
guage group, the next question to be solved is the Concord in
Bantu.
Like the prefixes, the Concord has been thus far considered
a linguistic phenomenon exclusively Bantu, except in so far as
Bleek has attempted to show that this Concord existed, to a
certain extcMit, in the Hottentot language. I am not prepared

to discuss this latter matter, which forms a subject in itself.

But there can be no doubt, that the Concord, as found in
Bantu, is certainly specifically a matter which has developed in
that language group itself, though the germs may be found in
earlier languages.
As an appendix to the first volume of their oh-quoted work
on the Pagan Races of the Malay Peninsula, Messrs. Skc.it and
Blagden have given us a number of Besisi and Blanda songs. I
*As a specimen of the careless way in which Bantu is handled 1 may quote
"
a book, called Kit-merits of Luganda C.ranunai ." written by a missionary in
t'^.-inda. Then- 1 t'md mi pu^e 4*. ckitabo = book. I nun >! the ki
class (sic) ! Naturally <-kitnln> is tin- HantuiM-d lurm n! \ralnan A/Vn6=book.
and the prefix is and not t'ki.
< Such things are bound to make Bantu
"
confusion worse confounded."
have carefully gone into these, but I must confess that, though I
"
have found a good deal of repetition of words," I have found
nothing like Concord, or even Alliteration, in them. Unfortu-
nately, I have not been able to obtain any songs of the non-
Aryan tribes of India, and I am thus unable to say whether the
germs of Concord are found there.
Before, however, entering into this question historically, we
" "
might first ask another question, viz., What is Concord ?
"
My answer to this is Concord is Alliteration systemised in
:

such a manner that the alliteration is dependent upon the

consonants, or any particular consonant, of the word, which
expresses the leading idea in the sentence. Usually that lead-
ing idea is contained in the subject of the sentence. Under
special circumstances the Concord can depend upon the vowel
of the leading word in the sentence."
I have just now called Concord a linguistic phenomenon, but
in reality this is a misnomer, for Concord has really nothing to
do with speech. Speech is connected primarily with the tongue,
the teeth, the throat, the palate, and the various oral organs ;

Alliteration and Concord are solely connected \vith the ear, and
hence it should be classed as a musical phenomenon. Allitera-
tion originated in Poetry, when Poetry was pure and simple
song. There is no doubt that originally such songs were
connected with religious rites.
The musical faculty in the Bantu is splendidly developed

and, as a rule, they have a remarkably good ear. That is. to a

certain extent, one of the reasons why the Bantu have de-
veloped the Concord to such a remarkable degree.
" "
In his Kilima-Njara Expedition (London, 1886), p. 460,
Sir H. H. Johnston, then, as it were, at the commencement of
his brilliant career, has attempted to show that there exists a
kind of Concord in Galla, and even in Arabic. But it is a ques-
tion open to doubt whether this Concord is the same as that of
the Bantu.
Placed as I am, I have not been able to obtain access to any
important ancient" Ugro- Altaic poetry, or even to Sumerian
poetry. But the Great Epic of Gilgamesh," with its magnifi-
cent Legend of the Deluge, was undoubtedly originally a
Sumerian production, of which we only possess to-day the
Babylonian adaptation. But even in that adaptation we find
clear traces of alliteration, showing that the Semitic translator
tried to introduce into Semitic a non-Semitic principle. In the
"
following passages, taken from the Legend of the Deluge," as
"
found in
Rosenberg's Assyrische Sprachlehre (Hartleben),"
Vienna, p. 60 and ff.. some examples of this attempt are given :

Line 13.
A/u su-u /abiirma i/aani kirbusu.
(r and /
being always interchangeable, would naturally fall
under the same principle of alliteration.)
 93

Line 27.
Su/ima zeeer naapsaati ka/ama ana /ibbi e/iippi.
Line 83.
Miimma isuu esiensi huraasu.
Line 85.
Uste/i ana /iibbi e/iippi ka/a kiimtua u sa/atiia.
Line 147.
Usesima suummatu umaassiir.
Line 163.
T/aani u//anuumma Bee/it i/aani ina kasadisu.

I have just picked out a few of the most prominent examples,

but many more might be quoted from the same poem, and it
seems evident that the Assyrian poet attempted to imitate the
alliteration of the Sumerian original, as far as the Semitic lan-

guage allowed him to do so.

Not having been able to obtain access to any Sumerian poetry
(a fact which has undoubtedly handicapped me considerably) I
had recourse to what is perhaps the nearest to it, namely,
"
the magnificent poetry of the Finns, as found in the Kale-
vala." The whole of trfis great epic I could not obtain, but
"
luckily I found many extracts from it in "A. Castren's Vor-
lesungen uber die Finnische Mythologie (translated by A.
Schiefner), a copy of which is possessed by the S.A. Library in
Cape Town. What I saw there was more than sufficient for
my purpose. For there I found not only pure Alliteration, but
" "
actually the first of the Bantu Concord.
dawnings I shall

only quote a few verses, in the original, without giving Schief-

ner 's German translations, because the only object is to show
how perfectly this alliteration exists in Finnish poetry :

I. Anna ucko uuhiansi

Anna oinahat omansi
Ukko kullainen kuningas
Tuuvos ilman tuusimata
Varomata vaaputtele.

II. Pikkuisessa pirttisessa

Kamarissa kaituisessa
Kiven kirjavan kylessa
Paaen paksun kainalossa.

III. Vaski oli hattu hartioilla

Vaski saappahat jalassa
Vaski kirjat kintahissa
Vaski-vyovyt vyolle vyotty
j
Vaski kirves vyon takana.
 94

IV. Kaistat kalaisen karjan

Taman nuotan nostimille,
Sata-lauan laskimille
Kalaisista kaartehista,
Lohisista loukeroista,
Suurilta selan navoilta,
Synkilta syvantehilta,
Taivan paistamattomilta,
Hiekan hieromattomilta.

V. Sorea on suonten vaimo

Suanetar sorea vaimo
Soma suonten kehrea^a
Sorealla kehrinpuulla
Vaskisella vatrtinalla
Rautaisella rattahalla
Tule tanne tarvitaissa
Kay tanne kutsuttaessa
Suoni sykkyra sylissa
Kalvo kaari kainalossa
Suonia sitelemahan
Paita suonten solmimahan
Haavoissa halennehissa
Rei'issa revennehissa !

Surely we have here alliteration with a vengeance ! But have

we here the Bantu concord ?
In Bantu the Concord consists of alliteration dependent upon
the prefix of the subject of the sentence, or better, of the word
expressing the leading idea of the sentence. But leaving the
consideration of the prefix question aside for the moment, the
Bantu concord is dependent practically upon the first letter of
the leading word (excluding the article). Thus in the Zulu
sentence given by Bleek, page 97 :

"
A&antu 6etu a&achle &aya-6onakala siftatanda,"
" "
the concord is dependent upon the first letter of the word

bantu, which is the subject of the principal sentence, and con-

veys the leading idea of the sentence.
In the Finnish examples given above we find in most cases
an alliteration of the initial consonant (and even in the first two
lines of Example I. of the vowels). In nearly all cases, the
alliteration is dependent upon the leading word in the sentence.
So in Example II. all the first words are the nouns with which
"
the verb to sit," in the first two lines (root form istua] agrees,
" " " "
while in the two last lines the adjectives gaudy and thick
also agree with the nouns. This is very clear in Example III.,
where the leading word is vaski = copper, and in the first line
the concord is with the letter i of the word, in the second with
 95

the s, with the k, in the fourth with the initial v,

in the third
and with both v and k, the initial letters of the two
in the last
syllables of Vas-ki, a very pretty piece of poetical alliteration.
In the IV. Example the concord agrees all through with the
leading idea in each line, though the word expressing it does
not in every case stand at the beginning of the line, but some-
" "
times, like karjan (" flock or here shoal of fishes "), stands
at the back.
We have thus here in Finnish something which comes very
near the Concord in Bantu, though as appears from the ex-
amples given, in Finnish that principle is not always carried
through with the unswerving consistency found in Zulu,
Kafir, and most of the other Bantu languages.
" "
It therefore, very probable that the
is, Concord as it
exists to-day in Bantu was developed and brought to its pre-
sent musical perfection within the Bantu itself, after the tribes
had settled in Africa. The facts that the Bantu had lost the
art of writing, that all great events were commemorated in
"
songs, and that in honour of their chief they sung his great
names," and his prowess, as the Zulu still did in the times of
Tjaka and Dingaan, and several Bantu tribes in Central Africa
are now accustomed to do, would greatly tend to develop the
Concord, which, as we have seen above, existed already in
Ugro-Altaic. That very fact of not possessing the art of writ-
ing, togetherwith the inborn love for musical sounds which the
Bantu have, was undoubtedly the cause that the Concord was
applied to the ordinary prose language of daily life, and thus
gradually became a great linguistic factor.
Under the present circumstances, and within the necessarily
narrow limits of this sketch of the Bantu origin, I cannot enter
more deeply upon this question of Concord, a matter which
really requires a volume in itself. For at present, the Concord
constitutes, as it were, the grammatical centre around which
the whole of the Bantu turns, and a consideration thereof would
lead us too far away out of my course. What has been said
above tends, however, to strengthen the bond which, I main-
tain, exists between Bantu and Ugro-Altaic.
In conclusion, a few words regarding this relation may not
be out of place.
To many it might seem absurd to connect the semi-, or some-
times wholly, savage Bantu with the highly-developed Finn,
or the energetic and pushing Japanese, but it should be remem-
bered that in the Finn, the Sirjcnian, the Votian and all the
other members of the Finnish group, we have probably a
strong Aryan element that in the Manchu and the Japanese
;

we have a specifically-developed Jakun element, which we

generally denominate Mongolian, but the history of which is
still shrouded in that, again, in the blood of the
mystery ;
 96
"
unspeakable Turk," and his relations, we have a strong
Semitic strain, and perhaps some Aryan drops.
The Bantu miss that foreign element. In them there is
probably no other blood than that of the Sakai and the Semang,
of the old Nagas or Lu, who ages ago populated India, but
who, after the Aryan conquest, seem to have made their
headquarters the present Beluchistan, and perhaps part of
Afghanistan. But before these Nagas could enter upon a new
development, the Bantu, who in any case seem to have be-
longed to a low Naga division, or may have even been out-
casts, left the country and started life anew in Africa. While in
the original home their brethren rose, step by step, in civilisa-
tion, and sent off branch after branch northwards to populate
nearly the whole of Asia north of the Himalaya, the Bantu in
Africa entered upon a career which could only have a most
baneful effect upon their development. In a strange country,
where Nature was fitful, where one year the earth produced
plenty, and the next even a drop of water might be a boon ;

having to struggle for an existence not only with Nature, but

also with human enemies, who surrounded them on all sides ;

split up in course of time in various divisions, whose sole object

in life was to prey upon each other, as to-day is still done
by his relatives in Beluchistan possessors of fertile lands
;

to-day, and wanderers in the desert to-morrow is it a wonder

that, under such circumstances, the Bantu sank lower and
lower in the scale of civilisation ? Low they did sink in some
;

cases they even reached cannibalism. Yet they never sank to the
depths that the Hottentot sank in the Bushman type they never
;

became isolated nomads, but always kept their tribal system in

full force. And it is this tribal system, with all its remarkable
institutions, social as well as political, which saved the Bantu
from utter deterioration. Immense charms as this aspect of the
Bantu has, I cannot discuss it here, but must leave that for a
more fitting opportunity.
In the above essay, an attempt has been made, clumsily
perhaps, to find a key which will open to us the secret of
Bantu philology not only, but the hidden mysteries of Bantu
character and Bantu institutions. As such this essay is, what
I have called it, a
preliminary work. The key I have tried to
find must be tested we must see whether it will fit upon the
;

locks of the great safe which holds what we wish to see and
observe for ourselves.
The few years of the rest of my life I intend to devote to this
task or, at all events, to some small part of it. Whether I shall
be enabled to carry out this intention depends, to a large extent,
upon others. For such work one must have not only leisure,
but one must be enabled to devote his whole energy to it. and
hence it is necessary that some provision should be made for
the sordid cares of mere animal existence.
 97

But the work cannot be done by one man, even if he had

half a century at his disposal. We want workers, earnest,
capable, and enthusiastic workers in this field we want a
;

number of them.
If this essay should be fortunate enough to induce some of
our smart young South Africans, who now grace the roll of our
I
"niversity calendars, to join the rank of workers in the study
of Bantu philology, Bantu institutions, and the kindred sub-
jects, the writer will consider himself amply rewarded for the
innumerable hours of close and hard study he has devoted to
this little work. Then only will he feel the truth of the motto
of a noble English family

PER ASPERA AD ASTRA.

1907.]
 University of Toronto
CO
W Library
0)
o
p
cd

-p
O

0)

<D O
<D C
Ct, bo
H
co o

h
O
O

H I>
O
< inerf

^O