the withe were known to them; so also were the bear and wolf, the hare, the mouse, and the snake, as well as the goose and raven, the quail and the owl. Cattle, sheep, goats, and swine were all kept; the dog had been domesticated, and in all probability also the horse. Last, but not least, boats were navigated by means of oars, the boats themselves being possibly the hollowed trunks of trees.

This account of the primitive community is necessarily imperfect. There must have been many words, like that for "river," which were once possessed by the parent-speech, but afterward lost in either the Eastern or Western branches of the family. Such words the comparative philologist has now no means of discovering, he must accordingly pass them over along with the objects or ideas which they represent. The picture he can give us of the speakers of the primeval Indo-European language can only be approximately complete. Moreover it is always open to correction. Some of the words we now believe to have been part of the original stock carried away by the derived dialects of Asia and Europe may hereafter turn out to have been borrowed by one of these dialects from another, and not to have been a heritage common to both. It is often very difficult to decide whether we are dealing with borrowed words or not. If a word has been borrowed by a language before the phonetic changes had set in which have given the language its peculiar complexion, or while they were in the course of progress, it will undergo the same alteration as native words containing the same sounds. The phonetic changes which have marked off the High German dialects from their sister tongues do not seem to go back beyond the fall of the Roman Empire, and words borrowed from Latin before that date will accordingly have submitted to the same phonetic changes as words of native origin. Indeed, when once a word is borrowed by one language from another and has passed into common use, it soon becomes naturalized, and is assimilated in form and pronunciation to the words among which it has come to dwell. A curious example of this is to be found in certain Latin words which made their way into the Gaelic dialects in the fourth or fifth century. We often find a Gaelic e corresponding to a Welsh p, both being derived from a la

bialized guttural or qu, and the habit was accordingly formed of regarding a c as the natural and necessary representative of a foreign p.

When, therefore, words like the Latin pascha and purpura were introduced by Christianity into the Gaelic branch of the Keltic family, they assumed the form of caisg and corcur.

It is clear that such borrowings can only take place where the speakers of two different languages have been brought into contact with one another. Before the age of commercial intercourse between Europe and India we cannot suppose that European words could have been borrowed by Sanskrit or Persian, or Sanskrit and Persian words by the European languages. But the case is quite otherwise if instead of comparing together the vocabularies of the Eastern and Western members of the Indo-European stock, we wish to compare only Western with Western, or Eastern with Eastern. There our difficulties begin, and we must look to history, or botany, or zoology for aid. From a purely philological point of view the English hemp, the Old High German hanf, the Old Norse hanpr, and the Latin cannabis might all be derived from a common source, and point to the fact that hemp was known to the first speakers of the Indo-European languages in North-western Europe. But the botanists tell us that this could not have been the case. Hemp is a product of the East which did not originally grow in Germany, and consequently both the plant itself and the name by which it was called must have come from abroad. So, again, the lion bears a similar name in Greek and Latin, in German, in Slavonic, and in Keltic. But the only part of Europe in which the lion existed at a time when the speakers of an Indo-European language could have become acquainted with it were the mountains of Thrace, and it must, accordingly, have been from Greek that its name spread to the other cognate languages of the West.

It has been needful to enter into these details before we can approach the question, What was the original home of the parent Indo-European language? They have been too often ignored or forgotten by those who have set themselves to answer the question, and to this cause must be ascribed the larger part of the misunderstandings and false conclusions to which the inquiry has given birth.

Until a few years ago I shared the old belief that the parent-speech had its home in Asia, probably on the slopes of the Hindu Kush. The fact that the languages of Europe and Asia alike possessed the same words for "winter" and "ice" and " snow," and that the only two trees whose names were preserved by both the "birch" and the pine"-were inhabitants of a cold region, proved that this home did not lie in the tropics. But the uplands of the Hindu Kush, or the barren steppes in the neighborhood of the Caspian Sea, or even the valleys of Siberia, would answer to the requirements presented by such words. Taken by themselves they were fully compatible with the view that the first speakers of the Indo-European tongues were an Asiatic people.

But when I came to ask myself what were the grounds for holding this view, I could find none that seemed to me satisfactory. There is much justice in Dr. Latham's remark that it is unreasonable to derive the majority of the Indo-European languages from a continent to which only two members of the group are known to belong, unless there is an imperative necessity for doing so. These languages have grown out of dialects once existing within the parent-speech itself, and it certainly appears more probable that two of such dialects or languages should have made their way into a new world, across the bleak plains of Tartary, than that seven or eight should have done so. The argument, it is true, is not a strong one, but it raises at the outset a presumption in favor of Europe. Before the dialects had developed into languages, their speakers could not have lived far apart; there is, in fact, evidence of this in the case of Sanskrit and Persian; and a more widely spread primitive community is implied by the numerous languages of Europe than by the two languages of Asia. A widely spread community, however, is less likely to wander far from its original seat than a community of less extent, more especially when it is a community of herdsmen, and the tract to be traversed is long and barren.

Apart from the general prejudice in favor of an Asiatic origin due to old theological teaching and the effect of the discovery of Sanskrit, I can find only two arguments which have been supposed to be of sufficient weight to determine the

choice of Asia rather than of Europe as the cradle of Indo European speech. The first of these arguments is linguistic, the second is historical, or rather quasi-historical.

On the one hand it has been laid down by eminent philologists that the less one of the derived languages has deflected from the parent-speech, the more likely it is to be geographically nearer to its earliest home. The faithfulness of the record is a test of geographical proximity. As Sanskrit was held to be the most primitive of the Indo-European languages, to reflect most clearly the features of the parent-speech, the conclusion was drawn that that parent speech had been spoken at no great distance from the country in which the hymns of the Rig-Veda were first composed. The conclusion was supported by the second argument drawn from the sacred books of Parsaism. In the Vendidâd the migrations of the Iranians were traced back through the successive creations of Ormazd to Airyanem Vaêjô, "the Aryan Power," which Lassen localized near the sources of the Oxus and Jaxartes. But Bréal and De Harlez have shown that the legends of the Vendidâd, in their present form, are late and untrustworthy

later, in fact, than the Christian era ; and even if we could attach any historical value to them, they would tell us only from whence the Iranians believed their own ancestors to have come, and would throw no light on the cradle of the IndoEuropean languages as a whole. The first argument is one which I think no student of language would any longer employ. As Professor Max Müller has said, it would suffice to prove that the Scandinavians emigrated from Iceland. But to those who would still urge it, I must repeat what I have said elsewhere. though in many respects Sanskrit has preserved more faithfully than the European languages the forms of primitive IndoEuropean grammar, in many other respects the converse is the case. In the latest researches into the history of Indo-European grammar, Greek holds the place once occupied by Sanskrit. The belief that Sanskrit was the elder sister of the family led

Al

* Bréal, "Mélanges de Mythologie et de Linguistique" (1878), pp. 187-215; De Harlez, "Introduction à l'Etude de l'Avesta," pp. cxcii., sqq. Compare Darmesteter's Introduction to the Zend-Avesta, pt. 1, in "The Sacred Books of the East.”

to the assumption that the three short vowels ǎ, ě, and Ŏ have all originated from an earlier ǎ. I was, I believe, the first to protest against this assumption in 1874, and to give reasons for thinking that the single monotonous à of Sanskrit resulted from the coalescence of three distinct vowels. The analogy of other languages goes to show that the tendency of time is to reduce the number of vocalic sounds possessed by a language, not the contrary. In place of the numerous vowels possessed by ancient Greek, modern Greek can now show only five, and cultivated English is rapidly merging its vowel sounds into the so-called "neutral"ə. Since my protest the matter has been worked out by Italian, German and French scholars, and we now know that it is the vocalic system of the European languages rather than of Sanskrit which most faithfully represents the oldest form of Indo-European speech. The result of the discovery, for discovery it must be called, has been a complete revolution in the study of Indo-European etymology, and still more of Indo-European grammar, and whereas ten years ago it was Sanskrit which was invoked to explain Greek, it is to Greek that the "new school" now turns to explain Sanskrit. The comparative philologist necessarily cannot do without the help of both; the greater the number of languages he has to compare the sounder will be his inductions; but the primacy which was once supposed to reside in Asia has been taken from her. It is Greek, and not Sanskrit, which has taught us what was the primitive vowel of the reduplicated syllable of the perfect and the augment of the aorist, and has thus narrowed the discussion into the origin of both.

Until quite recently, however, the advocates of the Asiatic home of the IndoEuropean languages found a support in the position of the Armenian language. Armenian stands midway, as it were, between Persia and Europe, and it was imagined to have very close relations with the old language of Persia. But we now know that its Persian affinities are illusory, and that it must really be grouped with the languages of Europe. What is more, the decipherment of the cuneiform inscriptions of Van has cast a strong light on the date of its introduction into Armenia. These inscriptions are the records of kings whose capital was at Van, and who

marched their armies in all directions during the ninth, eighth, and seventh centuries before our era. The latest date that can as yet be assigned to any of them is B.C. 640. At this time there were still no speakers of an Indo-European language in Armenia. The language of the inscriptions has no connection with those of the Indo European family, and the personal and local names occurring in the countries immediately surrounding the dominions of the Vannic kings, and so abundantly mentioned in their texts, are of the same Enguistic character at the Vannic names themselves.

The evidence of classical writers fully bears out the conclusions to be derived from the decipherment of the Vannic inscriptions. Herodotos* tells us that the Armenians were colonists from Phrygia, the Phrygians themselves having been a Thrakian tribe which had migrated into Asia. The same testimony was borne by Eudoxos, who further averred that the Armenian and Phrygian languages resembled one another. The tradition must have been recent in the time of Herodotos, and we shall probably not go far wrong if we assign the occupation of Armenia by the Phrygian tribes to the age of upheaval in Western Asia which was ushered in by the fall of the Assyrian Empire.

Professor Fick has shown that the scanty fragments of the Phrygian language that have survived to us belong to the European branch of the Indo European family, and thus find their place by the side of Armenian.

Instead, therefore, of forming a bridge. between Orient and Occident Armenian represents the furthermost flow of IndoEuropean speech from West to East. And this flow belongs to a relatively late period. Apart from Armenian we can discover no traces of Indo-European occupation between Media and the Halys until the days when Iranian Ossetes settled in the Caucasus and the mountaineers of Kurdistan adopted Iranian dialects. must reiterate here what I have said many years ago: if there is one fact which the Assyrian monuments make clear and indubitable, it is that up to the closing days of the Assyrian monarchy no Indo-European languages were spoken in the vast

* vii. 73.

I

† According to Eustathios (in Dion. v. 694).

tract of civilized country which lay between Kurdistan and Western Asia Minor. South of the Caucasus they were unknown until the irruption of the Phrygians into Armenia. Among the multitudinous names of persons and localities belonging to this region which are recorded in the Assyrian inscriptions during a space of several centuries there is only one which bears upon it the Indo-European stamp. This is the name of the leader of the Kimmerians, a nomad tribe from the northeast which descended upon the frontiers of Assyria in the reign of Esar-haddon, and was driven by him into Asia Minor. The fact is made the more striking by the further fact that as soon as we clear the Kurdish ranges and enter Median territory, names of Indo European origin meet us thick and fast. We can draw but one conclusion from these facts. Whether the Indo-European languages of Europe migrated from Asia, or whether the converse were the case, the line of march must have been northward of the Caspian, through the inhospitable steppes of Tartary and over the snow covered heights of the Ural mountains.

An ingenious argument has lately been put forward, which at first sight seems to tell in favor of the Asiatic origin of IndoEuropean speech. Dr. Penka has drawn attention to the fact that several of the European languages agree in possessing the same word for " eel," and that whereas the eel abounds in the rivers and lakes of Scandinavia, it is unknown in those cold regions of Western Asia where, as we have seen, it has been proposed to place the cradle of the IndoEuropean family. But it is a curious fact that in Greek and Latin, and apparently also in Lithuanian, the word for "eel" is a diminutive derived from a word which denotes a snake or snake-like creature. This, it has been urged, may be interpreted to mean that the primeval habitat of the Indo-European languages was one where the snake was known, but the eel was not. The argument, however, cannot be pressed. We all agree that the first speakers of the Indo-European languages lived on the land, not on the water, and that they were herdsmen rather than fishermen. Naturally, therefore, they would become acquainted with the snake before they became acquainted with the eel, however much it might abound in the

rivers near them, and its resemblance to the snake would lend to it its name. In Keltic the eel is called "a water-snake," and to this day a prejudice against eating it on the ground that it is a snake exists in Keltic districts. All we can infer from the diminutives anguilla, čyxeλvg, is that the Italians and Greeks in the first instance gave the name to the fresh-water eel, and not to the huge conger.

I cannot now enter fully into the reasons which have led me gradually to give up my old belief in the Asiatic origin of the Indo-European tongues, and to subscribe to the views of those who would refer them to a northern European birthplace. The argument is a complicated one, and is necessarily of a cumulative character. The individual links in the chain may not be strong, but collectively they afford that amount of probability which is all we can hope to attain in historical research. Those who wish to study them may do so in Dr. Penka's work on the "Herkunft der Arier," published in 1886. His hypothesis that Southern Scandinavia was the primitive Aryan home" seems to me to have more in its favor than any other hypothesis on the subject which has as yet been put forward. It needs verification, it is true, but if it is sound the verification will not be long in coming. A more profound examination of Teutonic and Keltic mythology, a more exact knowledge of the words in the several Indo-European languages which are not of Indo-European origin, and the progress of archæological discovery, will furnish the verification we need.

66

Meanwhile, it must be allowed that the hypothesis has the countenance of history. Scandinavia, even before the sixth century, was characterized as the manufactory of nations ;" and the voyages and settlements of the Norse Vikings offer a historical illustration of what the prehistoric migrations and settlements of the speakers of the Indo-European languages must have been. They differed from the latter only in being conducted by sea, whereas the prehistoric migrations followed the valleys of the great rivers. It was not until the age of the Roman Empire that the northern nations became acquainted with the sailing-boat; our Eng

vagina nationum :" Jordanes, De Getarum sive "Quasi officina gentium aut certe velut Gothorum origine, ed. Closs, c. 4.

"the little

lish sail is the Latin sagulum, cloak of the soldier, "borrowed by the Teutons along with its name, and used to propel their boats in imitation of the sails of the Roman vessels. The introduction of the sail allowed the inhabitants of the Scandinavian "hive" to push boldly out to sea, and ushered in the era of Saxon pirates and Danish invasions.

Dr. Penka's arguments are partly anthropological, partly archæological. He shows that the Kelts and Teutons of Roman antiquity were the tall, blue-eyed, fair haired, dolicho-cephalic race which is now being fast absorbed in Keltic lands by the older inhabitants of them. The typical Frenchman of to-day has but little in common with the typical Gaul of the age of Cæsar. The typical Gaul was, in fact, as much a conqueror in Gallia as he was in Galatia, or, as modern researches have shown, as the typical Kelt was in Ireland. It seems to have been the same in Greece. Here, too, the golden-haired hero of art and song was a representative of the ruling class, of that military aristocracy which overthrew the early culture of the Pelo. ponnese, and of whom tradition averred that it had come from the bleak North. Little trace of it now remains: it is rarely that the traveller can discover any longer the modern kinsfolk of the golden-haired Apollo or the blue-eyed Athênê.

If we would still find the ancient blonde race of Northern Europe in its purity we must go to Scandinavia. Here the prevailing type of the population is still that of the broad-shouldered, long-headed blondes who served as models for the Dying Gladiator.

And it is in Southern Scandinavia alone that the prehistoric tumuli and burying-grounds yield hardly any other skeletons than those of the same tall dolichocephalic race which still inhabits the country. Elsewhere such skeletons are either wanting or else mixed with the remains of other races. It is therefore reasonable to conclude that it was from Southern Scandinavia that those bands of hardy warriors originally emerged, who made their way southward and westward and even eastward, the Kelts of Galatia penetrating like the Phrygians before them into the heart of Asia Minor. The Norse migrations in later times were even more extensive, and what the Norse Vikings were able to achieve could have been achieved by their ancestors centuries before.

Now the Kelts and Teutons of the Roman age spoke Indo European languages. It is more probable that the subject populations should have been compelled to learn the language of their conquerors than that the conquerors should have taken the trouble to learn the language of their serfs. We know at any rate that it was so in Ireland. Here the old " Ivernian" population adopted the language of the small band of Keltic invaders that settled in its midst. It is only where the conquered possess a higher civilization than the conquerors, above all, where they have a literature and an organized form of religion, that Franks will adapt their tongues to Latin speech, or Manchus learn to speak Chinese. Moreover, in Southern Scandinavia, where we have archæological evidence that the tall blonde race was scarcely at any time in close contact with other races, it is hardly possible for it to have borrowed its language from some other people. The Indo-European languages still spoken in the country must, it would seem, be descended from languages spoken there from the earliest period to which the evidence of human occupation reaches back. The conclusion is obvious: Southern Scandinavia and the adjacent districts must be the first home and starting-point of the Western branch of the Indo-European family.

If we turn to the Eastern branch, we find that the further East we go the fainter become the traces of the tall blonde race and the greater is the resemblance between the speakers of Indo-European languages and the native tribes. In the highlands of Persia, tall long-headed blondes with blue eyes can still be met with, but as we approach the hot plains of India, the type grows rarer and rarer until it ceases altogether. An Indo-European dialect must be spoken in India by a dark-skinned people before it can endure to the third and fourth generation. As we leave the frontiers of Europe behind us we lose sight of the race with which Dr. Penka's arguments would tend to connect the parentspeech of the Indo-European family.

I cannot now follow him in the interesting comparison he draws between the social condition of the Southern Scandinavians as disclosed by the contents of the prehistoric "kitchen-middens," and the social condition of the speakers of the Indo-European parent-speech according to