Chapter I: The Physical Types

ἐπεὶ ἦ μάλα πολλὰ μεταξύ, οὔρεά τε σκιόεντα θάλασσά τε ἠχήεσσα Il. I. 156–7.

Ethnic isolation

IN respect of those decisive physical features which determine the course of the national movements of mankind, India may be described as an irregularly triangular or pear-shaped fortress, protected on two sides by the sea and guarded on the third by the great bulwark of mountain ranges of which the Himalaya forms the central and most impregnable portion.* As these ranges curve westward and southward towards the Arabian Sea, they are pierced by a number of passes, practicable enough for the march of unopposed armies, but offering small encouragement to the halting advance of family or tribal migration. On the east, though the conformation of the barrier is different, its secluding influence is equally strong. The ridges which take off from the eastern end of the Himalaya run for the most part north and south, and tend to direct the main stream of Mongolian colonization towards the river basins of Indo-China rather than towards India itself. On either frontier, where the mountains become less formidable, other obstacles intervene to bar the way. On the western or Iranian march the gap between the Suleiman range and the Arabian Sea is closed by the arid plateaux and thirsty deserts of Makran; to the east, the hills of the Turanian border rise in a succession of waves from a sea of trackless forest. On either side, again, at any rate within historic times, the belt of debatable land which veiled a dubious and shifting frontier has been occupied by races of masterless men knowing, in the west, no law save that of plunder and vendetta, and in the east, owning no obligation but the primitive rule that a man must prove his manhood by taking the stranger’s head. Along the coast line conditions of a different character tended equally to preclude immigration on a large scale. The succession of militant traders who landed on the narrow strip of fertile but malarious country which fringes Western India, found themselves cut off from the interior by the forest-clad barrier of the Western Ghâts; while on the eastern side of the peninsula, the low coast, harbourless from Cape Comorin to Balasore, is guarded by dangerous shallows backed by a line of pitiless surf.*

The country thus isolated by physical and historical causes comprises three main regions, the Himalaya or abode of snow; the Middle Land, or Madhyadesa, as the river plains of Northern India are called in popular speech; and the southern table-land of the Deccan with its irregular hill ranges rising out of undulating plains. Each region possesses an ethnic character of its own, and has contributed a distinct element to the making of the Indian people. The Deccan, itself one of the most ancient geological formations in the world, has, since the dawn of history, been the home of the Dravidians, the oldest of the Indian races. The most recent of the three regions, the alluvial plains of the north, formed in pre-historic times the highway of the Aryan advance into India, and a large section of its inhabitants still cherishes the tradition of remote Aryan descent. The influence of the Himalaya has been mainly negative. It has served as a barrier against incursions from the north, but all along the line of the hills, even among people whose speech is of Rajput origin, distinct traces may be observed of an intermixture of Mongolian blood.

The Empire of to-day has outgrown its ancient limits, and now embraces the Indo-Iranian region of Baluchistan and the Indo-Chinese region of Burma. If we speak of India as a fortress, these are the outworks which guard its flanks. Nor is it pressing metaphor too far to describe Baluchistan as a great natural glacis stretching westward from the crest of the ramparts of India till it loses itself in the plains of Kandahar. Its surface is a medley of rocky peaks, narrow passes, intricate ravines and broken ranges of barren hills, which bristle at every point with defensive positions. The people show no trace of Indian culture, and are as rugged as the land in which they dwell. Arab or Afghan by tradition, Scythian or Turki by type, but probably a blend of several stocks, they are fitting guardians of the inhospitable wastes which separate India from Iran.

The Eastern outpost, Burma, presents the sharpest of contrasts to Baluchistan. Broad stretches of alluvial rice-land fringe the coast strip and run up into the interior, gradually thinning out as they approach the highlands of earlier formation through which the great rivers have forced their way. Cut off from India by a series of forest-clad ranges, which restricted the interchange of population by land, Burma lay open on the north, east and south to the inroads of a succession of Mongolian races who bore rule in turn and combined to form the type which we know as Burmese. In the hands of a maritime power Burma commands the eastern gate of the Empire, and the growing Indian element in the population owes its existence to the English control of the sea.

External Factors.

These are the external factors of the problem of Indian ethnology. The main results of their influence are obvious enough. An unbroken chain of snow-clad peaks and of passes only practicable at certain seasons opposes an effectual obstacle to the fusion of contrasting types. Ranges of lower elevation, intersected by frequent valleys, form no bar to hostile incursions and yield but scanty protection to a weaker race. Long stretches of fertile plains, traversed by navigable rivers and lying open to the march of armies, lend themselves to that crushing out of racial distinctions which conquest brings in its train. Isolated hill ranges and lofty plateaux, guarded by fever-haunted forests and offering no prospect of profit or plunder, furnish an abiding refuge for tribes which are compact enough to emigrate en masse. Lastly, a coast line almost devoid of sheltering harbours, while it may invite a daring invader, fails to foster the maritime skill and enterprise which alone can repulse his landing.

Internal Factors.

For the internal factors—the races which lived and struggled within the environment roughly sketched above—we must depend to a great extent upon speculative data. Living organisms are more complex and less stable than their material surroundings. The hills may not be everlasting, as poets have imagined, but they outlive countless generations of men, and the changes that time works in their structure do impress on them some record, however imperfect, of processes which it has taken ages to complete. Man alone passes and leaves nothing behind. India in particular is conspicuous for the absence of the pre-historic evidence of which ethnologists in Europe have made such admirable use. There are no cave deposits, no sepulchral mounds or barrows, no kitchen middens, no lake dwellings, no ancient fortified towns such as modern research is now unearthing in Greece,* and no sculptured bones or weapons portraying the vicissitudes of the life of primitive man. The climate and the insects have obliterated all perishable vestiges of the past, and what nature may have spared a people devoid of the historic sense has made no effort to preserve. To fill the blank we are thrown back mainly on conjecture. Yet in India conjecture starts from a more solid basis than in the progressive countries of the Western world. For here we have before our eyes a society in many respects still primitive, which preserves, like a palimpsest manuscript, survivals of immemorial antiquity. In a land where all things always are the same we are justified in concluding that what is happening now must have happened, very much in the same way, throughout the earlier stages of human society in India. Observation of the present is our best guide to the reconstruction of the past.

The race basis of Indian society

On a stone panel forming part of one of the grandest Buddhist monuments in India, the great tope at Sānchi, a carving in low relief depicts a strange religious ceremony.* Under trees with conventional foliage and fruits, three women, attired in tight clothing without skirts, kneel in prayer before a small shrine or altar. In the foreground, the leader of a procession of monkeys bears in both hands a bowl of liquid and stoops to offer it at the shrine. His solemn countenance and the grotesquely adoring gestures of his comrades seem intended to express reverence, devotion, and humility. In the background four stately figures, two men and two women of tall stature and regular features, clothed in flowing robes and wearing elaborate turbans, look on with folded hands in apparent approval of this remarkable act of worship. Antiquarian speculation has for the most part passed the panel by unnoticed, or has sought to associate it with some pious legend of the life of Buddha. A larger interest, however, attaches to the scene, if it is regarded as the sculptured expression of the race sentiment of the Aryans towards the Dravidians, which runs through the whole course of Indian tradition and survives in scarcely abated strength to the present day. In this view the carving would belong to the same order of ideas as the story in the Rāmāyaṇa of the army of apes who assisted Rāma in the invasion of Ceylon. It shows us the higher race on friendly terms with the lower, but keenly conscious of the essential difference of type and taking no active part in the ceremony at which they appear as sympathetic but patronizing spectators. An attempt is made in the following pages to show that the race sentiment which inspired this curious sculpture, rests upon a foundation of facts which can be verified by scientific methods; that it supplied the motive principle of caste; that it continues, in the form of fiction or tradition, to shape the most modern developments of the system; and, finally, that its influence has tended to preserve in comparative purity the types which it favours.

It is a familiar experience that the ordinary untravelled European, on first arriving in India, finds much difficulty in distinguishing one native of the country from another. To his untrained eye all Indians are black; all have the same cast of countenance; and all, except the “decently naked” labouring classes, wear loose garments which revive dim memories of the attire of the Greeks and Romans. An observant man soon shakes off these illusions and realizes the extraordinary diversity of the types which are met with everywhere in India. The first step in his education is to learn to tell a Hindu from a Muhammadan. A further stage is reached when it dawns upon him that the upper classes of Hindus are much fairer than the lower and that their features are moulded on finer lines. Later on, if opportunity favours him, he comes to recognize at a glance the essential differences between the Punjabi and the Bengali, the Pathān and the Gurkha, the Rajput and the “Jungly” tea coolie: he will no longer take a Marāthā Brāhman for a Madrasi, or an Oriya for a native of Kashmir. He learns, in short, to distinguish what may be called the provincial types of the people of India, the local, racial, or linguistic aggregates which at first sight seem to correspond to the nations of Europe. But the general impressions thus formed, though accurate enough so far as they go, are wanting in scientific precision. They cannot be recorded or analyzed; no description can convey their effect; they melt away in the attempt to fix them, and leave nothing behind.

The Data of Ethnology.

The modern science of ethnology endeavours to define and to classify the various physical types, with reference to their distinctive characteristics, in the hope that when sufficient data have been accumulated it may be possible in some measure to account for the types themselves, to determine the elements of which they are composed, and thus to establish their connexion with one or other of the great families of mankind. In India, where historical evidence can hardly be said to exist, the data ordinarily available are of three kinds—physical characters, linguistic characters, and religious and social usages. Of these the first are by far the most trustworthy. Most anthropologists, indeed, are now inclined to adopt without much question the opinion of the late Sir William Flower, who wrote to me some years ago that “physical characters are the best, in fact the only true tests of race, that is, of real affinity; language, customs, etc., may help or give indications, but they are often misleading.”

The claims of language to share in the settlement of questions of race cannot, however, be dismissed in a single sentence.

Language and Race.

Nearly twenty years ago, when the ethnographic survey of Bengal was in progress, the late Professor Max Müller sent me a long letter, since published in his collected works, in which he protested against “the unholy alliance” of the two sciences of ethnology and comparative philology. At first sight it is hard to understand why two lines of research, dealing with different subjects and working towards different ends, should be charged with nefarious collusion for the purpose of perverting the truth. A clue to the grounds of the accusation is, however, furnished by Sir Henry Maine’s remark that the study of the sacred languages of India has given to the world “the modern science of Philology and the modern theory of Race.” The study of Sanskrit received its first impetus from the publication by Sir William Jones of translations of Kālidāsa’s Sakuntala in 1789 and of the Institutes of Manu in 1794.* The discovery was announced and its importance emphasised in Friedrich von Schlegel’s treatise on the Language and Wisdom of the Hindus; but even with this assistance the fresh ideas took more than a generation to spread beyond the narrow circle of Orientalists and to impress themselves upon the main current of European thought. The birth of a new science, based upon an ancient language of which most people then heard for the first time, was inaugurated by Friedrich Bopp’s Comparative Grammar of the Indo-European languages. The editions of this work extend over the period 1833—1852, so that the beginnings of Comparative Philology coincide in point of time with the popular upheaval which found expression in the revolutionary movements of 1848. The belief that linguistic affinities prove community of descent was one which commended itself alike to populations struggling for freedom and to rulers in search of excuses for removing a neighbour’s landmark. The old idea of tribal sovereignty seemed almost to have revived when Napoleon III. assumed the title of Emperor of the French and justified his annexation of Savoy by the plea that territory where French was spoken ought to belong to France. As the principle gained strength and was invoked on a larger scale it gave rise to the political aspirations implied in the terms Pan-Teutonism, Pan-Hellenism, Pan-Slavism; it helped the cause of German unity; it was appealed to in the name of united Italy; and, if carried to its logical conclusion, it may some day contribute to the disruption of the Austrian Empire.

Thus we find Comparative Philology, in the hands of ardent patriots and astute diplomatists, trespassing on the domain of ethnology and confusing for political purposes the two distinct conceptions of race and nationality. But the ethnologists themselves were not free from blame. So far from resisting the encroachment on their territory they lent their authority to the prevailing tendency and based their classification of races mainly upon linguistic characters. For this they may well be held to have had some substantial excuses. In the first place linguistic data are far easier to collect on a large scale, and far easier to examine when collected, than the physical observations which form the main basis of ethnological conclusions. The vast array of languages and dialects which fill the sixteen volumes of Dr. Grierson’s Linguistic Survey of India was brought together from the most distant corners of the Empire by the simple device of circulating for translation the parable of the Prodigal Son (the fatted calf, in deference to Hindu sentiment, being discreetly transformed into a goat), together with a small number of common words and phrases. But to have recorded the physical characters of the people on a similar scale would have cost an immense sum; the operations would have extended over many years; and the results would probably have been vitiated by the personal divergencies of the numerous observers whom it would have been necessary to employ.

Secondly, languages lend themselves far more readily to precise classification than the minute variations of form and feature which go to make up an ethnic type. Thirdly,—and this is perhaps the most important point of all—while there are practically no mixed languages, there are hardly any pure races. Judged by the only sound test, that of grammatical structure as distinguished from mere vocabulary, all languages may be regarded as true genera and species from which no hybrid progeny can arise. Words may be borrowed on a larger or smaller scale, but the essential structure of the language remains unchanged, the foreign elements being forced into an indigenous mould. Thus French people who have taken to afternoon tea have evolved the verb “five o’cloquer”; a Bengali clerk who is late for office will say āmi miss-train kariyāchhi, converting a mangled English phrase into a characteristic verbal noun; and a Berlin tram-conductor, who was explaining to me how his working hours had come to be reduced, summed up the position with the words “wir haben nämlich streikirt.” In each case a foreign phrase has been taken to express an imported idea; but this phrase has been absorbed and dealt with in accordance with the genius of the language, and there is no approach to structural hybridism. Races, on the other hand, mix freely; they produce endless varieties; and it can hardly be said even now that any satisfactory agreement has been arrived at as to the system on which such varieties should be classified.

These considerations go some way towards accounting for the “unholy alliance” which politics and the spirit of classification have combined to bring about between two distinct sciences. They fail, however, to give us much assistance in the solution of the main question—what are the true relations between Ethnology and Philology? Within what limits can we argue from correspondences of language to community of race or from differences of language to diversity of race? Are we to hold with Schwiker and Hale that language is the only true test of racial affinities; or should we follow Sayce’s opinion that “identity or relationship of language can prove nothing more than social contact”? The mere fact that speech is a physiological function, depending in the last resort on the structure of the larynx, suggests that the latter view may be too absolutely expressed. That some races produce sounds which other races can only imitate imperfectly is a matter of common observation, and may reasonably be ascribed to differences of vocal machinery. The clicks of the Bushman and Hottentot, the gutturals of Arabic and the dental and cerebral consonants of the Indian vernaculars present varying degrees of difficulty to the average European. Similar differences of phonetic capacity may be observed among the Indian races. Bengalis, as Dr. Grierson has pointed out, “cannot pronounce a clear s but make it sh”; the natives of Western India tend to turn v into w; and nearly all Orientals find a difficulty in starting a word like Smith without prefixing a vowel and turning it into “I-Smith.” Even within the range of a single language, dialectic variations occur which may be due to physical causes. The gobbling speech of the people of Chittagong and Eastern Bengal, and their inability to negotiate certain consonants, seem to suggest that their original tongue belonged to the Tibeto-Burman family, and that their vocal apparatus must differ materially from that of their Western neighbours.* Whether it will ever be possible to define these variations, and to correlate them with racial characteristics, is a question for students of the physiological side of the modern science of phonetics.

The truth as to the relation between race and language probably lies somewhere between the extreme views noticed above, but it can only be reached by an examination of the facts. There are four possible cases:— (1) where both language and physical type have been changed by contact with other races or communities, as have happened with the Bengali-speaking Kochh, who have lost their tribal language while their original Mongoloid type, still clearly discernible among their congeners in Assam, has been modified by intermixture with a Dravidian element ; (2) where the language has changed but the racial type has remained the same, as with the Gauls, Normans, and Lombards in Europe, the Negroes in America, and the Ahoms, Bhumij and many others in India ; (3) where the original language has been retained but the racial type has changed, as with the Basques and Magyars in Europe, the Khas in Nepal, and a large proportion of the Rajputs all over India ; (4) where both language and physical type are unchanged, as with the Andamanese, the Santals, the Mundas, the Manipuris and many others.

In the first two cases an appeal to language would clearly be ineffectual unless historical evidence were forthcoming to show what the original language had been. In India the genius loci has not turned to history, and almost the only instance in which ancient records throw light upon the origin of a tribe is that of the Ahoms, a Shan people who entered Assam early in the thirteenth century and within the next three hundred years conquered and gave their name to the country. Towards the end of the seventeenth century they embraced Hinduism, lost their original language, and “became, like Brāhmans, powerful in talk alone.” Their chronicles (buranji or “store of instructions for the ignorant”) were kept up by their priests in Ahom, “an old form of the language which ultimately became Shan,” and are the chief authority for the early history of Assam.

To the remaining two cases we may apply a canon which I suggested to Dr. Grierson some two years ago, and which he has embodied in his chapter on Language in the Census Report of 1901. I would now state it somewhat more fully thus :—

(1) In areas where several languages are spoken, one or more of them will usually be found to be gaining ground, while others are stationary or declining: the condition of stable equilibrium is comparatively rare. The former may be described in relation to any given area as dominant, the latter as decadent or subordinate languages. What languages belong to either class is, in each case, a matter of observation. (2) The fact that a particular tribe or people uses a dominant language does not of itself suggest any inference as to their origin. (3) The fact that such a group speaks a decadent language may supply evidence of their origin, the value of which will vary with circumstances.

It must be admitted, however, that these propositions do not carry us very far, and that in their application to particular cases they tend to break down just at the point where the enquiry begins to be interesting. Of course it is obvious enough that the fact that the Rājbansi-Kochh and the Bhumij both speak Bengali does not prove them to be of Indo-Aryan descent. On this point their physical type would be conclusive, even if we had not independent evidence that a few generations ago they spoke tribal languages of their own. Similarly, when one finds two small and isolated communities in Bengal, the Siyālgirs of Midnapur and the Kichaks of Dacca, speaking Bhil dialects of Gujarātī, one is naturally disposed to infer that these people must have come from Gujarāt, and are probably related in some way to the Bhils. But here again there is room for doubt. Although both Kichaks and Siyālgirs are now of settled habits, the traditions of the former, and the usages and occupations of both, suggest that at no very distant date they formed part of that miscellaneous multitude of gipsy folk whose origin is no less of a mystery in India than in other parts of the world. To people of their habits—the Kichaks say that their ancestors were dacoits, and the Siyālgirs are credited with thievish proclivities—the possession of a special argot would be an obvious convenience, and it seems simpler to suppose that this circumstance led to the wide diffusion of the dialect than to argue that the small groups which make use of it in Bengal must be fragments of a distant and compact tribe like the Bhils. Thieves’ patters have a family likeness all the world over, but no one has yet attempted to trace the speakers to a common ancestor.

Other minor instances deserve passing mention. The Vaidu herbalists of Poona, who speak Marāthi to their neighbours, explain the fact that they use Kanarese among themselves by the tradition that they were brought from the Kanara country by one of the Peshwas and settled in Kirki. The Kāsār coppersmiths of Nasik speak Gujarāti at home and Marāthi out of doors. The men dress like Marāthas, but the women still wear the characteristic petticoat (ghāgra) of Gujarāt instead of the Marātha sāri. In both these cases linguistic evidence points to a migration; but the value of the deduction is small. For we know historically that the migration must have been a recent one and it could probably be established on independent grounds. Nor do linguistic considerations throw any light upon the curious question how it is that the Mundas and Oraons, two distinct tribes of identical physical type, speak languages which differ widely in respect of structure and vocabulary.

But perhaps the most notable illustration of the weakness of the argument from affinity of language to affinity of race is afforded by Brāhūi. One of the maps in Dr. Grierson’s chapter on language in the Census Report for India in 1901, shows the distribution of the Dravidian languages. Most of the Dravidian-speaking areas are massed in the south of India, while a few outlying patches represent Gond in the Central Provinces and Kandh, Kurnkh, and Malto in Bengal. Otherwise the map is blank save for Brāhūi, a tiny island of Dravidian speech far away in Baluchistan where it is surrounded on all sides by Indo-Aryan languages. As to the Dravidian affinities of the Brāhūi language, I understand that there is practical agreement among linguistic authorities. Concerning the conclusions to be drawn from this fact opinions differ widely. One school founds upon it the hypothesis that the Dravidians entered India from beyond the north-west frontier, while another regards the Brāhūi as an outpost of the main body of Dravidians in Southern India. Both assume identity of race, and both ignore the essential fact that, as is shown at length below, few types of humanity can present more marked physical differences than the Brāhūi and the Dravidian. How then can we explain the resemblances of language? Surely only by assuming that at some remote period the two races must have been in contact and that the speech of one influenced that of the other. Thus what seems at first sight to be a crucial instance serves merely to bring out the uncertainty that besets any attempt to argue from language to race. Here, if anywhere, is a decadent and isolated language; here, if anywhere, it ought to tell a plain tale; and here, when confronted with other evidence, it conspicuously fails us. Thus we end very much where we began, with the rather impotent conclusion that in questions of racial affinity, while the testimony of language should certainly be considered, the chances are against its telling us anything that we did not know already from other and less dubious sources.

Indefinite physical characters.

For ethnological purposes physical characters may be said to be of two kinds—indefinite characters which can only be described in more or less appropriate language, and definite characters which admit of being measured and reduced to numerical expression. The former class, usually called descriptive or secondary characters, includes such points as the colour and texture of the skin; the colour, form, and position of the eyes; the colour and character of the hair; and the form of the face and features. Conspicuous as these traits are, the difficulty of observing, defining, and recording them is extreme. Colour, the most striking of them all, is perhaps the most evasive, and deserves fuller discussion as presenting a typical instance of the shortcomings of the descriptive method. Some forty years ago the French anthropologist Broca devised a chromatic scale consisting of twenty shades, regularly graduated and numbered, for registering the colour of the eyes, and thirty-four for the skin. The idea was that the observer would consult the scale and note the numbers of the shades which he found to correspond most closely with the colouring of his subjects. Experience, however, has shown that with a scale so elaborate as Broca’s the process of matching colours is not so easy as it looks; that different people are apt to arrive at widely different conclusions; and that even when the numbers have been correctly registered no one can translate the result of the observations into intelligible language. For these reasons Broca’s successor Topinard reverted to the method of simple description, unaided by any scale of pattern colours. He describes, for example, the mud-coloured hair so common among the peasants of Central Europe as having the colour of a dusty chestnut. In the latest edition of the Anthropological Notes and Queries published under the auspices of the British Association, an attempt is made to combine the two systems. A greatly simplified colour scale is given, and each colour is also briefly described. I doubt, however, whether it is possible to do more than to indicate in very general terms the impression which a particular colour makes upon the observer. In point of fact the colour of the skin is rather what may be called an artistic expression, dependent partly upon the action of light, partly on the texture and transparency of the skin itself, and partly again on the great variety of shades which occur in every part of its surface. It is hopeless to expect that this complex of characters can be adequately represented by a patch of opaque paint which is necessarily uniform throughout and devoid of any suggestion of light and shade.

The difficulty which besets all attempts to classify colour is enhanced in India by the fact that, for the bulk of the population, the range of variation, especially in the case of the eyes and hair, is exceedingly small. The skin, no doubt, exhibits extreme divergencies of colouring which any one can detect at a glance. At one end of the scale we have the dead black of the Andamanese, the colour of a blackleaded stove before it has been polished, and the somewhat brighter black of the Dravidians of Southern India, which has been aptly compared to the colour of strong coffee unmixed with milk. Of the Irulas of the Nilgiri jungles, some South Indian humourist is reported to have said that charcoal leaves a white mark upon them. At the other end one may place the flushed ivory skin of the typical Kashmiri beauty and the very light transparent brown—“wheat-coloured” is the common vernacular description—of the higher castes of Upper India, which Emil Schmidt compares to milk just tinged with coffee and describes as hardly darker than is found in members of the swarthier races of Southern Europe. Between these extremes we find countless shades of brown, darker or lighter, transparent or opaque, frequently tending towards yellow, more rarely approaching a reddish tint, and occasionally degenerating into a sort of greyish black which seems to depend on the character of the surface of the skin. It would be a hopeless task to attempt to register and to classify these variations. Nor, if it were done, should we be in a position to evolve order out of the chaos of tints. For even in the individual minute gradations of colour are comparatively unstable, and are liable to be affected not only by exposure to sun and wind, but also by differences of temperature and humidity. Natives of Bengal have assured me that people of their race, one of the darkest in India, become appreciably fairer when domiciled in Hindustan or the Punjab; and the converse process may be observed not only in natives of Upper India living in the damp heat of the Ganges delta, but in Indians returning from a prolonged stay in Europe, who undergo a perceptible change of colour during the voyage to the East. The fair complexion of the women of the shell-cutting Saṅkāri caste in Dacca is mainly due to their seclusion in dark rooms, and the Liṅgāyats of Southern India who wear a box containing a tiny phallus tied in a silk cloth round the upper arm, show, when they take it off, a pale band of skin contrasting sharply with the colour of the rest of the body.

Still less variety is traceable in the character of the eyes and hair. From one end of India to the other, the hair of the great mass of the population is black or dark brown, while among the higher castes the latter colour is occasionally shot through by something approaching a tawny shade. Straight hair seems, on the whole, to predominate, but hair of a wavy or curly character appears in much the same proportion as among the races of Europe. The Andamanese have woolly or frizzy hair, oval in section and curling on itself so tightly that it seems to grow in separate spiral tufts, while in fact it is quite evenly distributed over the scalp. Although the terms woolly and frizzy have been loosely applied to the wavy hair not uncommon among the Dravidians, no good observer has as yet found among any of the Indian races a head of hair that could be correctly described as woolly. Throughout India the eyes are almost invariably dark brown. Occasional instances of grey eyes are found among the Konkanasth Brāhmans of Bombay, and the combination of blue eyes, auburn hair, and reddish blonde complexion is met with on the northwestern frontier. On the Malabar coast in the south, Mr. Thurston had noticed several instances of pale blue and grey eyes combined with a dark complexion and has even seen a Syrian Christian baby of undoubted native parentage with bright carroty hair. The Syrian Christians of South Travancore say, indeed, that they differ from Northerners in having a red tinge to the moustache.

Definite physical characters.

When we turn to the definite or anthropometric characters we find ourselves upon firmer ground. The idea of applying instruments of precision to the measurement of the human body was familiar to the Egyptians and the Greeks, both of whom appear to have made extensive experiments with the object of arriving at a “canon” or ideal type, showing the proportions which various parts of the body should bear to the entire figure and to each other. Such canons were usually expressed either in terms of a particular member of which the rest were supposed to be multiples, or in fractional parts of the entire stature. Thus, according to Lepsius, the Egyptian canon is based on the length of the middle finger and this measure is supposed to be contained nineteen times in the full stature, three times in the head and neck, eight times in the arm, and so forth. The Greek canon, on the other hand, as restored by Quetelet, expresses the limbs and other dimensions in thousandth parts of the entire stature. Concerning this canon a curious story is told by Topinard, not without interest in its bearings upon the relations of Egyptian and Greek art. In 1866, the eminent French anthropologist Broca was asked on behalf of an artist who was engaged in the attempt to reconstruct the Greek standard, to provide a skeleton corresponding in its proportions to certain measurements derived from an examination of the Belvedere Apollo. After some search Broca found in the Museum of the Anthropological Society at Paris a skeleton of the type required. It was that of a Soudanese negro named Abdullah, and from this Broca concluded that the famous statue of Apollo had been modelled on the Egyptian canon, which in his opinion had been derived by Egyptian sculptors from the study of the Nubian negroes whom they employed as models.

The Roman canon handed down in the treatise De Architectura of Vitruvius was taken up and developed in the early days of the Renaissance by Leo Battista Alberti, himself, like Vitruvius, an architect, and a curious enquirer into the secret ways of nature and of the human frame. Forty years later Leonardo da Vinci, in his Trattato della pittura, expressed the general opinion that the proportions of the body should be studied in children and adults of both sexes, and refuted the opinion of Vitruvius that the navel should be deemed the centre of the body. Following Leonardo’s suggestions, Albrecht Dürer addressed himself to the task of working out the proportions of the body for different ages and sexes, for persons of different heights, and for different types of figure. In his “Four books on the proportions of the human figure,” published at Nürnberg in 1528, the year of his death, Dürer discussed the difficult question of the so-called “orientation” or adjustment of the head in an upright position, and he is believed by the authors of the Crania ethnica to have also anticipated Camper’s invention of the facial angle. Jean Cousin, a French contemporary of Dürer’s, took the nose as his unit of length and represented the ideal head as measuring four noses, and the ideal stature as equivalent to eight heads or thirty-two noses. Cousin’s system, slightly modified by Charles Blanc, holds its own at the present day as the canon des ateliers of French artists, preference, however, being given in ordinary parlance to the head rather than the nose as the unit of length.

The data now available.

All these canons, it will be observed, approach the subject purely from the artistic point of view; and so far from taking account of the distinctive characters of particular races, incline to sink these in the attempt to frame a general canon of the proportions of the body which should hold good for the whole of mankind. Such an endeavour would be foreign to the purpose of anthropology, which fixes its attention on points of difference rather than of resemblance, and seeks by examination and analysis of such differences to form hypotheses concerning the genesis of the distinct race stocks now in existence. It would perhaps be fanciful to trace the germs of anthropometric research in the statement of Herodotus that the skulls of the Persian soldiers slain at the battle of Platæa were thin, and those of the Egyptians were thick, or to cite his explanation, that the former lived an indoor life and always wore hats, while the latter shave their heads from infancy and exposed them to sun without covering, as the earliest instance of the modern scientific doctrine of the influence of external conditions. But when Ctesias speaks of the small stature, black complexion, and snub noses of the inhabitants of India, we feel that the description is precise enough to enable us to identify them with the Dasyus and Nishādas of early Sanskrit literature, and we are almost tempted to wonder whether the Greek physician, who was doubtless acquainted with the canon of Polycletus, may not have devised some accurate method of recording the racial characteristics of which he was so close an observer. Curiously enough the famous potter, Bernard de Palissy, was the first to throw out, in a humorous dialogue published in 1563, the idea of measuring the skull for purposes other than artistic. The passage quoted by Topinard is too quaint to be omitted here:—“Quoy voyant il me print envie de mesurer la teste d’un homme pour sçavoir directement ses mesures, et me semble que la sauterelle, la régle, et le compas me seroient fort propres pour cest affaire, mais, quoy qu’il en soit, je n’y sceu jamais trouver une mesure osseuse, parce que les folies qui estaient en ladite teste luy faisaient changer ses mesures.”

Method of treatment adopted.

Palissy, however, cannot be seriously put forward as the founder of scientific craniometry, and that title perhaps most properly belongs to the Swedish naturalist, Anders Retzius, who in 1842 hit upon the device of expressing one of the chief characters of the skull by the relation of its maximum breadth to its maximum length, the latter being taken to be one thousand. In this way he distinguished two forms of skull—the dolicho-cephalic, or long-headed type, in which the length exceeds the breadth by about one-fourth, and the brachy-cephalic, or short-headed type, in which the length exceeds the breadth by a proportion varying from one-fifth to one-eighth. Thus according to Retzius the Swedes are long-headed in the proportion 773:1000, and the Lapps short-headed in the proportion 865:1000. He also distinguished two types of face—the orthognathic, in which the jaws and teeth project either not at all, or very little beyond a line drawn from the forehead, and the prognathic, in which this projection is very marked. His classification of races was based upon these characteristics. In 1861 Broca improved Retzius’ system by expressing it in hundredths instead of thousandths, by introducing an intermediate group, called mesati-cephalic or medium-headed and ranging from 77.7 to 80 per cent., and by giving the name of cephalic index to the relation between the two diameters. Numerous other measurements, which are described in the literature of the subject, have since been introduced.

Craniometry and anthropometry.

In the earlier days of anthropology, it was natural that the attention of students should have been directed mainly to the examination of skulls. Craniometry seemed to offer a solution of the problems regarding the origin and antiquity of the human race which then divided the scientific world. Its precise method promised to clear up the mystery of the prehistoric skulls discovered in the quaternary strata of Europe, and to connect them on the one side with a possible Simian ancestor of mankind and on the other with the races of the present day. The latter line of research led on to the measurements of living subjects, which have since been undertaken by a number of enquirers on a very large scale. Anthropometry which deals with living people, while craniometry is concerned exclusively with skulls, possesses certain advantages over the elder science. For reasons too technical to enter upon here, its procedure is in some respects less precise and its results less minute and exhaustive than those of craniometry. These minor shortcomings are, however, amply made up for by its incomparably wider range. The number of subjects available is practically unlimited; measurements can be undertaken on a scale large enough to eliminate, not merely the personal equation of the measurer, but also the occasional variations of type arising from intermixture of blood; and the investigation is not restricted to the characters of the head, but extends to the stature and the proportions of the limbs. A further advantage arises from the fact that no doubts can be cast upon the identity of the individuals measured. In working with skulls, whether prehistoric or modern, this last point has to be reckoned with. The same place of sepulture may have been used in succession by two different races, and the skulls of conquering chiefs may be mixed with those of alien slaves or of prisoners slain to escort their captors to the world of the dead. The savage practice of head-hunting may equally bring about a deplorable confusion of cranial types; famine skulls may belong to people who have wandered from no one knows where; and even hospital specimens may lose their identity in the process of cleaning. In the second of his elaborate monographs on the craniology of the people of India Sir William Turner observes * that among the Oriya skulls belonging to the Indian Museum, which were lent to him for examination, some crania partake “of Dravidian, others of Aryan characters,” while in others again there is “a trace of Mongolian or other brachy-cephalic intermixture.” He surmises, therefore, that “no proper history of the dead had been obtained, and that in consequence the skulls had not been accurately identified.” As a matter of fact most of these skulls were acquired during the Orissa famine of 1866, and the only description they bear is “Oriya” or “Orissa,” the word “Hindu” being occasionally added. To any one who is acquainted with the conditions which prevailed in Orissa at that time it is obvious that a given skull may have belonged to a broad-nosed Dravidian from the hill tracts, to a high caste Hindu of the coast strip, or to a Mongoloid pilgrim from Nepal who died of starvation or cholera while seeking salvation at Jagannath. The characters of the skulls themselves render it probable that all of these indefinite groups are represented in the collection.

Anthropometry in India.

Scientific anthropometry was introduced into India on a large scale twenty years ago in connexion with the ethnographic survey of Bengal then in progress. The survey itself was a first attempt to apply to Indian ethnography the methods of systematic research sanctioned by the authority of European anthropologists. Among these the measurement of physical characters occupies a prominent place, and it seemed that the restrictions on intermarriage, which are peculiar to the Indian social system, would favour this method of observation, and would enable it to yield peculiarly clear and instructive results. A further reason for resorting to anthropometry was the fact that the wholesale borrowing of customs and ceremonies which goes on among the various social groups in India makes it practically impossible to arrive at any certain conclusions by examining these practices. Finally, the necessity of employing more precise methods was accentuated by Mr. Nesfield’s * uncompromising denial of the truth of “the modern doctrine which divides the population of India into Aryan and aboriginal,” and his assertion of the essential unity of the Indian race, enforced as it was by the specific statements that “the great majority of Brahmans are not of lighter complexion or of finer and better bred features than any other caste,” and that a stranger walking through the class rooms of the Sanskrit College at Benares “would never dream of supposing” that the high caste students of that exclusive institution “were distinct in race and blood from the scavengers who swept the roads.” A theory which departed so widely from literary tradition, from the current beliefs of the people, and from the opinions of most independent observers called for the searching test which anthropometry promised to furnish, and the case was crucial enough to put the method itself on its trial. The experiment has been justified by its results.

In 1890 I published in the Journal of the Anthropological Institute,* under the title “The Study of Ethnology in India,” a summary of the measurements of eighty-nine characteristic tribes and castes of Bengal, the United Provinces of Agra and Oudh, and the Punjab. These measurements were taken in accordance with a scheme approved by the late Sir William Flower of the British Museum and Professor Topinard of Paris. Topinard’s instruments were used, and his instructions were closely followed throughout. Analysis of the data rendered it possible to distinguish in the area covered by the experiment three main types, which were named provisionally Aryan, Dravidian, and Mongoloid. The characteristics of these types will be discussed fully below. Here it is sufficient to remark that the classification was accepted at the time by Flower, Beddoe, and Haddon in England, by Topinard in France, and by Virchow, Schmidt, and Kollmann in Germany. It has recently been confirmed by the high authority of Sir William Turner, who has been led by the examination of a large number of skulls to the same conclusions that were suggested to me by measurements taken on living subjects, and has been good enough to quote and adopt my descriptions of the leading types in his monographs† on the subject. Similar confirmation is furnished in the case of the Punjab by the craniometric researches of Lieutenant-Colonel Sir Havelock Charles.‡ Great additions have since been made to the number of measurements on living subjects by the exertions of Mr. Edgar Thurston, Superintendent of Ethnography for Southern India, under the comprehensive scheme of research sanctioned by Lord Curzon; by Sir T. H. Holland, Director of the Geological Survey of India, who has contributed important data for the Coorgs and Yeruvas of Southern India and the Kanets of Kulu and Lahoul; * by my anthropometric assistants, Rai Sahib Kumud Behāri Sāmanta and Mr. B. A. Gupte, who have carried out under my instructions an extensive series of measurements in Baluchistan, Rajputana, Bombay, Orissa, and Burma; and by Lieutenant-Colonel Waddell, C.B., C.I.E., of the Indian Medical Service, who has published some valuable data for Assam, and parts of Bengal in the Journal of the Asiatic Society of Bengal.

It is clearly impossible, within the compass of this sketch, to enter upon a full analysis of all the measurements which have been collected. I have therefore selected three characters, the proportions of the head, the proportions of the nose, and the stature, and have included them in the tables appended to this volume. For two groups I have also taken the orbito-nasal index, which affords a very precise test of the comparative flatness of face, determined mainly by the prominence or depression of the root of the nose in relation to the bones of the orbit and cheek, which is a distinctive characteristic of the Mongolian races. The measurements are arranged under the seven types, into which I now propose to divide the population; in every case the average and the maximum and minimum indices or dimensions are shown; and for each type diagrams are given, showing the seriation of the data for the tribes or castes selected as characteristic of the type. It need hardly be added that the conclusions which I have ventured to put forward are necessarily provisional, and will be of use mainly as a guide to research, and as an indication of the progress made up to date in this line of enquiry. During the next few years the data will be greatly added to by the ethnographic survey, and we may then hope to be in a position to make some approach to a final classification of the people of India on the basis of their physical characters.

General classification of mankind: the three primary types.

Meanwhile, it may be of service to point out that no natural classification of the varieties of the human species has as yet been arrived at. Certain extreme types can, of course, be readily distinguished. No one can fail to recognize the enormous structural differences between an Andamanese and a Chinaman, an Englishman, and a Negro, or a Patagonian and a Hottentot. But owing to the tendency of individuals to vary, and to the intermixture of races, which has gone on more or less at all times, and is continually increasing with modern improvements in communications, the apparently impassable gulf between the extreme types is bridged over by a number of intermediate or transitional forms, which shade into each other by almost imperceptible degrees. It is therefore practically impossible to divide mankind into a number of definite groups in one or other of which every individual will find a place. Even as regards the primary groups there has been great diversity of opinion, and the number suggested by different writers ranges from two to more than sixty. In the main, however, as Flower has pointed out, there has always been a tendency to revert to the four primitive types sketched out by Linnæus—the European, Asiatic, African, and American, reduced by Cuvier to three by the omission of the American type. Flower himself is of opinion “that the primitive man, whatever he may have been, has, in the course of ages, divaricated into three extreme types, represented by the Caucasian of Europe, the Mongolian of Asia, and the Ethiopian of Africa,” and “that all existing individuals of the species can be ranged around these types, or somewhere or other between them.” He therefore adopts as the basis of his classification the following three types:— I. The Ethiopian, Negroid, or black type with dark or nearly black complexion; frizzly black hair, a head almost invariably long (dolicho-cephalic); a very broad and flat nose; moderate or scanty development of beard; thick, everted lips; large teeth; and a long forearm.

The Negroid type is again sub-divided into four groups, with only one of which we are concerned here. This is the Negrito, represented within the Indian Empire by the Andamanese enumerated for the first time in the Census of 1901 and possibly by the Semangs of the jungles of Malacca, some of whom may have wandered up into the Mergui district of Burma.* In respect of colour and hair, the Andamanese closely resemble the Negro, but they have broad heads, their facial characters are different, and they form a very distinct group which has not been affected by intermixture with other races.

II. The Mongolian, Xanthous, or yellow type, with yellow or brownish complexion. These races have coarse straight hair without any tendency to curl; they are usually beardless or nearly so; they are mostly broad-headed; the face is broad and flat with projecting cheek-bones; the nose small, and conspicuously depressed at the root; the eyes sunken and the eyelids peculiarly formed so as to give the eye itself the appearance of slanting downwards; the teeth of moderate size.

The Northern or Mongolo-Altaic group of Mongolians includes the nomadic races of Central Asia whose influence on the population of India will be discussed later on. The Tibetans and Burmese are members of the Southern Mongolian group.

III. The Caucasian, or white type, has usually a fair skin; hair fair or dark, soft, straight or wavy; beard fully developed; the head-form is long or medium; the face narrow; the nose narrow and prominent; the teeth small and the forearm short.

Following Huxley, Flower divides the Caucasians into two groups:—

(a) The Xanthochroi or blonde type, with fair hair, light eyes and fair complexion. They “chiefly inhabit Northern Europe, but, much mixed with the next type, they extend as far as Northern Africa and Afghanistan.”

(b) Melanochroi, “with black hair and eyes, and skin of almost all shades from white to black.” Flower includes in this group not only the great majority of the inhabitants of Southern Europe, Northern Africa, and South-West Asia, consisting mainly of the Aryan, Semitic, and Hamitic families, but also the Dravidians of India, and the Veddahs of Ceylon.

Here we are confronted at once with the drawbacks which attend all attempts at systematic arrangement. It is difficult not to distrust a classification which brings together in the same category people of such widely different appearance, history, and traditions as the modern Greeks and Italians, and the black, broad-nosed Dravidians of Central and Southern India. Peschel’s arrangement seems to be in closer accordance with the facts established by recent observations. He divides the Caucasian type into (a) Indo-Germans, (b) Semites, (c) Hamites or Berbers, and includes the “Hindus” (non-Dravidian Indians) in the first of these groups. The Dravidians are classed with Sinhalese and Veddahs as people of uncertain origin. Huxley treats them as Australoid.

Their application to India.

In respect of classification the general position in India is closely parallel to that described above. It is easy enough to distinguish certain well-marked types. Our difficulties begin when we attempt to carry the process of classification further and to differentiate the minor types or subtypes which have been formed by varying degrees of intermixture between the main types. The extremes of the series are sharply defined, but the intermediate types melt into each other, and it is hard to say where the dividing line should be drawn. Here measurements are of great assistance, especially if they are arranged in a series so as to bring out the relative preponderance of certain characters in a large number of the members of particular groups. This is well illustrated by the diagrams in Appendix III., and will be more fully dwelt upon below. We are further assisted by the remarkable correspondence that may be observed at the present day in all parts of India, except the Punjab, between variations of physical type and differences of grouping and social position. This, of course, is due to the operation of the caste system, which in its most highly developed form, the only form which admits of precise definition, is, I believe, entirely confined to India.

Conditions favourable to anthropometry.

Nowhere else in the world do we find the population of a large continent broken up into an infinite number of mutually exclusive aggregates, the members of which are forbidden by an inexorable social law to marry outside of the group to which they themselves belong. Whatever may have been the origin and the earlier developments of caste, this absolute prohibition of mixed marriages stands forth now as its essential and most prominent characteristic, and the feeling against such unions is so deeply engrained in the people that even the theistic and reforming sect of the Brahmo Samāj has found a difficulty in freeing itself from the ancient prejudices, while the Lingāyats of Western and Southern India have transformed themselves from a sect into a caste within recent times. In a society thus organized, a society putting an extravagant value on pride of blood and the idea of ceremonial purity, differences of physical type, however produced in the first instance, may be expected to manifest a high degree of persistence, while methods which seek to trace and express such differences find a peculiarly favourable field for their operations. In this respect India presents a remarkable contrast to most other parts of the world, where anthropometry has to confess itself hindered, if not baffled, by the constant intermixture of types obscuring and confusing the data ascertained by measurements. Thus in Europe, as Topinard observes, there is nothing to prevent the union “of the blonde Kymri with the dark-haired dweller on the Mediterranean, of the broad-headed Celt with the long-headed Scandinavian, of the tiny Laplander with the tall Swede.” In fact, all the recognized nations of Europe are the result of a process of unrestricted crossing which has fused a number of distinct tribal types into a more or less definable national type. In India the process of fusion has long ago been arrested, and the degree of progress which it had made up to the point at which it ceased to operate is expressed in the physical characteristics of the groups which have been formed. There is consequently no national type and no nation or even nationality in the ordinary sense of these words.

Shape of the head.

The measurements themselves require a few words of explanation, which will be given in as popular language as the nature of the subject permits. The form of the head is ascertained by measuring in a horizontal plane the greatest length from a definite point on the forehead (the glabella) to the back of the head, and the greatest breadth a little above the ears. The proportion of the breadth to the length is then expressed as a percentage, called the cephalic index, the length being taken as 100. Heads with a breadth of 80 per cent. and over are classed as broad or brachy-cephalic; those with an index under 80, but not under 75, are called medium heads (meso- or mesati-cephalic); long or dolicho-cephalic heads are those in which the ratio of breadth to length is below 75 per cent.

Its value as a test of race.

It is not contended that these groupings correspond to the primary divisions of mankind. Long, broad and medium heads are met with in varying degrees of preponderance among the white, black, and yellow races. But within these primary divisions the proportions of the head serve to mark off important groups. Topinard shows how the form expressed by the index separates the long-headed Scandinavian people from the broad-headed Celts and Slavs; while the Esquimaux are distinguished on similar grounds from the Asiatic Mongols, and the Australians from the Negritos. All authorities agree in regarding the form of the head as an extremely constant and persistent character, which resists the influence of climate and physical surroundings, and (having nothing to do with the personal appearance of the individual) is not liable to be modified by the action of artificial selection. Men choose their wives mainly for their faces and figures, and a long-headed woman offers no greater attractions of external form and colouring than her short-headed sister. The intermixture of races with different head-forms will, of course, affect the index, but even here there is a tendency to revert to the original type when the influence of crossing is withdrawn. On the whole, therefore, the form of the head, especially when combined with other characters, is a good test of racial affinity. It may be added that neither the shape nor the size of the head seems to bear any direct relation to intellectual capacity. People with long heads cannot be said to be cleverer or more advanced in culture than people with short heads.

Shape of the head in India.

In relation to the rest of Asia, India may be described as an area of mainly long-headed people separated by the Himalaya and its offshoots from the Mongolian country, where the broad-headed types are more numerous and more pronounced than anywhere else in the world. At either end of the mountain barrier, broad heads are strongly represented in Assam and Burma on the east, and in Baluchistan on the west, and the same character occurs in varying degrees in the Lower Himalayas and in a belt of country on the west of India extending from Gujarāt through the Deccan to Coorg, the limits of which cannot at present be defined precisely. In the Punjab, Rajputana, and the United Provinces, long heads predominate, but the type gradually changes as we travel eastwards. In Bihar medium heads prevail on the whole, while in certain of the Bengal groups a distinct tendency towards brachy-cephaly may be observed, which shows itself in the Muhammadans and Chandāls of Eastern Bengal, is more distinctly marked in the Kāyasths, and reaches its maximum development among the Bengal Brāhmans. In Peninsular India south of the Vindhya ranges, the prevalent type seems to be mainly long-headed or medium-headed, short heads appearing only in the western zone of country referred to above. But the population of the coast has been much affected by foreign influence, Malayan or Indo-Chinese on the east, Arab, Persian, African, European and Jewish on the west, and the mixed types thus produced cannot be brought under any general formula.

Shape of the nose and nasal index.

The proportions of the nose are determined on the same principle as those of the skull. The length and breadth are measured from certain specified points, and the latter dimension is expressed as a percentage of the former. The nasal index, therefore, is simply the relation of the breadth of the nose to its length. If a man’s nose is as broad as it is long—no infrequent case among the Dravidians—his index is 100. The results thus obtained are grouped in three classes—narrow or fine noses (leptorrhine) in which the width is less than 70 per cent. of the length; broad noses (platyrrhine) in which the proportion rises to 85 per cent. and over, and medium noses (mesorrhine) with an index of from 70 to 85. The index, as Topinard points out, expresses with great accuracy the extent to which the nostrils have been expanded and flattened out or contracted and refined, the height in the two cases varying inversely. It thus represents very distinctly the personal impressions which a particular type conveys to the observer. The broad nose of the Negro or of the typical Dravidian is his most striking feature, and the index records its proportions with unimpeachable accuracy. Where races with different nasal proportions have intermixed, the index marks the degree of crossing that has taken place; it records a large range of variations; and it enables us to group types in a serial order corresponding to that suggested by other characters. For these reasons the nasal index is accepted by all anthropologists as one of the best tests of racial affinity.

Correspondence with social groupings.

Speaking generally, it may be said that the broad type of nose is most common in Madras, the Central Provinces and Chutia Nāgpur; that fine noses in the strict sense of the term are confined to the Punjab and Baluchistan, and that the population of the rest of India tends to fall within the medium class. But the range of the index is very great. It varies in individual cases from 122 to 53, and the mean indices of different groups differ considerably in the same part of the country. The average nasal proportions of the Māl Pahāria tribe of Bengal are expressed by the figure 94·5, while the pastoral Gūjars of the Punjab have an index of 66·9, the Sikhs of 68·8 and the Bengal Brāhmans and Kāyasths of 70·4. In other words, the typical Dravidian, as represented by the Māl Pahāria, has a nose as broad in proportion to its length as the Negro, while this feature in the Indo-Aryan group can fairly bear comparison with the noses of sixty-eight Parisians, measured by Topinard, which gave an average of 69·4. Even more striking is the curiously close correspondence between the gradations of racial type indicated by the nasal index and certain of the social data ascertained by independent enquiry. If we take a series of castes in Bengal, Bihar, the United Provinces of Agra and Oudh, or Madras, and arrange them in the order of the average nasal index, so that the caste with the finest nose shall be at the top, and that with the coarsest at the bottom of the list, it will be found that this order substantially corresponds with the accepted order of social precedence. Thus in Bihar or the United Provinces the casteless tribes, Kols, Korwas, Mundās and the like, who have not yet entered the Brahmanical system, occupy the lowest place in both series. Then come the vermin-eating Musahars and the leather-dressing Chamārs. The fisher castes, Bauri, Bind, and Kewat, are a trifle higher in the scale; the pastoral Goālā, the cultivating Kurmi, and a group of cognate castes from whose hands a Brāhman may take water, follow in due order, and from them we pass to the trading Khatris, the landholding Bābhans and the upper crust of Hindu society. Thus, for those parts of India where there is an appreciable strain of Dravidian blood it is scarcely a paradox to lay down, as a law of the caste organization, that the social status of the members of a particular group varies in inverse ratio to the mean relative width of their noses. Nor is this the only point in which the two sets of observations—the social and the physical—bear out and illustrate each other. The character of the curious matrimonial groupings for which the late Mr. J. F. McLennan devised the useful term exogamous, also varies in a definite relation to the gradations of physical type. Within a certain range of nasal proportions, these sub-divisions are based almost exclusively on the totem. Along with a somewhat finer form of nose, groups called after villages and larger territorial areas, or bearing the name of certain tribal or communal officials, begin to appear, and above these again we reach the eponymous saints and heroes who in India, as in Greece and Rome, are associated with a certain stage of Aryan progress.

Shape of face orbitonasal index.

The comparative flatness of the Mongolian face is a peculiarity which cannot fail to strike the most casual observer. On closer examination this characteristic will be seen to be intimately connected with the formation of the cheek-bones, the margins of the bony sockets of the eyes, and the root of the nose. No precise measurements can be made of the cheek-bones on the living subject, for it is impossible to fix any definite points from which the dimensions can be taken. Some years ago, however, Mr. Oldfield Thomas devised a method of measuring the relative projection of the root of the nose above the level of the eye-sockets, which expresses very accurately the degree of flatness of face met with in different types. It was used by him for skulls, but it has the great advantage of being equally applicable to living persons, and at Sir William Flower’s suggestion it has been extensively used in India, especially among hill tribes and wherever there was reason to suspect an intermixture of Mongolian blood. The principle on which it proceeds can be described without resorting to technical language. Any one who looks at a Gurkha in profile will readily observe that the root of the nose rises much less above the level of the eye-sockets than is the case with Europeans or natives of Upper India. The object is to determine the comparative elevation of the lowest point on the root of the nose above the plane of the eye-sockets. This is done by marking a point on the front surface of the outer edge of each orbit and a third point on the centre of the root of the nose where it is lowest. The distance between the two orbital dots is then measured in a direct line and also the distance from each of these to the dot on the bridge of the nose. The former dimension represents the base of a triangle, and the latter its two sides. The index is formed by calculating the percentage of the latter dimension on the former. If, as is sometimes the case, the bridge of the nose is let down so low that it does not project at all beyond the level of the orbits, the two dimensions will obviously be of equal length and the index will be 100. If, on the other hand, the elevation of the bridge of the nose is marked, the index may be as high as 127 or 130. In the paper already referred to, which dealt only with skulls, Mr. Thomas proposed the division of the index into three classes:—

The experience gained in India, which extends to a large number of castes and tribes in all parts of the country, has led me to adopt the following grouping for the living subject:—

This brings the Mongoloid people of Assam and the Eastern Himalayas within the platyopic group, and effectually differentiates them from the broad-headed races of Baluchistan, Bombay and Coorg. It also separates the Indo-Aryans from the Aryo-Dravidians.

Stature in Europe and India.

Topinard’s classification of stature, which is generally accepted, comprises four groups:—

Much has been written on the subject of the causes which affect the stature. The general conclusion seems to be that in Europe the question is a very complicated one, and that the influence of race is to a great extent obscured by other factors, such as climate, soil, elevation, food supply, habits of life, occupation, and natural or artificial selection. Most of these causes also come into play in India, but not necessarily to the same extent as in Europe. The influence of city life, which in civilized countries as a rule tends to reduce the stature and to produce physical degeneracy, is comparatively small in India, where from fifty to eighty-four per cent. of the population are engaged in agriculture and live an outdoor life. Nor are the conditions of factory industries in India so trying or so likely to affect growth as in Europe. The operatives do not attend so regularly nor do they work so hard, and many of them live in the country for a great part of the year, coming into the mills only when there is nothing to be done in the fields. Some of the indigenous hand-loom weavers, however, show the lowest mean stature yet recorded—a fact which is probably due to the unwholesome surroundings in which they live. In India, as in Europe, the dwellers in the hills are generally shorter than the people of the plains, and within the hill region it may in either case be observed that the stature is often greater at high than at moderate altitudes—a fact which has been ascribed to the influence of a rigorous climate in killing off all but vigorousindividuals. In India the prevalence of malaria in the lowerlevels and the less healthy conditions of life would probablytend to bring about the same result. On the whole, however,the distribution of stature in India seems to suggest that racedifferences play a larger part here than they do in Europe.The tallest statures are massed in Baluchistan, the Punjab, andRajputana; and a progressive decline may be traced down thevalley of the Ganges until the lowest limits are reached amongthe Mongoloid people of the hills bordering on Assam. In thesouth of India the stature is generally lower than in the plainsof the north. The minimum is found among the Negritos ofthe Andaman Islands, whose mean stature is given by Denikeras 1485 mm. or 4 feet 10½ inches.

The seven physical types.

These physical data enable us to divide the people of theIndian Empire into seven main physicaltypes, the distribution of which is shownin the coloured map at the end of thisvolume.* If we include the Andamanese, the number of typesis eight, but for our present purpose this tiny group of Negritosmay be disregarded. Curious and interesting as they are fromthe point of view of general anthropology, the Andamanesehave had no share in the making of the Indian people. Theysurvive—a primitive outlier—on the extreme confines of theEmpire to which they belong merely by virtue of the accidentthat their habitat has been selected as a convenient location fora penal settlement. I have, however, thought it worth whileto take this opportunity of publishing the measurements of200 Andamanese, 100 males and 100 females, which were takensome years ago by Major Molesworth, I.M.S., then Surgeon atPort Blair, in the hope that they may be of service to anyonewho has the leisure to undertake a monograph on the subject.The conclusions suggested by Major Molesworth’s measure-ments of living subjects seem to coincide with those arrivedat by the late Sir William Flower from an examination of aseries of forty-eight skulls, and confirmed by Sir WilliamTurner in the monograph referred to above. These observersagree in describing the Andamanese as short-headed, andbroad-nosed, with a low cranial capacity. Their heads differin essential particulars from those of the Dravidians, and SirWilliam Turner considers that no direct evidence of either apast or a present Negrito population in India has yet been obtained.

Counting from the western frontier of India, we may determine the following distinctive types :—

I. The Turko-Iranian type, represented by the Baloch, Brāhūī, and Afghāns of the Baluchistan Agency and the North-West Frontier Province; probably formed by a fusion of Turki and Persian elements in which the former predominate. Stature above mean; complexion fair; eyes mostly dark, but occasionally grey; hair on face plentiful; head broad; nose moderately narrow, prominent, and very long.

II. The Indo-Aryan type, occupying the Punjab, Rajputana, and Kashmir, and having as its characteristic members the Rajputs, Khatris, and Jāts. This type approaches most closely to that ascribed to the traditional Aryan colonists of India. The stature is mostly tall; complexion fair; eyes dark; hair on face plentiful; head long; nose narrow and prominent, but not specially long.

III. The Scytho-Dravidian type of Western India, comprising the Marātha Brāhmans, the Kunbis, and the Coorgs; probably formed by a mixture of Scythian and Dravidian elements, the former predominating in the higher groups, the latter in the lower. The head is broad; complexion fair; hair on face rather scanty; stature medium; nose moderately fine and not conspicuously long.

IV. The Aryo-Dravidian type found in the United Provinces of Agra and Oudh, in parts of Rajputana, in Bihar and Ceylon, and represented in its upper strata by the Hindustāni Brāhman and in its lower by the Chamār. Probably the result of the intermixture, in varying proportions, of the Indo-Aryan and Dravidian types, the former element predominating in the lower groups and the latter in the higher. The head-form is long with a tendency to medium; the complexion varies from lightish brown to black; the nose ranges from medium to broad, being always broader than among the Indo-Aryans; the stature is lower than in the latter group, and is usually below the average by the scale given above.

V. The Mongolo-Dravidian type of Lower Bengal and Orissa, comprising the Bengal Brāhmans and Kāyasths the Muhammadans of Eastern Bengal, and other groups peculiar to this part of India. Probably a blend of Dravidian and Mongoloid elements with a strain of Indo-Aryan blood in the higher groups. The head is broad; complexion dark; hair on face usually plentiful; stature medium; nose medium with a tendency to broad.

VI. The Mongoloid type of the Himalayas, Nepal, Assam, and Burma, represented by the Kanets of Lahoul and Kulu, the Lepchās of Darjeeling, the Limbus, Murmis and Gurungs of Nepal, the Bodo of Assam, and the Burmese. The head is broad; complexion dark with a yellowish tinge; hair on face scanty; stature small or below average; nose fine to broad; face characteristically flat; eyelids often oblique.

VII. The Dravidian type extending from Ceylon to the valley of the Ganges and pervading the whole of Madras, Hyderabad, the Central Provinces, most of Central India, and Chutia Nāgpur. Its most characteristic representatives are the Paniyans of the South Indian hills and the Santāls of Chutia Nāgpur. Probably the original type of the population of India, now modified to a varying extent by the admixture of Aryan, Scythian, and Mongoloid elements. In typical specimens the stature is short or below mean; the complexion very dark, approaching black; hair plentiful with an occasional tendency to curl; eyes dark; head long; nose very broad, sometimes depressed at the root, but not so as to make the face appear flat.

Limitations of the scheme.

Before proceeding to describe the types in further detail, a few words of preliminary explanation are essential. In the first place, it must be clearly understood that the areas occupied by the various types do not admit of being defined as sharply as they are shown on the map. They melt into each other insensibly, and although at the close of a day’s journey from one ethnic tract to another, an observer whose attention had been directed to the subject would realise clearly enough that the physical characteristics of the people had undergone an appreciable change, he would certainly be unable to say at what particular stage in his progress the transformation had taken place. Allowance, therefore, must be made for the necessary conditions of map-making, and it must not be supposed that a given type comes to an end as abruptly as the patch of colour which indicates the area of its maximum prevalence. Secondly, let no one imagine that any type is alleged to be in exclusive possession of the locality to which it is assigned. When, for example, Madras is described as a Dravidian and Bengal as a Mongolo-Dravidian tract, that does not mean that all the people of Madras and Bengal must of necessity belong to the predominant type. From time immemorial in India a stream of movement has been setting from west to east and from north to south—a tendency impelling the higher types towards the territories occupied by the lower. In the course of this movement representatives of the Indo-Aryan type have spread themselves all over India as conquerors, traders, landowners, or priests, preserving the original characteristics in varying degrees, and receiving a measure of social recognition dependent in the main on the supposed purity of their descent from the original immigrants.* Family and caste traditions record countless instances of such incursions, and in many cases the tradition is confirmed by the concurrent testimony of historical documents and physical characteristics. Even in the provinces farthest removed from the Indo-Aryan settlements in North-Western India, members of the upper castes are still readily distinguishable by their features and complexion from the mass of the population, and their claims to represent a different race are thrown into relief by the definition now for the first time attempted of the types which predominate in different parts of India. Until the existence of a lower type has been established, no special distinction is involved in belonging to a higher one. Thirdly, it may be said that the names assigned to the types beg the highly speculative question of the elements which have contributed to their formation. The criticism is unanswerable. One can but admit its truth, and plead by way of justification that we must have some distinctive names for our types, that names based solely on physical characters are no better than bundles of formulæ, and that if hypotheses of origin are worth constructing at all, one should not shrink from expressing them in their most telling form.

The Turko-Iranian type is in practically exclusive possession of Baluchistān and the North-West Frontier Province. Its leading characteristics are the following :—

(i) The head is broad, the mean indices ranging from 80 in the Baloch of the Western Punjab to 85 in the Hazāra of Afghanistan. I put aside as doubtful cases the Hunzas, Nagars, and Kāfirs and the Pathāns of the North-Western Punjab. For the first three the data are scanty, and it is possible that further enquiry might lead to their inclusion in the Indo-Aryan type. In the case of the last the individual indices vary from 69 to 87, and although broad heads preponderate on the whole, there is a sufficient proportion of long heads to warrant the suspicion of some mixture of blood.

(ii) The proportions of the nose (nasal index) are fine or medium, the average indices running from 67.8 in the Tarin to 80.5 in the Hazāra. Some of the individual indices are high and one Hazāra attains the remarkable figure of 111. These abnormalities may probably be accounted for by the importation of Abyssinian slaves. The proportions of the nose, however, are less distinctive of the type than its great absolute length, which varies in individual cases from 56 mm. among the Hazāras to 65 among the Brāhūi. The one feature indeed that strikes one in these people is the portentous length of their noses, and it is probably this peculiarity that has given rise to the tradition of the Jewish origin of the Afghāns. Some of the Scythian coins exhibit it in a marked degree. As M. Ujfalvy* has pointed out, the lineaments of Kadphises II survive in the Dards of to-day, and the remark holds good of most of the people whom I have ventured to include in the Turko-Iranian type.

(iii) The mean orbito-nasal index, which measures the relative flatness of the face, ranges with the Turko-Iranians from 111 in the Hazāra to 118 in the Baloch, Brāhūi, and Dehwār. The highest individual index (131) occurs among the Pathāns of the North-Western Punjab and the lowest (118) among the Kāfirs. The type as a whole is conspicuously proopic, and there are no signs of that depression of the root of the nose and corresponding flatness of the cheek bones to which the appearance popularly described as Chinese or Mongolian is due. In respect of this character the Hazāras seem to be an exception. In them the individual indices form a continuous curve of striking regularity from 103 to 120, and it is a question whether the tribe ought not to be included in the Mongoloid type. I prefer, however, to show them as Turko-Iranian, for it seems possible that they partake of the elements of both types and represent the points of contact between the two.

(iv) The average stature varies from 162 in the Baloch of Makrān to 172 to the Achakzai Pathān of Northern Baluchistān. The figure for the Hazāra is 168, which makes for their inclusion in the Turko-Iranian rather than in the Mongoloid group; but the subjects measured belonged to one of the regiments at Quetta and were probably rather above the average stature of the tribe.

The Indo-Aryan type predominates in Rajputana, the Punjab, and the Kashmir valley, though in parts of these areas it is associated to a varying extent with other elements. It is readily distinguishable from the Turko-Iranian. Its most marked characteristics may be summarised as follows :—

(i) The head-form is invariably long, the average index ranging from 72·4 in the Rajput to 74·4 in the Awān. The highest individual index (86) is found among the Khatris and the lowest (64) among the Rajputs. The seriations bring out very clearly the enormous preponderance of the long-headed type and present the sharpest contrast with those given for the Turko-Iranians.

(ii) In respect of the proportions of the nose there is very little difference between the two types. The Indo-Aryan index ranges from 66·9 in the Gūjar to 75·2 in the Chuhrā, and there are fewer high individual indices ; but between the seriations there is not much to choose. On the other hand the Indo-Aryans, notwithstanding their greater stature, have noticeably shorter noses than the Turko-Iranians.

(iii) Concerning the orbito-nasal index there is little to be said. All the members of the Indo-Aryan type are placed by their average indices within the pro-opic group ; their faces are free from any suggestion of flatness, and the figures expressing this character run in a very regular series. The highest index (117·9) occurs among the Rajputs and the lowest (113·1) amongst the Khatris.

(iv) The Indo-Aryans have the highest stature recorded in India, ranging from 174·8 in the Rajput to 165·8 in the Arora. Individual measurements of Rajputs rise to 192·4 and of Jāts (Sikhs) to 190·5. Stature alone, therefore, were other indications wanting, would serve to differentiate the Indo-Aryan from the Aryo-Dravidian type of the United Provinces and Bihar.

The most important points to observe in the Indo-Aryan series of measurements are the great uniformity of type and the very slight differences between the higher and the lower groups. Socially, no gulf can be wider than that which divides the Rajputs of Udaipur and Mārwār from the scavenging Chuhrā of the Punjab. Physically, the one is cast in much the same mould as the other ; and the difference in mean height which the seriations disclose is no greater than might easily be accounted for by the fact that in respect of food, occupation, and habits of life, the Rajput has for many generations enjoyed advantages, telling directly on the development of stature, which circumstances have denied to the Chuhrā. Stature we know to be peculiarly sensitive to external influences of this kind. Other and more subtle influences re-act upon environment and tend to modify the type. Sikhism has transformed the despised Chuhrā into the soldierly Mazhabi. Who shall say that military service might not have the same effect on groups belonging to the lower social strata of the Punjab, whose physical endowment is hardly inferior to that observed at the top of the scale?

Scytho-Dravidian type.

The Scytho-Dravidian type occurs in a belt of country on the west of India extending from Gujarat to Coorg. It is represented at one extreme of this belt by the Nāgar Brāhmans of Gujarat and at the other by the remarkable people who have given their name to the little province of Coorg. Excluding the Kātkaris, who really belong to the Dravidian type, the leading characteristics of the Scytho-Dravidians are the following :—

(i) The head-form ranges from 76·9 in the Deshasth Brāhmans to 79·7 in the Nāgar Brāhmans and 79·9 in the Prabhus and the Coorgs, while the maximum individual indices rise as high as 92 with the Marātha Kunbis and the Shenvi Brāhmans. In the case of the three type specimens—the Nāgar Brāhmans, the Prabhus, and the Coorgs—the mean index is virtually 80, and the predominance of the broad-headed type is unmistakable. The seriations show that the gradation of the type is fairly regular, and a comparison with the diagrams of the Indo-Aryans brings out marked differences of head-form, where the features and complexion taken by themselves would appear to point to an identical origin. Both indices and maxima are noticeably lower than among the Turko-Iranians.

(ii) In the proportions of the nose there is nothing much to remark. The mean indices vary from 72·0 in the Coorg to 81·9 in the Mahār, the Nāgar Brāhman giving 73·1 and the Prabhu 75·8. The length of the nose, whether we look to the averages or the maxima, is distinctly less than among the Turko-Iranians, the type most closely allied to the Scytho-Dravidian.

(iii) The mean orbito-nasal index varies from 113·1 in the Son-Koli to the very high figure of 120 in the Coorg. It deserves notice, however, that the minimum indices run very low, and that the range between the highest maximum (132) and the lowest minimum (103) is considerable and points to some mixture of blood.

(iv) The mean stature varies from 160 in the case of the Kunbis in 168.7 in the Coorgs, and an examination of the figures will show that it is, on the whole, lower than among the Turko-Iranians.

The type is clearly distinguished from the Turko-Iranian by a lower stature, a greater length of head, a higher nasal index, a shorter nose and a lower orbito-nasal index. All of these characters, except perhaps the last, may be due to a varying degree of intermixture with the Dravidians. In the higher types the amount of crossing seems to have been slight; in the lower the Dravidian elements are more pronounced, while in the Kātkari the long head and wide nose are conspicuous.

Aryo-Dravidian-type.

The Aryo-Dravidian or Hindustāni type extends from the eastern frontier of the Punjab to the southern extremity of Bihar, from which point onwards it melts into the Mongolo-Dravidian type of Bengal Proper. It occupies the valleys of the Ganges and Jumna, and runs up into the lower levels of the Himalayas on the north and the slopes of the Central Indian plateau on the south. Its higher representatives approach the Indo-Aryan type, while the lower members of the group are in many respects not very far removed from the Dravidians. The type is essentially a mixed one, yet its characteristics are readily definable, and no one would take even an upper class Hindustāni for a pure Indo-Aryan, or a Chamār for a genuine Dravidian. Turning now to details, we find the following results :—

(i) The head-form is long, with a tendency towards medium. The average index varies from 72.1 in the Kāchhi and Koiri of Hindustan to 76.8 in the Dosādh of Bihar and 76.7 in the Bābhan. The highest individual index (90) occurs among the Bābhans of Bihar, and the lowest (62) among the Bhars of Hindustan. But the head-form throws little light upon the origin and affinities of the type, and would of itself barely serve to distinguish the Aryo-Dravidian from the Indo-Aryan. Nor, indeed, would one expect it to do so, for the pure Dravidians are themselves a long-headed race, and the Hindustāni people might well have derived this character from the Dravidian element in their parentage.

(ii) The distinctive feature of the type, the character which gives the real clue to its origin, and stamps the Aryo-Dravidian as racially different from the Indo-Aryan, is to be found in the proportions of the nose. The average index runs in an unbroken series from 73.0 in the Bhuinhār or Bābhan of Hindustan and 73.2 in the Brāhman of Bihar to 86 in the Hindustāni Chamār and 88.7 in the Musahar of Bihar. The order thus established corresponds substantially with the scale of social precedence independently ascertained. At the top of the list are the Bhuinhārs, who rank high among the territorial aristocracy of Hindustan and Bihar; then come the Brāhmans, followed at a slight but yet appreciable interval by the clerkly Kāyasths with an index of 74.8; while down at the bottom the lower strata of Hindu society are represented by the Chamār, who tans hides and is credibly charged with poisoning cattle, and the foul-feeding Musahar who eats pigs, snakes, and jackals, and whose name is popularly derived from his penchant for field-rats. The seriations tell the same tale as the averages, and mark the essential distinction between the Aryo-Dravidian and Indo-Aryan types. The Hindustāni Brāhmans, with a slightly lower mean index than the Chuhṛās of the Punjab, have a far larger proportion of the broad noses, which point to an admixture of Dravidian blood.

(iii) The statistics of height lead to a similar conclusion. The mean stature of the Aryo-Dravidians ranges from 166 centimetres in the Brāhmans and Bhuinhārs to 159 in the Musahar, the corresponding figures in the Indo-Aryan being 174.8 and 165.8. The one begins where the other leaves off.

Mongolo-Dravidian-type.

The Mongolo-Dravidian or Bengali type occupies the delta of the Ganges and its tributaries from the confines of Bihar to the Bay of Bengal. It is one of the most distinctive types in India, and its members may be recognized at a glance throughout the wide area where their remarkable aptitude for clerical pursuits and their keen sense of family obligations have procured them employment. Within its own habitat the type extends to the Himalayas on the north and Assam on the east, and probably includes the bulk of the population of Orissa. The western limit coincides approximately with the hilly country of Chutia Nāgpur and Western Bengal.

(i) The broad head of the Bengali, of which the mean index varies from 79.0 in the Brāhman to 83.0 in the Rājbaṅsi Magh, effectually differentiates the type from the Indo-Aryans orAryo-Dravidians. The seriation of the cephalic index for the Brāhmans of East Bengal is very regular in its gradations, and presents a striking contrast with the corresponding diagrams for the Hindustāni Brāhmans and the Rajput. Here, as elsewhere, the inferences as to racial affinity suggested by the measurements are in entire accord with the evidence afforded by features and general appearance. For example, it is a matter of common knowledge that the Rājbaṅsi Magh of Chittagong, who is in great demand as a cook in European households in India and usually prospers exceedingly, resembles the upper class Bengali of Eastern Bengal so closely that it takes an acute observer to tell the difference between the two.

(ii) The mean proportions of the nose range from 70.3 in the Brāhmans and Kāyasths to 84.7 in the Mals of Western Bengal and 80 in the Kochh. The number of high individual indices brings out the contrast with the Indo-Aryans, and points to the infusion of Dravidian blood. In the Brāhman seriation the finer forms predominate, and it is open to any one to argue that, notwithstanding the uncompromising breadth of the head, the nose-form may, in their case, be due to the remote strain of Indo-Aryan ancestry to which their traditions bear witness.

(iii) The stature varies from 167 in the Brāhmans of Western Bengal to 159 in the Kochh of the Sub-Himalayan region. The seriations of the Kochh deserve special notice for the indications which they give of the two elements that have combined to form the Mongolo-Dravidian type. In writing about them fifteen years ago I ventured, on the evidence then available, to describe them as a people of Dravidian stock who, being driven by pressure from the west into the swamps and forests of Northern and North-Eastern Bengal, were there brought into contact with the Mongoloid races of the Lower Himalayas and the Assam border, with the result that their type was affected in a varying degree by intermixture with these races. On the whole, however, I thought that Dravidian characteristics predominated among them over Mongolian. My conclusions, which coincided in the main with those of Colonel Dalton and other observers, have been questioned by Lieutenant-Colonel Waddell, C.B., C.I.E., in a paper on the Tribes of the Brahmaputra Valley.* Colonel Waddell, who has observed and measured the Kochh both in North-Eastern Bengal and in Assam, denies their Dravidian origin, and describes them as “distinctly Mongoloid though somewhat heterogeneous.” For purposes of comparison I have included both his measurements and my own in the same diagram. As regards the head-form and the stature, the two sets of observations are practically identical. In the case of the nose, Colonel Waddell’s data show a far higher proportion of broad noses than mine, and clearly point to a strong Dravidian element. On the other hand, the orbito-nasal index exhibits, though in a less degree, some distinctive Mongoloid characteristics. One can ask for no better illustration of the efficacy of the method of anthropometry in its application to a mixed or transitional type than the fact that, while two independent observers have formed different opinions as to the relative preponderance of its component elements, the data obtained by them from two distinct series of individuals correspond to the remarkable extent indicated by the Kochh diagram. There is, of course, no real conflict of opinion between Colonel Waddell and myself. The whole question turns upon the point of view of the observer. Take the Kochh in Dinajpur and Rangpur, and they strike you as in the main Dravidian; travel further east, and include in your survey the cognate Kachāri of Assam, and there is no mistaking the fact that Mongoloid characteristics predominate. The same may be said of the Bengali type as a whole. In Western Bengal the Dravidian element is prominent; in Dacca and Mymensingh the type has undergone a change, which scientific methods enable us to assign to the effect of intercourse with a Mongolian race.

Mongoloid type.

On its northern and eastern frontier India marches with the great Mongolian region of the earth. The effect of this contact with an almost exclusively broad-headed population is indicated in yellow on the map, and a glance will show how the area within which this particular foreign influence has impressed itself upon India widens gradually from west to east. The Punjab and Hindustan are left virtually untouched; the Bengalis exhibit a type sensibly modified in the direction of Mongolian characters; the Assamese are unmistakably Mongoloid, and in Burma the only non-Mongolian elements are the result of recent immigration from India. This condition of things is of course mainly due to the intervention of the great physical barrier of the Himalayas, “the human equator of the earth,” as an American anthropologist* has called it, which throughout its length offers an impassable obstacle to the southward extension of the Mongolian races. But other causes also enter in. No one who is acquainted with the population of the Lower Himalayas can have failed to observe that in the west there has been a substantial intermixture of Indo-Aryan elements, while in the east the prevailing type down to the verge of the plains is exclusively Mongoloid. The reason seems to be that the warlike races of the Punjab and Hindustan invaded the pleasant places of the hills and conquered for themselves the little kingdoms which once extended from the Kashmir valley to the eastern border of Nepal. The Dogrās or Hill Rajputs of Kāngra, and the Khas of Nepal form the living record of these forgotten enterprises. Further east the conditions were reversed, neither Bengalis nor Assamese have any stomach for fighting; they submitted tamely to the periodical raids of the hill people, and the only check upon the incursions of the latter was their inability to stand the heat of the plains. They occupied, however, the whole of the lower ranges and held the Duārs or “gates” of Bhutan until dispossessed by us. Thus in the Eastern Himalayas none of the plains people made good a footing within the hills, which remain to this day in the exclusive possession of races of the Mongoloid type.

The summaries of measurements given in the appendix relate to a fairly large number of subjects and the type is distinct.

(i) The prevalent head-form is broad, but the mean indices show some remarkable departures from this type. The Jaintia index is 72.9, thus falling within the long-headed category, and several tribes have indices between 75 and 80. These low indices are, however, based upon a comparatively small number of subjects, and it seems not unlikely that a larger series of measurements may sensibly modify the average. In any case a great deal of work will have to be done before we are in a position to determine the probable affinities of the numerous Mongoloid tribes who inhabit the hilly region between India and China.

(ii) The nose-form appears at first sight to show a great range of variations, but on closer examination it will be seen that the higher indices are for the most part confined to tribes for which the data are scanty. In the larger groups the mean index ranges from 67.2 for the Lepchās to 84.5 for the Chakmās and 86·3 for the Khasiās; the Tibetans (73·9) and the Murmis (75·2) falling between these extremes. The highest mean index (95·1) occurs among the Mānde or Gāro, in one of whom, according to Lieutenant-Colonel Waddell, the width of the nose exceeds its height to an extent indicated by the surprising ratio of 117. But only 34 Gāros have been measured, and looking to the possibilities of crossing one can scarcely regard the figures as conclusive. On the measurements given in the table there may be some question whether the Mānde should not be classed as Mongolo-Dravidian, and this view may be thought to derive some support from Buchanan’s description of them as a wild section of the Kochh.*

(iii) Under the head of stature there is nothing much to remark. The Gurungs (169·8) are the tallest and the Miris (156·4) the shortest of the tribes included in the table. The 106 Tibetans show an average of 163·3, which may be regarded as fairly typical. The tallest individuals (176) are found among the Tibetans and Murmis; the shortest (141) are the Khambus and the Khasiās.

(iv) The characteristic orbito-nasal index, which measures the relative flatness or prominence of the root of the nose and the adjacent features, yields singularly uniform results. The average varies in the large groups, which alone are worth considering, from 106·4 in the Chakmā to 109·1 in the Tibetan. For the Lepchās Lieutenant-Colonel Waddell’s observations yield a mean index of 105·8, with a maximum of 119 and a minimum of 92, against my average of 101·8 ranging from 133 to 103. As my figures relate to a larger number of subjects (57 against 36), I have selected them in preference to his for inclusion in the diagram showing seriation. A glance at the diagrams given for the Lepchās of Darjeeling and the Chakmās of the Hill Tracts of Chittagong will show how regularly the gradations of the indices are distributed, and will bring out better than any description the correspondences and divergences of type.

Dravidian type.

The Dravidian race, the most primitive of the Indian peoples, occupies the oldest geological formation in India, the medley of forest-clad ranges, terraced plateaux, and undulating plains which stretches, roughly speaking, from the Vindhyas to Cape Comorin. On the east and west of the peninsular area the domain of the Dravidian is conterminous with the Ghāts; while farther north it reaches on one side to the Aravallis and on the other to the Rajmahal hills. Where the original characteristics have been unchanged by contact with Indo-Aryan or Mongoloid people the type is remarkably uniform and distinctive. Labour is the birthright of the pure Dravidian, and as a coolie he is in great demand wherever one meets him. Whether hoeing tea in Assam, the Duārs and Ceylon, planting sugar-cane in far Fiji, cutting rice in the swamps of Eastern Bengal, or doing scavenger’s work in the streets of Calcutta, Rangoon, and Singapore, he is recognizable at a glance by his black skin, his squat figure and the negro-like proportions of his nose. In the upper strata of the vast social deposit which is here treated as Dravidian these typical characteristics tend to thin out and disappear, but even among them traces of the original stock survive in varying degrees. We must look to the researches of Mr. Thurston,* who is conducting the ethnographic survey of Southern India, to define and classify the numerous sub-types thus established and to determine the causes which have given rise to them.

Turning now to the actual measurements we find the following specific characters:— (i) The head-form is usually medium with a tendency in the direction of length. The mean indices range in Southern India from 71·7 in the Badaga of the Nilgiris and 72·9 in the Kādir of the Anamalai Hills to 76·6 in the Shānāns of Tinnevelly. The Tiyans (73), Nāyars (73·2), Cheruman (73·4), Palli (73), Parāyan or Pariah (73·6), Irula (73·1) and several others also fall well within the long-headed group. In Chutia Nāgpur, on the other hand, the type is uniformly medium. Among the large groups the Chik (73·8), the Munda (74·5), the Māle (74·8), the Kharia (74·5), and the Korwa (74·4) are just included in the long-headed division; while for all the others the mean index ranges about 75 and 76. In this part of India the physical conformation of the country, the vast stretches of fever-haunted jungle, the absence of roads, and the compact tribal organization and independent spirit of the Dravidian races have tended to preserve them singularly free from the intrusion of foreign influence, and for these reasons I believe that their measurements may be taken as fairly typical. The seriation given for the Santāls shows how regularly the individual indices are graduated.

(ii) In Southern India the mean proportions of the nose vary from 69.1 in the Lambādis of Mysore, and 73.1 in the Vellālas of Madras to 95.1 in the Paniyans of Malabar. In Chutia Nāgpur and Western Bengal the range of variation is less marked, and the mean indices run from 82.6 in the Kurmi of Mānbhum in a gradually ascending series to 94.5 in the Māle of the Santāl Parganas. The Asur figure of 95.9 may be left out of account as it relates only to two subjects. In both regions the mean proportions of the nose correspond in the main to the gradations of social precedence, and such divergences as occur admit of being plausibly accounted for. At the head of the physical series in Southern India stand the Lambādi with a mean index of 69.1. They do not employ the local Brāhmans as priests, and their touch is held to convey ceremonial pollution. But there is reason to believe that they are a nomadic people from Upper India, and that their social rank is low merely because they have not been absorbed in the social system of the South. Next come the Vellālas, the great cultivating caste of the Tamil country, with a mean index of 73.1. They are classed as Sat or pure Sudras; the Brāhmans who serve them as priests will take curds and butter from their hands and will cook in any part of their houses. The Tamil Brāhmans themselves belong, indeed, to a lower physical type; but their mean index of 76.7 has probably been affected by the inclusion in the group of some tribal priests, who obtained recognition as Brāhmans when their votaries insensibly became Hindus. Then follow the Palli (77.9), a large group mainly employed in agriculture, who claim twice-born rank and frequently describe themselves as Agnikula or fire-born Kshatriyas. Low down in the social as in the physical scale are the Parāiyan or Pariah, with an index of 80, whose mere vicinity pollutes, but whose traditions point to the probability that their status was not always so degraded as we find it at the present day. This conjecture derives some support from the fact that the Kādir, Mukkuvan and Paniyan with substantially broader noses yet take higher social rank.

(iii) Among the Dravidians of Southern India the mean stature ranges from 170 in the Shānān of Tinnevelly to 153 in the Pulaiyan of Travancore; and individual measurements vary from 182.8 in the former group to 143.4 in the latter. Mr. Thurston has drawn my attention to the well-marked correlation between stature and the proportions of the nose which is brought out by the following statement:—

In Chutia Nāgpur and Western Bengal the stature is more uniform, varying from 162.7 in the Orāon of Ranchi to 157.7 in the Māl Pahāria and Māle of the Santāl Parganas, and the correlation with the proportions of the nose, though traceable, is less distinct.

Origins of types.

The origins of these types are hidden in the mist which veils the remote era of the Aryan advance into India. Within that dim region evidence is sought for in vain. Our only guides are tradition and conjecture, aided by the assumption, which the history of the East warrants us in making, that in those distant ages types were formed by much the same processes as those that we find in operation to-day. Such are our materials for a study of the evolution of the Indian people. At the best the picture can present but shadowy outlines. All that can be demanded of it is that it should accord in the main with the scanty data furnished by what passes for history in India, and at the same time should offer a consistent and plausible explanation of the ethnic conditions which prevail at the present time.

The oldest of the seven types is probably the Dravidian. Their low stature, black skin, long heads, broad noses, and relatively long fore-arm distinguish them from the rest of the population, and appear at first sight to confirm Huxley’s surmise that they may be related to the aborigines of Australia. Linguistic affinities, especially the resemblance between the numerals in Mundāri and in certain Australian dialects, and the survival of some abortive forms of the boomerang in Southern India, have been cited in support of this view, and an appeal has also been made to Sclater’s hypothesis of a submerged continent of Lemuria, extending from Madagascar to the Malay Archipelago, and linking India with Africa on the one side and Australia on the other. But Sir William Turner’s comparative study of the characters of Australian and Dravidian crania has not led him to the conclusion that these data can be adduced in support of the theory of the unity of the two peoples.

Dravidian.

The facts which cast doubt on the Australian affinities of the Dravidians finally refute the hasty opinion which seeks to associate them with the tiny, broad-headed, and woolly-haired Negritos of the Andamans and the Philippines. This is the last word of scientific authority, and here we might leave the subject, were it not that another theory of the origin of the Dravidians was adopted by Sir William Hunter in the account of the non-Aryan races of India given by him in The Indian Empire. According to this view there are two branches of the Dravidians—the Kolarians speaking dialects allied to Mundāri, and the Dravidians proper whose languages belong to the Tamil family. The former entered India from the north-east and occupied the northern portion of the Vindhya table-land. There they were conquered and split into fragments by the main body of Dravidians, who found their way into the Punjab through the north-western passes and pressed forward towards the south of India. The basis of this theory is obscure. Its account of the Dravidians seems to rest upon a supposed affinity between the Brāhūi dialect of Baluchistan and the languages of Southern India; while the hypothesis of the north-eastern origin of the Kolarians depends on the fancied recognition of Mongolian characteristics among the people of Chutia Nāgpur. But in the first place the distinction between Kolarians and Dravidians is purely linguistic and does not correspond to any differences of physical type. Secondly, it is extremely improbable that a large body of very black and conspicuously long-headed types should have come from the one region of the earth which is peopled exclusively by races with broad heads and yellow complexions. With this we may dismiss the theory which assigns a trans-Himalayan origin to the Dravidians. Taking them as we find them now it may safely be said that their present geographical distribution, the marked uniformity of physical characters among the more primitive members of the group, their animistic religion, their distinctive languages, their stone monuments and their retention of a primitive system of totemism justify us in regarding them as the earliest inhabitants of India of whom we have any knowledge.

The Indo-Aryan. Its non-Indian origin.

Upon the interminable discussions known as the Aryan controversy there is no need to enter here. Whether anything that can properly be described as an Aryan race ever existed; whether the heads of its members were long, according to Penka, or short according to Sergi; whether its original habitat was Scandinavia, the Lithuanian steppe, South-Eastern Russia, Central Asia, or India itself, as various authorities have held; or again whether the term Aryan is anything more than a philological expression denoting the heterogeneous group of peoples whose languages belong to the Aryan family of speech—these are questions which may for our present purpose be left unanswered. We are concerned merely with the fact that there exists in the Punjab and Rajputana at the present day, a definite physical type, represented by the Jāts and Rajputs, which is marked by a relatively long (dolicho-cephalic) head; a straight, finely cut (leptorrhine) nose; a long, symmetrically narrow face; a well-developed forehead, regular features, and a high facial angle. The stature is high and the general build of the figure is well proportioned, being relatively massive in the Jāts and relatively slender in the Rajputs. Throughout the group the predominant colour of the skin is a very light transparent brown, with a tendency towards darker shades in the lower social strata. Except among the Meos and Mīnas of Rajputana, where a strain of Bhīl blood may perhaps be discerned, the type shows no signs of having been modified by contact with the Dravidians; its physical characteristics are remarkably uniform; and the geographical conditions of its habitat tend to exclude the possibility of intermixture with the black races of the south. In respect of their social characters the Indo-Aryans, as I have ventured to call them, are equally distinct from the bulk of the Indian people. They have not wholly escaped the contagion of caste; but its bonds are less rigid among them than with the other Indian races; and the social system retains features which recall the more fluid organization of the tribe. Marriage in particular is not restricted by the hard and fast limits which caste tends to impose, but is regulated within large groups by the principle of hypergamy or ‘marrying up’ which was supposed to govern the connubial relations of the four original classes (varna) in the system described by Manu. Even now Rajputs and Jāts occasionally intermarry, the Rajputs taking wives from the Jāts, but refusing to give their own maidens in return. What is the exception to-day is said to have been the rule in earlier times. In short, both social and physical characters are those of a comparatively homogeneous community which has been but little affected by crossing with alien races.

The mode of its entry into India.

The uniformity of the Indo-Aryan type can be accounted for only by one of two hypotheses, (1) that its members were indigenous to the Punjab, (2) that they entered India in a compact body or in a continuous stream of families from beyond the north-west frontier. It is clear that they could not have come by sea, and equally clear that they could not have found their way into India round the Eastern end of the Himalayas. The theory that the Punjab was the cradle of the Aryan race was propounded by a writer in the Journal of the Royal Asiatic Society * about fifty years ago, on the basis of some rather crude linguistic speculations; but it met with no acceptance, and the opinion of European scholars from Von Schlegel down to the present time is unanimous in favour of the foreign origin of the Indo-Aryans. The arguments appealed to are mainly philological. Vedic literature, indeed, as Zimmer † admits, throws but scanty light upon the subject, for no great weight can be laid upon the identification of the River Rasā with the Araxes, the name by which the Jaxartes was known to Herodotus. Following authority, however, we may assume for our present purpose that the ancestors of the Indo-Aryans came into India from the north-west, and that at the time of their arrival the peninsula, as far as the valley of the Ganges and Jumna, was in the possession of the Dravidians. The only indication of the latter people having extended further to the west, is to be found (as has been mentioned already) in the survival of Brāhūi, an island of supposed Dravidian speech, among the Iranian languages of Baluchistan. But the present speakers of Brāhūi are certainly not Dravidians by race; and we find no traces of Dravidian blood among the Indo-Aryans of to-day. It seems probable, therefore, that when the Indo-Aryans entered the Punjab they brought their own women with them, and were not reduced to the necessity of capturing Dravidian brides. On no other supposition can we explain the comparative purity of their type.

Now, if the physical and social conditions of the Indian Borderland had been the same in those remote ages as we find them at the present day, it is difficult to see how the slow advance of family or tribal migration could have proceeded on a scale large enough to result in an effective occupation of the Punjab. The frontier strip itself, a mere tangle of barren hills and narrow valleys, is ill-adapted to serve as an officina gentium ; while a pastoral people, moving by clans or families from more favoured regions further west, would have found their way barred by obstacles which only the strongest members of the community could have surmounted. The women and children must have been left behind or they would have perished by the way. Again, given the present rainfall and climate of the countries adjacent to India, where should we find to-day, within a measurable distance of the frontier, the favoured region that would give off the swarm of emigrants required to people the Punjab? Surely not in south-eastern Persia, with its inhospitable deserts of shifting sand; nor on the dreary Central Asian steppes where only a scanty nomadic population finds a meagre subsistence. But is it certain that during the three or four thousand years that may have elapsed since the Aryans began to press forward into India the climate of the countries through which they passed may not have undergone a material change? There is an appreciable amount of evidence, the value of which I am anxious not to overrate, in favour of this supposition. The late Mr. W. T. Blanford, writing in 1873,* thought it probable that the rainfall both in Central Asia and Persia had fallen off greatly in modern times, and that owing mainly to this cause, and in a less degree to the destruction of trees and bushes, the climate had become perceptibly drier, cultivation had fallen off and the population had greatly declined in numbers. Nearly thirty years later, we find Mr. Blanford’s views confirmed and developed by Mr. E. Vredenburg in his geological sketch of the Baluchistan Desert and part of Eastern Persia.† Mr. Vredenburg applies to the problem the known principles of physical geography and shows how, given a dwindling rainfall in a tract situated like Eastern Persia and Baluchistan, evaporation is bound to produce the present condition of perennial drought. As the rainfall declines fertile plains relapse into deserts; lakes are transformed into hideous salt marshes; the springs in the hills dry up and an era of desolation sets in. No human agency, however corrupt, no mere misgovernment, however colossal, could bring about such widespread disaster. The village communities, give them but earth and water, would outlast the conqueror and the marauder, as they have done in India. The forces of nature alone could defeat their patient industry. It is the great merit of Mr. Vredenburg’s paper that it indicates the true cause of the facts observed and exposes the fallacy of the belief, countenanced by a long series of travellers, that oriental inertia and corruption are solely or chiefly answerable for the present condition of Baluchistan. In illustration of the state of things which must have existed in some former age, he tells us how in the desolate valleys of the State of Khārān there exist hundreds of stone walls, known locally as gorbands or “dams of the infidels,” which mark the edges of ancient terraced fields, and retain even now remnants of soil which once was cultivated. A legend still survives that the builders of these walls carried the earth in bags on their backs from the alluvial desert on the south, a form of labour which the indolent Baloch would regard as degrading to the dignity of a man. Toil of this sort, whether the soil was transported by beasts of burden or by men, can only have been undertaken in the certain hope of a substantial return. No one would construct fields in a rainless wilderness of ravines, or build walls which have lasted for centuries to retain water where water there was none. Nor is it likely that the cultivation was confined to the hills. Arguing from what one sees in India, it seems far more likely that these terraced fields represent the overflow of a flourishing agricultural community driven up into the hills by the pressure of population in the plains. Gradually as the climate changed, the level alluvial tracts, deprived of rainfall, lapsed into desert; the bulk of the population drifted on into the Punjab, while those who remained behind turned their ploughshares into swords and eked out by pillage the meagre livelihood to be won from patches of soil in the hills. Last of all, the springs on which this scanty cultivation depended shrank and disappeared, till nothing was left but the stone walls to recall the labours of the forgotten people who built them.

The picture which these observations enable us to construct of a country of great lakes and fertile plains extending from the centre of Persia to the western confines of India, or let us say from the Dasht-i-Kavir in western Khorāsān to the deserts of Registān and Khārān, may help to throw light upon the problem of the Indo-Aryan advance into the Punjab. The population of such a tract, as they began to press on their own means of subsistence or were pushed forward by incursions from the west, would naturally have moved on by tribes and families, without any disturbance of their social order, and would have occupied the valley of the Indus. Arriving there as an organized society, like the children of Israel when they entered Palestine, they would have had no need and no temptation to take to themselves any Dravidian daughters of Heth, and they would have preserved their type as distinct as we find it in the Punjab to-day. The movement must, of course, have been gradual and must have extended over many centuries, during which time the climate continued to dry up and the possibilities of agriculture to decline. When the new conditions had become fully established the north-western frontier of India was closed to the slow advance of family or tribal migration and remained open only to bands of fighting men or adventurous nomads, who could force their way through long zones of waterless deserts ending in a maze of robber-haunted hills. Armed invasion took the place of peaceful colonization. But the invaders, however great their strength, could in any case bring relatively few women in their train. This indeed is the determining factor both of the ethnology and of the history of India. As each wave of conquerors, Greek, Scythian, Arab, Moghal, that entered the country by land became more or less absorbed in the indigenous population, their physique degenerated, their individuality vanished, their energy was sapped, and dominion passed from their hands into those of more vigorous successors. Ex Occidente Imperium; the genius of Empire in India has come to her from the West; and can be maintained only by constant infusions of fresh blood from the same source.

The scanty glimpses that are obtained of the history of this region in the distant past bear out the conclusions of the scientific observer. Three hundred years before the Christian era, Alexander’s lieutenant Krateros conducted half of the army which had invaded India, consisting of some fifty thousand men encumbered with elephants, invalids and heavy baggage, from Quetta to Kandahar and thence by the Helmund Valley to Narmashir in Seistān. The route which he followed crossed the southern end of the Dasht-i-Lūt or Desert of Desolation, and traversed nearly two hundred miles of what is now an absolute waste “either waterless or supplied with the most brackish wells.”* Arrian’s account of the march makes no mention of disaster, and Krateros appears to have joined Alexander without any material loss either of elephants or invalids. Strabo again, who described Kirmān about 20 B.C. in a treatise on geography for the use of Roman administrators, speaks of it as a fertile and well-wooded country watered by rivers and producing everything.

Yet when Major Sykes passed through a part of the same tract in 1893–94 he found it covered with ancient ruins and had difficulty in procuring forage for the camels of his small party numbering only about twenty men. Clearly the whole face of the country must have been transformed in the interval. Was this the work of nature or of man? Has the disappearance of the population been brought about by physical causes, such as diminished rainfall, the shifting of river courses, the inroads of wind-driven sand, and the shrinking of the crust of the earth? Or need we look no further than the familiar incidents of Oriental misgovernment—incessant wars, general lawlessness, official corruption and neglect of natural resources? To these questions an answer is supplied by Mr. Ellsworth Huntington’s paper on the Basin of Eastern Seistan and Persia, which forms part of the report of the explorations conducted in Turkestan and Persia in 1903 with the support of the Carnegie Institution of Washington.* Here it is shown that the main cause of the desolation now prevailing is the aridity of the climate due to the high mountains which “on every side shut out the moisture of the sea and shut in the people.” Ever since the end of the Tertiary era the geological history of the country has been marked by a series of epochs of “prolonged rivers and expanded lakes,” alternating with epochs when the rivers were curtailed and the lakes contracted; while throughout the period earth-movements have taken place tending to elevate the barren hills and extend their area and to reduce both the size and the productive capabilities of the habitable basins which they enclose. By the side of these overwhelming physical forces the influence of mere human agencies, such as foreign invasions and native misgovernment, sinks into insignificance. The argument is clinched by the effective comparison which Mr. Huntington draws between the four provinces of Khorāsān, Azerbaijan, Kirmān and Seistān, all of which are equally badly governed. The two former have been devastated by repeated invasions of the most savage character, but they enjoy a relatively abundant rainfall; the two latter have suffered less severely from war, but are afflicted by more or less permanent drought. Khorāsān and Azerbaijan are the most populous and flourishing provinces of Persia; Seistān and Kirmān have been depopulated almost beyond hope of recovery. In Persia, as in India, nature is stronger than man.

The Aryo-Dravidians: Dr. Hoernle’s theory.

For the origin of the Aryo-Dravidian type we need not travel beyond the ingenious hypothesis put forward by Dr. Hoernle twenty-six years ago and confirmed by the recent researches of Dr. Grierson’s Linguistic Survey. This theory supposes that after the first swarm of Indo-Aryans had occupied the Punjab, a second wave of Aryan-speaking people, the remote ancestors of the Aryo-Dravidians of to-day, impelled by some ethnic upheaval, or driven forward by the change of climate in Central Asia to which we have referred above, made their way into India through Gilgit and Chitrāl and established themselves in the plains of the Ganges and Jumna, the sacred Middle-land (Madhyadesa) of Vedic tradition. Here they came in contact with the Dravidians; here by the stress of that contact, caste was evolved; here the Vedas were composed, and the whole fantastic structure of orthodox ritual and usage was built up. For the linguistic evidence in favour of this view I must refer the reader to Dr. Grierson’s chapter on language in the Report on the Census of India, 1901. For my present purpose it is sufficient to note that the record of physical characters bears out the conclusions suggested by philology. The type of the people now dwelling in the Middle-land is precisely what might have been expected to result from the incursion of a fair long-headed race, travelling by a route which prevented women from accompanying them, into a land inhabited by dark-skinned Dravidians. The men of the stronger race took to themselves the women of the weaker, and from these unions was evolved the mixed type which we find in Hindustan and Bihar. The degree of intermixture varied to the extent indicated in the tables of measurements; at one end of the scale the type approaches the Indo-Aryan, at the other it almost merges in the Dravidian.

It may be said that the theory of a second wave of Aryans, resting as it does on the somewhat uncertain data of philology, is not really required for the purpose of explaining the facts. Why should we not content ourselves by assuming that the original Indo-Aryans outgrew their settlements on the Indus and threw off swarms of emigrants who passed down the Ganges valley, modifying their type as they went by alliances with the Dravidian inhabitants? But on this view of the problem it is difficult to account for the marked divergence of type that distinguishes the people of the Eastern Punjab from the people of Western Hindustan. If there had been no second and distinct incursion coming in like a wedge behind the original colonists, no such sharp contrast would now be discernible. One type would melt into the other by imperceptible gradations, and scientific observation and popular impressions would not concur, as they do, in affirming that a marked change takes place somewhere about the longitude of Sirhind—a name which itself preserves the tradition of an ethnic frontier. Nor is this the only point in favour of Dr. Hoernle’s hypothesis. That theory further explains how it is that the Vedic hymns contain no reference to the route by which the Aryans entered India or to their earlier settlements on the Indus; and it accounts for the antagonism between the eastern and western sections and for the fact that the latter were regarded as comparative barbarians by the more cultured inhabitants of the Middle-land.

The Monogolo-Dravidians.

When we leave Bihar and pass on eastward into the steamy-rice-fields of Bengal, the Indo-Aryan element thins out rapidly and appears only in a sporadic form. The bulk of the population is Dravidian, modified by a strain of Mongoloid blood which is relatively strong in the east and appreciably weaker in the west. Even in Bengal, however, where the Indo-Aryan factor is so small as to be hardly traceable, certain exceptions may be noticed. The tradition cherished by the Brāhmans and Kāyasths of Bengal that their ancestors came from Kanauj at the invitation of King Adisur to introduce Vedic ritual into an unhallowed region is borne out to a substantial degree by the measurements of these castes, though even among them indications are not wanting of occasional intermixture with Dravidians.* If, however, the type is regarded as a whole the racial features are seen to be comparatively distinct. The physical degeneration which has taken place may be due to the influence of a relaxing climate and an enfeebling diet, and still more perhaps to the practice of marrying immature children, the great blot on the social system of the upper classes of Bengal.

The Scytho-Dravidian type: its history.

Of the foreign elements that have contributed to the making of the Indian people two have now been passed in review. We have seen the Indo-Aryan type maintaining a high degree of purity in the Punjab and Rajputana, transformed by an increasing admixture of Dravidian blood in Hindustan and Bihar, and vanishing beyond recognition in the swamps of Lower Bengal. We have found the Mongoloid races predominant on the eastern and northern frontiers, confined to the hills where the people of the plains were strong, but further east, where they came in contact with feebler folk, mixing with the Dravidian element to form the type characteristic of the mass of the population of Bengal and Assam. A third foreign element still remains to be accounted for. It has long been known, mainly from Chinese sources supplemented by the evidence of coins and the uncertain testimony of Indian tradition, that long after the settlement of the Indo-Aryans in the Punjab successive swarms of nomadic people, vaguely designated Sakas or Scythians, forced a way into India from the west, and established their dominion over portions of the Punjab, Sind, Gujarat, Rajputana, and Central India. The impulse which started them on their wanderings may be traced in some instances to tribal upheavals in far distant China, while in other cases hordes already on the move were pushed forward from Central Asia. All these people came from regions which, so far as we know, have from time immemorial been occupied by broadheaded races.

In the time of the Achæmenian kings of Persia the Scythians, who were known to the Chinese as Ssē, occupied the regions lying between the lower course of the Sillis or Jaxartes and Lake Balkash. We learn from Herodotus that according to the opinion of classical antiquity these Scythians were riding people who wore breeches and used bows of a fashion of their own. It may be gathered from other sources that their empire extended up to the plains of Eastern Turkestan. In the sixth century B.C. the Scythians, who were then renowned for their valour and their riches, came within the scope of the ambitious policy of Cyrus. Their king Amorges was made prisoner, but Sparethra, his wife, rallied the remains of the army, repulsed the Persians, and compelled them to surrender her husband in exchange for the prisoners she had taken. Notwithstanding this temporary success, the Scythians were nevertheless recognised as tributaries of the Persians, and the portion of Turkestan which they occupied formed the twentieth Satrapy of the Persian Empire. Later on they seem to have regained their independence, for at the battle of Arbela we find them fighting on the Persian side no longer as subjects but allies. The fragments of early Scythian history which may be collected from classical writers are supplemented by the Chinese annals, which tell us how the Ssē, originally located in Southern China, occupied Sogdiana and Transoxiana at the time of the establishment of the Græco-Bactrian monarchy about the year 165 B.C. Dislodged from these regions by the Yuechi, who had themselves been put to flight by the Huns, the Ssē invaded Bactriana, an enterprise in which they were frequently allied with the Parthians. To this circumstance, says Ujfalvy, may be due the resemblance which exists between the Scythian coins of India and those of the Parthian kings.

Its possible origin.

At a later period the Yuechi made a further advance and drove the Scythians or Sakas out of Bactriana, whereupon the latter crossed the Paropanisus and took possession of the country called after them Sakastān, comprising Segistān, Arachosia, and Drangiana. But they were left in possession only for a hundred years, for in the year 25 B.C. the Yuechi disturbed them afresh. A body of Scythians then emigrated eastward and founded a kingdom in the western portion of the Punjab. The route they followed in their advance upon India is uncertain, but to a people of their habits who were already located in Sakastān it would seem that the march through Baluchistan and Kachhi would have presented no serious difficulty. Among the sculptured figures on the rock of Behistun there is one which bears the name of Sakuka, the Scythian. Khanikoff, writing in 1866, professed to recognise in this figure the features of a Kirghiz of the present day. Ujfalvy, however, regards the statement as doubtful. He says that he has never seen a Kirghiz with such a luxuriant beard, and the physiognomy of the figure in question appears to him to be Turko-Tartar presenting a mixture of Mongolian and Aryan lineaments.

The Indo-Scythian Yuechi, afterwards known as the Tokhari, while settled in Eastern Turkestan to the south of the Tian Shan range were defeated by the Hiung-nu or Huns in 201-265 B.C. They fled towards the west, crossed the mountains and took possession of the part of Bactriana inhabited by the Tajiks. A portion of them remained in Eastern Turkestan in the mountainous country to the south-west of Khotan. The Chinese called these people the Siao or Little Yuechi, in order to distinguish them from the others, whom they designated the Ta or Great Yuechi. The Yuechi occupied Central Asia and the north-west of India for more than five centuries from 136 B.C. to 425 A.D. The Hindus called them Sakas and Turushkas, but their kings seem to have known of no other dynastic title than that of Kushan. The Chinese annals tell us how Kitolo, Chief of the Great Kushans, whose name is identified with the Kidara of the coins, giving way before the incursions of the Ephthalites, crossed the Paropanisus and founded in the year 425 of our era the kingdom of Gandhāra, of which in the time of his son Peshawar became the capital. Fifty years later the Ephthalites took possession of Gandhāra and forced the Kushans to retreat into Chitrāl, Gilgit, and Kashmir.

Just at the time when the Kushans were establishing themselves in Gandhāra, the Ephthalites or Hoa of the Chinese annals, who were then settled on the north of the Great Wall of China, being driven out of their territory by the Juan-Juan, started westward and overran in succession Sogdiana, Khwarizm, Bactriana, and finally the north-west portion of India. Their invading movements reached India in the reign of Skanda Gupta, 452—480, and brought about the disruption of the Gupta Empire. The Ephthalites were known in India as Huns. The leader of the invasion of India, who succeeded in snatching Gandhāra from the Kushans, and established his capital at Sakala, is called by the Chinese Laelih, and the inscriptions enable us to identify him with the original Lakhan Udayaditya of the coins. His son Toramana (490—510) took possession of Gujarat, Rajputana and a portion of the Ganges valley, and in this way the Huns came into possession of the ancient Gupta Kingdom. Toramana’s successor Mihirakula (510—540) added at the beginning of his reign Kashmir to his kingdom, but eventually succumbed to the combined attack of a confederation of the Hindu princes of Malwa and Magadha.*

These are the historical data. Scanty as they are, they serve to establish the fact that during a long period of time swarms of nomadic people, whose outlandish names are conveniently summed up in the generic term Scythian, poured into India, conquered and governed. Their coins are now the sole memorial of their rule, but their inroads probably began many centuries before coins were struck or annals compiled. Of the people themselves all traces seem to have vanished, and the student who enquires what has become of them finds nothing more tangible than the modern conjecture that they are represented by the Jāts and Rajputs. But the grounds for this opinion are of the flimsiest description and consist mainly of the questionable assumption that the people who are called Jāts at the present day must have something to do with the people who were known to Herodotus as Getae. Now apart from the fact that resemblances of names are mostly misleading—witness the Roman identification of these very Getae with the Goths—we have good historical reasons for believing that the Scythian invaders of India came from a region occupied exclusively by broad-headed races and must themselves have belonged to that type. They were, by all accounts, nations or hordes of horsemen, with broad faces and high cheek-bones, short and sturdy of stature, and skilled in the use of the bow. In their original homes on the Central Asian steppes their manner of life was that of pastoral nomads; and their instincts were of the predatory order. It seems therefore primâ facie unlikely that their descendants are to be looked for among tribes who are essentially of the long-headed type, tall, heavy men without any natural aptitude for horsemanship, settled agriculturists with no traditions of a nomadic and marauding past. Still less probable is it that waves of foreign conquerors, entering India at a date when the Indo-Aryans had long been an organized community, should have been absorbed by them so completely as to take rank among their most typical representatives, while the form of their heads, the most persistent of racial distinctions, was transformed from the extreme of one type to the extreme of another without leaving any trace of the transitional forms involved in the process. Such are the contradictions which beset the attempt to identify the Scythians with the Jāts and Rajputs. The only escape from them seems to lie in an alternative hypothesis which is suggested by the measurements summarised in the Scytho-Dravidian table. These data show that a zone of broad-headed people may still be traced southwards from the region of the Western Punjab, in which we lose sight of the Scythians, right through the Deccan till it attains its furthest extension among the Coorgs. Is it not conceivable that this may mark the track of the Scythians who first occupied the great grazing country of the Western Punjab and then, pressed upon by later invaders and finding their progress eastward blocked by the Indo-Aryans, turned towards the south, mingled with the Dravidian population and became the ancestors of the Marāthas ? The physical type of the people of the Deccan accords fairly well with this theory, while the arguments derived from language and religion do not seem to conflict with it. For, after entering India the Scythians readily adopted an Aryan language written in the Kharosthi character and accepted Buddhism as their religion. These they would have carried with them to the south. Their Prākrit speech would have developed into Marāthi, while their Buddhistic doctrines would have been absorbed in that fusion of magic and metaphysics which has resulted in popular Hinduism. Nor is it wholly fanciful to discover some aspects of Marātha history which lend it incidental support. On this view the wide-ranging forays of the Marāthas ; their guerrilla methods of warfare ; their unscrupulous dealings with friend and foe ; their genius for intrigue, and their consequent failure to build up an enduring dominion ; and finally the individuality of character and tenacity of purpose which distinguish them at the present day—all these may be regarded as part of the inheritance which has come to them from their Scythian ancestors.

Footnotes

[* The geographical isolation of India has probably been overestimated (V. A. Smith, History of Fine Art in India and Ceylon, 1911, p. 377). Commercial intercourse with the Tigris-Euphrates valley was active during the period 700-300 B.C. (J. Kennedy, “The Early Commerce of Babylon with India,” Journal Royal Asiatic Society, 1908, pp. 241-88). At the close of the 1st century A.D., while slaves were imported into Western India, and the trade in Abyssinian slaves has left evidence of negro blood among some castes in the same region (Periplus Maris Erythraei, ch. 49 ; Bombay Gazetteer, xi., 1883, p. 433 n.).]

• Professor Huxley’s comparison of the shape of India to “the diamond on a pack of cards, having a north angle at Ladakh, a south angle at Cape Comorin, a west angle near the mouth of the Indus, and an east angle near that of the Ganges,” is possibly more accurate than that adopted in the text. It brings out the great projections of the Punjab and Kashmir towards the north and the long straight line of frontier which forms the northwestern side of the diamond. On the whole, however, the triangular aspect seems to catch the eye more as one looks at a map and is thus better suited for descriptive purposes. Huxley’s description is to be found in the first volume of the Journal of the Ethnological Society of London. His simile is curiously analogous to the “rhomboid” of Eratosthenes and other Greek geographers.
• In an instructive paper recently published Professor Kabbadias, Director of Antiquities in Greece, shows that in pre-historic times fortified towns occupied the place taken in other countries by pile-dwellings, Man, Decr., 1904, No. 112.

• Transactions of the Royal Society of Edinburgh, Vol. XL. Pars I. (No. 6).
• Nesfield’s Brief View of the Caste System of the North-West Provinces and Oudh.
• J. A. I., XX, 235. † “Contributions to the Craniology of the People of the Empire of India.” Transactions of the Royal Society of Edinburgh, Vol. XXXIX., Part III. (No. 28) ; Vol. XL., Part I. (No. 6). ‡ Journal of Anatomy and Physiology, Vol. XXVII., p. 20.

• An effective parallel might be drawn between the predatory invasions of the Rajputs and the settlements effected by the Normans in Sicily, Southern Italy and Greece. Both sets of movements arose from similar impulses, both have left unmistakable traces behind, and both ended in the comparative absorption of the conquering race.
• L’Anthropologie, IX., 407. Mémoire sur les Huns blancs.
• J. A. S. B., Vol. LXIX., Part III., 190. [Major A. Playfair, The Garos, 1909, pp. 18 et seq.]
• Ripley. The Races of Europe, p. 45.

• J. R. A. S., XVI., 172-200. † Zimmer, Altindisches Leben, pp. 15 and 101.
• Quart. Jour. Geol. Soc., XXIX. (1873). † Mem. Geol. Survey of India, XXXI., Pt. 2.

• Mr. Romesh Chandra Dutt, C.I.E., pointed out long ago that ‘ aboriginal blood enters largely in [sic] the existing Brahman community of Bengal.’ Calcutta Review, LXXV., p. 238.