to Umbrians, and the course of Sabine migration, it is certain that Latium was many times conquered. Pliny names its occupants in the following order: (iii. 9,) Aborigines, Pelasgians, Arcadians, Siculians, Auruncans, Rutulians. Though nothing can be made of the details, we cannot be wrong in believing that it was a very mixed population, having, in all probability, at least four elements, viz., 1. Some primitive race-say Enotrians; 2. Colonists from the Grecian seas; 3. Siculians; 4. Oscans. To all these, in Rome itself, the Sabines were superadded. Under such circumstances, it would seem to be a miracle if the ultimate Roman language was not extremely composite. No doubt, we may be told that Siculian, Oscan, Greek -yes, and Celtic-all belong to the Indo-European stock, and have many words and principles in common; that words may really be native, which at first sight seem to be imported; that Latin has a sensible fraction of its primitive vocabulary in common with Sanscrit; indeed, words which do not exist in Greek, as ignem, ensem, regem-Sansk. agnim, asim, rājam ; and that we do not imagine an Indian migration into Italy, in order to impart these terms to the Romans. All this we admit. But, to put a parallel case, supposing that European literature, earlier than the fifteenth century, had utterly perished, but that tradition preserved the fact of the Saxon and of the Norman conquests of Britain, we surely should be justified in the à priori inference that the English tongue contains at least three distinct elements, British, Saxon, and Norman. Which of these would predominate, nothing could be said à priori; and, in the actual business of separating them, we should be liable to various errors of detail. We might easily suppose that Night is of Norman origin, or that Air is of Welsh-awyr, while, in fact, night is independent of nuit; and the Welsh awyr was first expelled by the Saxon lyft; and this in turn by the Norman air. Knowing how complicated a problem we have in hand, we must learn extreme caution, and much diffidence, as to special points. But a combined etymological argument is of such a kind, (however logicians may explain it,) that the whole is stronger than any of the parts; and it is often reasonable to speak more confidently of the conclusion than of the premises separately.

In endeavouring to analyze the Latin tongue, and mark off its several constituents, we must expect to fall upon words which were common to them all, and are not rightly to be referred to any special one; yet, through the fragmentary state of our know

ledge, we shall not always be aware of this. Such cases, nevertheless, will be quite exceptive, if the colliding languages differed in something more than dialect; and in many instances we may judge almost by inspection whether a word is imported or not. That rex is native Latin, we discern from its connection with rego; a verb which has the senses, (1.) to point or guide; as regio and dirigo prove; (2.) to rule; and unless all are native, the whole family has been transplanted and acclimated with deceptive success. Hence, though the Erse has righ, and the Sanscrit rāja, we are not hereby tempted to doubt that rex is native. On the contrary, Bacteùç has no sources in Greek, and does not appear likely to be itself a root; finding, then, a SyroArabian verb, mashal, (to rule,) we are led to believe that the Hellenes, during their residence in Asia, before they reached Greece, picked up this and many other words from some people who spoke a Syro-Arabian dialect. We may apply such principles unskilfully; but the principles are sound. There are ways of discriminating words certainly native, and words probably imported; and, upon applying such methods to the Latin tongue, we find a great mass of words of which no account can be given. When they are un-Greek, they are not forthwith "barbarian," or intrusive; indeed, on comparing rex and Baaλɛù, we must judge the Greeks to have talked barbarously; though, as opéyw stands for rego, perhaps we may believe that they have still kept the congener of rex, under the form opxauos. But, if we find in Latin two or three words which bear a certain similarity of sound, but none of sense, or if their forms stray from all laws of the language, while, in sense, they are connected,―if, on the contrary, in some other tongue, we see the representatives of these words connected by closer links of sound and sense,—or if, in the one language, we find a well-developed family, in the other an isolated word,-we have a pretty sure mark in what direction the current of language set. Sometimes the mere circumstance, that one tongue has merely the secondary sense of a word, another has both the primary and the secondary, will indicate that the former has borrowed from the latter. If a Chinese were informed what abstraction means in Latin and in English, he would be able (without any other knowledge of either language,) to form a probable opinion, that the English have borrowed from the Latin, and not conversely. These principles are well known; the difficulty is to practise them. Etymo

logy is a quagmire, where a careless walker is easily swallowed up; but it has firm ground for those who know how to pick it.

To return; if we use the word Siculian in an extended sense, we may say that Latin is made up of Siculian, Greek, and Sabine, overlapping as well as combining. But, under the word Siculian, is concealed, not merely everything that the Siculians may have picked up in Italy itself, from Umbrians, Enotrians, or Oscans, but possibly other heterogeneous material. For the similarities of language force us to believe that these tribes migrated from the far east, where they once lived in close company with the progenitors of Persians and Bengalees. If the Siculian stream of migration passed along the continent to the north of Greece, there is a great à priori probability that they were often in close contact with northern peoples, Celts, Scandinavians, Teutons, or even Albanians, Lithuanians, Slavonians. If stray words are found in Latin which seem to belong to any of these, we need not be surprised; much less ought we to adduce them as a reductio ad absurdum of the argument which alleged an Italian corruption of Latin. To summon a tribe of Slavonians so far south is unnecessary to account for Slavonian words, if clear cases of such are found. The communication may have taken place on the banks of the Danube, or elsewhere.

It has for some years been recognized, at least by several English scholars, that there is a remarkable similarity between the CELTIC languages and Latin. In the case of Welsh, it was, I believe, at first supposed that the words must have been introduced by the Roman dominion in Britain; but when the likeness was found to exist in the Erse, and that the Erse was even more like to Latin (as regards the consonants) than the Welsh is, this idea, of course, fell to the ground. The scholar and physiologist who first pressed into notice the strong similarities of the Celtic to the Indo-European languages, and claim

In reasoning thus, we are not seeking to determine “the original” home, or to mount up at a single stride to the very beginning of mankind, as Niebuhr seems to think, (vol. 1. pp. 53, 54, 4th Eng. Ed.) Languages cannot have grown I up on the soil, and have such likenesses and such unlikenesses as we now see. If any reasoning at all on these subjects is trustworthy, the likeness of Gothic and

Latin to Sanscrit proves locomotive transmission of language from common points, and this can only have been by migrations.

7 Welsh and Irish scholars have, I believe, long declared that the first population of Italy must have been Celtic. But their principles of reasoning were so incautious, that their inferences passed for nothing.

ed a place for Celtic within that group,-Dr. Prichard,—has naturally fixed his attention with so much strength on the primitive relations of all these tongues as to be jealous and suspicious of an argument, which alleges that one has borrowed from the other. Some ten years ago, by his favour, I read a M.S. of a vocabulary, (the composition of Dr. Stratton, formerly of Aberdeen,) which compared the Gaelic with the Latin tongue in alphabetical order, without comment or development. From this vocabulary, Prichard gives an extract in his chapter on the Italian nations, and finds it entirely to confirm his views, that the Roman language has not suffered any large admixture by a foreign action. What is or was Dr. Stratton's opinion, I never heard. His vocabulary first suggested to me the value of this inquiry, but that is all. Having now been led to a fuller examination of the Welsh and Gaelic dictionaries, I find not only a far greater abundance of material (especially in the Welsh) than I could have imagined, but also that, by grouping the words aright, conclusions result such as I had not expected, and adverse to those of Dr. Prichard.

It may be imagined that the Welsh is certain to have received a large stock of words from the Romans, even if it be allowed that the Erse and Gaelic can have admitted few except ecclesiastical ones. But in order to remove any incredulity as to the value of Welsh to us when it stands alone, a few examples shall be first produced of a special kind :

·

Ffun, breath, spirit. [Eng. Fun?]

Ffen, flowing princi

ple, air.

from Italy; but, for that very reason, Ffynetr is likely to be native; and when we see its sense, and the whole family of words, there is no longer doubt of it. Ffynetr is formed regularly from the root Ffun, (the Welsh representative of лvéш,) and means an air-hole of a certain kind, viz. a chimney. Fe

nestra has no derivation in Latin. It is probable that the Latins imported Fenetr (with Fons) from a Celtic people, and (slightly modifying its sound) changed its sense from an air-hole or chimney to mean a window. Afterwards, Fenestra was perhaps carried back to the Celts, with its new sense adhering to it. As for Fons, the gushing of a spring is not unnaturally called a breathing forth; and the relation of Ffynnon to Ffen and Ffyned seems unquestionable. No one could have guessed this of Fons and Fenestra. Surely the Welsh preserves for us here an earlier and a less fractured state of the language.

Space will not allow to comment as fully on other similarities. [Welsh ƒ is often found for Latin m.]

These illustrations have been all taken from the letter Ff, by opening the dictionary at random: yet I can hardly hope that they are every where as abundant. It will be observed, that

8 Cf. German Fassen, to hold or clasp, English Fasten, Fast. Our word Fast combines two roots; 1. Welsh Ffêst, quick, from Ffes, subtlety, power of pene

trating; 2. Welsh Ffas, to tie and fasten, whence also the idea of withholding from food.