was driven by priestly tyranny to Africa and the shores of the Levant. Yet here, to this day, are names on the map of Spain, current in the mouths of the people, which imply the existence of all these different populations, and substantiate the fact of all these changes of dynasty and of race.

It is curious to compare with William von Humboldt's masterly essay another attempt of the same kind to trace an ancient population by means of local names. We allude to the pamphlet of Ludwig Steub on the original inhabitants of Rætia and their connexion with the Etruscans. Unfortunately the

conditions of the problem to be solved were very different in the two cases. Humboldt had (as we have in the Welsh) a living language to refer to for the meaning of the words and syllables which he found composing the names of rivers and of places. Steub, on the other hand, had nothing to fall back on except the scanty remains of Etruscan, which never have been, and probably never will be, successfully interpreted. He describes the feeling which led him to undertake his task in the following manner :

'When I was passing, last summer, in the Rætian Alps, I was more than ever attracted by those singular and fine-sounding names which accompany the traveller along the high-road, and meet him in the most remote valleys and on the wildest peaks. At Bludenz, in the Vorarlberg, I heard men speak of the Alps, Tilisuna and Blisadona; in the Vintschgau I saw places called Naturns, Schluderns, and Schlanders; the peaks Firmisaun and Similaun were pointed out to me; and I heard of the villages of Villanders, Velthurns, and Gufidaun or Altrans, Sistrans, and Axams: everywhere I went I was met by the sound of these mysterious names.'

In the first place, we may assume the presence of a Tuscan population in Ratia, for we have a distinct statement in Pliny and Justin, that the Rætians were supposed to be Etruscans driven by the Gauls from their homes, in the valley of the Po. Niebuhr and Otfried Müller adopt the view that this people came thence, as from the cradle of their race, rather than were driven thither.† It seems to us that these last writers maintain their view rather on the ground of what they suppose to be an à priori probability than on express testimony of any kind; but this point is immaterial for our present purpose. the second place, we find a mass of strange sounding names, inexplicable by a reference to any known language, all crowded together in this corner of the Alps. In the third place, without

In

* Ueber die Urbewohner Rætiens und ihren Zusammenhang mit den Etruskern." München, 1843. Vorrede, s. 111.

Pliny, iii. 20; Justin, xx. 5; Livy, v. 33; Niebuhr, 'Röm. Gesch.' (3te. Ausg.), B. i. s. 127; Otfried Müller, Etrusker,' s. 163.

6

attaching

attaching undue value to guesses at the possible meaning of particular syllables, these names easily fall into combinations of sounds, or letters closely analogous to the combinations presented by acknowledged Tuscan names, and by the remains of Etruscan inscriptions. It must be admitted, however, that the corruption which some of the Rætian names have gone through makes the process of reduction one of a somewhat arbitrary character; yet, on the whole, we think Mr. Steub has established the great probability of the theory he has adopted. We hesitate to go with him when he invades foreign territory, and claims as Etruscan a good many elements of local names which are explicable by us, with more chance of being right, by a reference to the Celtic. The case of the Rhine is one of these.

Let us now turn to a subject which has an interest of a more special kind for Englishmen, and to which Mr. Taylor has deservedly devoted a large portion of his book. We mean the local nomenclature of Great Britain.

It will be found, as we have already observed, that almost all the names of rivers in Great Britain are of Celtic origin. Our author says:

'One class of local names is of special value in investigations relating to primæval history. The river-names, more especially the names of important rivers, are everywhere the memorials of the very earliest races. These river-names survive where all other names have changed-they seem to possess an almost indestructible vitality. Towns may be destroyed, the sites of human habitations may be removed, but the ancient river-names are handed down from race to race; even the names of the eternal hills are less permanent than those of rivers. Over the greater part of Europe-in Germany, France, Italy, Spain-we find villages which bear Teutonic or Roman names, standing on the banks of streams which still retain their ancient Celtic appellations. Throughout the whole of England there is hardly a river-name which is not Celtic. By a reference to the map prefixed to this volume, it will be seen that those districts of our island which are dotted thickly with Anglo-Saxon and Scandinavian village-names are traversed everywhere by red lines, which represent the rivers, whose names are now almost the sole evidence that survives of a once universal Celtic occupation of the land.'—p. 204.

He then divides the elements of these river-names into two classes the one consisting of words which originally meant simply 'river' or 'water,' and the other containing those which he calls adjectival,' consisting of the Celtic words applicable as epithets to streams."

It appears that the names of the larger rivers of Europe, and of many of the smaller streams, contain one of the five following Celtic roots for water' or 'river.' These roots are

1. Avon,

1. Avon, or aon (river).

2. Dwr, or ter (water).

3. Uisge, or wysg, wy, is, eo, oise, usk, esk, ex, ax (water). 4. Rhe, or Rhin (swift motion or current).

5. Don, or dan (uncertain).

The word Stour occurs also constantly. It is found, perhaps, in the Stura, or Astura of Latium, mentioned by Festus and by Pliny, and in other English and Continental river-names, whilst its origin seems very doubtful.

*

The adjectival elements of Celtic river-names are principally the following:

1. Garw (rough).

2. All (Gaelic, white).

3. Ban (Gaelic, white).

4. Tam, Taw (spreading-quiet).

5. Cam (crooked).

6. Clith (Gaelic, strong), or perhaps Clyd (Welsh, warm). 7. Dhu (black).

We cannot undertake to illustrate in detail the application of these words, which meet us in various combinations in the river-names, not only of our own country, but of the whole continent of Europe-thus attesting the wide extent of ground once occupied by one or other of the great divisions of the Celtic race. Our readers no doubt know many Avons in England, in Scotland, and in Ireland. In the last country we have the river (Avon dhu) immortalised by Spenser as—

'Swift Awniduff, which of the English-men

Is cal'de Blackwater.'

The same element probably forms the root of the name Avignon, where the Durance (derived from dwr) falls into the Rhone. If we doubt our author's etymology of Trent as a contracted form of Derwent, or Darent, we are unable to supply a better. At any rate, we should reject that adopted by Milton, when he speaks

'Of Trent, who like some earth-born giant spreads
His thirty arms along the indented meads.'

We are of opinion, however, that in some cases an attempt has been made to give significance to the last syllable of rivernames, as in the case of the Calder, where that syllable is a mere termination. The Norse etymology of kaldr-cold or coolseems more plausible than the Celtic one of winding water,' in which the der' would represent the root 'dwr.' We may

Pliny, i., 152, 16. Festus (Müller), p. 317. Compare W. von Humboldt, 'Prüfung,' s. 114.

remark

6

remark too in passing that the 'Roy' in Invernessshire is more likely to mean the red stream' than to be connected with the Rhe' or 'Rhin.' With regard to the root 'don,' it is said that it still exists with the sense of river or water among the tribes of the Caucasus.

It is curious to observe that in very many cases the original word, as it became unintelligible in its first sense to the subsequent inhabitants of a region, was overlaid by a synonyme in their own language. Thus Durbeck in Nottinghamshire and Durbach in Germany contain the Celtic root dwr' with the Teutonic or Northern 'beck' or 'bach' added to it. Dourwater

is a translation of the same kind.* In the case of Wansbeckwater,' as our author observes, these synonymous elements are accumulated-Wan' is probably a corruption of Avon; 'Beck' is Saxon or Norse; and Water' is an English addition. There is the history of three races in this single word. It is perhaps going too far to regard the s as the remains of the Gaelic uisge,' and we should think it far more likely to be the mark of the genitive case. Brindon Hill, in Somersetshire; Mon-gibello, with its Arabic suffix; and the Puente de Alcantara,' in Spain, are examples of the same process of accumulation of synonymes. There is one great English river which we think Mr. Taylor has omitted to notice, and which deserves a passing word-we mean the Severn. In Ptolemy the name which is given to it is 'Sabriana;' in Tacitus it is called 'Sabrina.'† In Nennius the river is called 'Habren;' and we believe that the Welsh name, for a part of it at least, is still 'Hafren'-the S being represented by H, as in many other instances. Whether this word is connected with the Welsh word 'Hafru,' to render sluggish, in allusion to its tranquil stream as compared with a mountain torrent, we will not undertake to determine.

Mr. Taylor proceeds next to deal with the names of another great natural feature of all lands. He says

In antiquity and immutability the names of mountains and hills come next in value to the names of rivers. The names of these great landmarks have been transmitted from race to race very much in the same way and from the same causes as the names of rivers.

The modern Welsh names for the head and the back are "Pen" and "Cefn." We find these words in a large number of mountain names. The Welsh " cefn" (pronounced keven), a back or ridge, is

* See Palgrave, English Commonwealth,' vol. i. p. 450, n.

t'Annal. xii., 31. There is little doubt but that the correction of Aufona for Antona' must be received, and that the two rivers meant are the Severn and the Avon ; but the passage is corrupt; probably Ernesti's conjecture of 'cinctisque ' for 'cunctaque' is right.

Chapter 68. Compare Zeuss, 'Gramm. Celtica,' p. 144.

very common in local names in Wales, as in the case of Cefn Coed or Cefn Bryn. In England it is found in the "Chevin," a ridge in Wharfdale; in Keynton, a name which occurs in Shropshire, Dorset, and Wilts; in "Chevening" on the great ridge of North Kent; in Chevington in Suffolk and Northumberland; in Chevy Chase, and the Cheviot Hills; in the "Gebenna Mons," now "les Cevennes" in France; in "Cape Chien" in Brittany.'-p. 230.

'The Welsh Pen, a head, and by metonymy the usual name for a mountain, is widely diffused throughout Europe. The south-easterly extension of the Cymric race is witnessed by the names of the Pen-nine chain of the Alps, the A-penn-ines, a place called Penne, anciently Pinna, in the high Apennines, and Mount Pindus in Greece. The ancient name of Penilucus, near Villeneuve, is evidently Pen-y-llwch -the head of the lake. We find Pen-herf and the headland of Penmarch in Brittany; and there is a hill near Marseilles which is called La Penne. In our own island, hills bearing this name are very numerous. We have Penard, Penhill, and Pen, in Somerset; Upper and Lower Pen in Staffordshire; and Pann Castle near Bridgenorth. The highest hill in Buckinghamshire is called Pen. One of the most conspicuous summits in Yorkshire is called Pennigant. We have Pendleton and Penketh in Lancashire; Penshurst in Sussex; in Cumberland Penrith, the head of the ford; and in Herefordshire Pencoid, the head of the wood. In Cornwall and Wales the root Pen is of perpetual occurrence, as in the cases of Penrhyn and Pendennis (Pen Dinas) in Cornwall, and Penmaenmawr and Penrhos in Wales.'pp. 231-2.

In a note to the passage just quoted our author suggests with a query, that penny,' the coin, is derived from this root; and he repeats this conjecture at p. 455, assuming that it might be so called from the head upon the coin, like 'tester,' from 'teste.' We are bound to say that we think this etymology cannot be supported. The old Norse word 'Penningr' signified the tenth part not only of the ounce (eyrir) but of the ell of cloth, which was one of the articles of daily barter.* It is very difficult to suppose that the Scandinavians derived a word of this kind from the Cymric branch of the Celts. The word is moreover used by Otfried-that is to say, as early as the year 870. Another statement in the same note requires to be noticed. Mr. Taylor observes that Peña is Spanish for rock, and that Penna, in Italian, means a mountain summit; and a reference is made to Diez's Etymological Dictionary of the Romance Languages. On turning to Diez, however, the reader will find that

* See Egilsson, Lexicon Poeticum Antiquæ Linguæ Septentrionalis,' in v. According to a statement made by Mr. Marsh, in his Lectures on the English Language,' p. 127, note, 'penny' was used also to denote a duodecimal part; but the reader must bear in mind that the tenth part of the later tirætt hundrað' (100) was equal to the twelfth part of the old 'tolfrætt hundrað' (120).

he