I also bring in here, as much suggestively as determinately, the following.

According to the opinion of Zeuss and Gluck, the Danube, (ant. Danubius and Danuvius, Mod. Germ. Donau,) would come in here. These writers derive it from Gael. dan, Ir. dana, fortis, audax, in reference to its strong and impetuous current. This is no doubt the most striking characteristic of the river, but it might also not inappropriately be placed to the root tan, to extend, whence the names of some other large rivers. Gluck considers the ending vius to be simply derivative, and suggests that the Germans, with a natural striving after a meaning, altered this derivative ending into their word ava, aha, ach, or au, signifying river. Though Gluck is a writer for whose opinion I have great respect, and though this is the principle for which I myself have been all along contending, yet I am rather inclined to think that in Danuvius, as in Conovius (the Conway), there is contained a definite appellative, qualified by a prefixed adjective: this seems to me to be brought out more clearly in the Medway, and in the names connected with it.

The word Ister, which, according to Zeuss, is the Thracian name of the Danube, I have elsewhere referred to the Armorican ster, a river. Not that I mean to infer therefrom that the name is Celtic, because ster is only a particular form of an Indo-European word sur. If we refer the prefix is to the Old Norse isia, proruere, then Ister would have the same meaning as that given above to Danubius. But the derivation of Mone, who explains it by y, the Welsh definite article, and ster, a river, making Ister = "The river," I hold with Gluck to be—like other derivations proceeding on the same principle—opposed to all sound philology.

Among the rivers noted for the slowness of their course, the most conspicuous is the Arar or Saone. Cæsar (de Bell. Gall.) describes it as flowing "with such incredible gentleness that the eye can scarcely judge which way it is going." Seneca adopts it as a type of indecision—"the Arar in doubt which way to flow." Eumenius multiplies his epithets—"segnis et cunctabundus amnis, tardusque." The name Sauconna, Sagonna, Saonna, Saone, does not appear before the 4th cent., yet there does not seem any reason to doubt that it is as old as the other. Zeuss (Die Deutschen) and the Editor of "Smith's Ancient Geography" take this as the true Gallic name. And though Armstrong explains both the Arar and the Saone from the Celtic—referring the former to the Obs. Gael. ar, slow, and the latter to Gael. sogh, tranquil or placid, in which he may probably be correct, yet it by no means follows that the name of the Arar is Celtic, for ar is an ancient root of the Indo-European speech. To the same root as the Saone I also put the Seine (Sequăna), and the Segre (Sicŏris), comparing them with Lat. seg-nis. The former of these rivers is navigable for 350 miles out of 414, and the latter is noted in Lucian as "stagnantem Sicorim." Some other rivers, in which the characteristic is less distinct, I also venture to place here, separating this root as well as I can from another p. 58.

Perhaps allied in its root to the last is the Gael. saimh, quiet, tranquil, to which I put the following.

In the Gael. ar, slow, (whence the Arar, p. 118,) is to be found, as I take it, the root of the Welsh araf, mild, gentle. From this Zeuss (Gramm. Celt.), derives the name of the Arrăbo, now the Raab. This root is liable to mix with another, arv, p. 109, of precisely opposite meaning.

I bring in here the word aram or arm, which, both in the names of rivers, and in the ancient names of men, as the German hero Arminius, needs explanation. The authority of Dr. Donaldson may probably have been the cause of the reproduction, even in some of the latest English works, of the mistake of confounding the name Armin, Ermin, or Irmin, with the word hermann, warrior, (from her, army, mann, homo). That it is not so is shown by its appearance in the ancient names of women, as Ermina, Hermena, and Irmina,[51] (daughter of Dagobert the 2nd). And by the manner in which it forms compounds, as Armenfred, Irminric, Irminger,[52] Ermingaud, Irminher, &c. For we may take it as a certain rule that no word, itself a compound, forms other compounds in ancient names. Indeed, the last of the five names, Irminher, (which is found as early as the 7th cent.), is formed from the word her, army, so that, according to the above theory, it would be Her-mann-her. The fact then, as I take it, is that, both in the names of rivers and of men, the root is simply arm or irm, and armin or irmin an extended form, like those found all throughout these pages. As to its etymology, the word aram, arm, in the Teutonic dialects signifying poor or weak, is in itself unsuitable, but I think that the original meaning may perhaps rather have been mild or gentle. The root seems to be found in the Gael. ar, slow; and aram may be a corresponding word to the Welsh araf. Baxter, who, though his general system of river-names I hold to be fallacious, was, for his time, no contemptible etymologist, suggests something of the sort.

In this place I am inclined to bring in the Medway, and some other names connected with it. Among the various derivations which have been suggested for this name, that of Grimm deserves the first place, though I much fear that it is too poetical to be true. He observes, (Gesch. d. Deutsch. Sprach.), comparing it with another name—"In Carl's campaign, A.D. 779, there is a place mentioned in the vicinity of the Weser, called Medofulli, Midufulli; medoful means poculum mulsi, (Hel. 62, 10); it appears to have been a river, which at present bears some other name. Of just a similar meaning is the name of the river Medway flowing through the county of Kent into the Thames—i.e., Ang.-Sax. Meadovaege, Medevaege Medvaege (Cod. Dip.), from vaege, Old Sax. wêgi, Old Norse veig, poculum.... I suggest here a mythological reference: as the rivers of the Greeks and Romans streamed from the horn or the urn of the river-god, so may also the rivers and brooks of our ancestors, in a similar mythic fashion, have sprung from the over-turned mead-cup."

It is a pity to disturb so poetical a theory, coming too as it does from the highest authority, but I much fear that on a comparison of this name with all its related forms, it can hardly be substantiated. For the word does not stand alone—the prefix med is found in several names in which the second part can hardly be taken to mean poculum, and the ending way is found in several names of which the former part cannot mean mulsum. In any case, it seems to me that a Saxon derivation can hardly be sustained. For Medoăcus, (=Medwacus), occurs as the ancient name of a river in Venetia—this appears to be precisely the same name as that of the Medwag or Medway—and in Venetia we can account for a Celtic element, but not for a German. In Nennius the name stands as Meguaid or Megwed; and comparing this with a river called the Medvied(itza) or Medviet(za) in Russia, it would seem rather probable that the form is not altogether false, but that only it should be Medwed instead of Megwed. In that case it would probably be only another form of Medweg, for d and g sometimes interchange in the Celtic dialects, as in the Gaelic uidh and uigh, via, a word which indeed I take to be related to the one in question. Again, in the Meduāna of France and the English Medwin, we have a third form of ending, wân or win. And this may probably only be one of those extended forms in n so common in the Celtic languages.[53] So that the endings way, wân, wied, in Medway, Meduāna, Medvied(itza), may be slightly differing forms of a common appellative (p.p. 62, 63), qualified by the prefix med, which we have next to consider. In Gibson's "Etymological Geography" med is explained as medius—Medway = medium flumen—the river flowing through the middle of the county of Kent—and this I think is the general acceptation. In the case of the Medina, (ant. Mede), which divides the Isle of Wight into two equal parts, I should readily accept such a derivation, but in the case of the Medway it seems to me a feature scarcely sufficiently obvious to give the name. And I should on the whole prefer a derivation from the same root as mead, mulsum, viz., Sansc. mid, to soften, Lat. mitis, Gael. meath, soft, mild—finding in Old Norse mida, to move slowly or softly, the word most nearly approximating to the sense, and thus deriving the name of the Medway from its gentle flow.

Nevertheless it must be observed that as well as the supposed river Medofulli referred to as above by Grimm, we find in a charter of the 10th cent., a river called Medemelacha, which seems evidently to contain the Gael. mealach, sweet, and to mean "sweet as mead." This river is near Medemblik on the Zuyder-zee, and I suppose that the name of the place is corrupted from it.

The following names I place here, though with uncertainty in the case of some of them.

I think, upon the whole, that the general meaning of the root lam, lem, lim, is smoothness. Though the root-meaning seems rather that of clamminess or adhesiveness, as found in Sansc. limpas, Gr. λιπος, Lat. limus, Old Sax. lêmo, Mod. Germ. lehm, Eng. lime, &c.[54] In the Gr. λιμνη, lake, the sense becomes that of smooth or standing water: this, as I take it, is in effect the word found in the Lake Leman, Loch Lomond, &c. Though the word most immediately concerned is the Gaelic liobh, liomh, Welsh llyfnu, to smooth; and the Loch Lomond, (properly Lomon), was also formerly called, as the river which issues from it is still, Leven, being just another form of the same word—v and m interchanging as elsewhere noticed. Hence the Welsh llifo, to pour, p. 46, might be apt to intermix in the following. The Lat. lambo, the primitive meaning of which is to lick, is applied to the gentle washing of a river against its banks—"Quæ loca lambit Hydaspes,"—Horace. Dugdale observes that "at this day divers of those artificial rivers in Cambridgeshire, anciently cut to drain the fens, bear the name of Leam, being all muddy channels through which the water hath a dull or slow passage." In the following names the sense may be sometimes then that of muddiness, though in general, as I take it, that of sluggishness.

From the above form lam, lem, lim, I take to be formed by metathesis alm, elm, ilm. And the lake Ilmen in Russia I take to be in effect the same word as the lake Leman in Switzerland. In the name of another lake in Russia, the Karduanskoi-ilmen, it seems to occur as an appellative. A certain amount of doubt is imported by the coincidence of two names in which we find a sacred character—the river Almo, which was sacred to Cybele, and a sacred fountain Olmius mentioned in Hesiod. The coincidence, however, may be only accidental.

Perhaps from the Gael. foil, slow, gentle, we may get the following.

In the third division of this chapter I put the names in which the sense of spreading seems to be found. This sense may have three different acceptations—first, that, generally, of a wide river—secondly, that of a river relatively broad and shallow—thirdly, that of a river forming an estuary at its mouth.

I bring in here the Padus or Po, which, by Metrodorus Scepsius, a Greek author quoted by Pliny, has been derived from the pine-trees, "called in the Gallic tongue padi," of which there were a number about its source. A derivation like this jars with common sense, for it is unreasonable to suppose that the Gauls, coming upon this fine river, gave it no name until they had tracked it up to its source, and there made the not very notable discovery that it was surrounded by pine-trees. Much more probable is it that they came first upon its mouth, and much more striking would be the appearance that would be presented to them. For, as Niebuhr observes, "the basin of the Po, and of the rivers emptying themselves into it was originally a vast bay of the sea," which by gradual embanking was confined within its present channels. As then the mouth of the Padus was a vast estuary, so in the Gael. badh, a bay or estuary, I find the explanation of the name. The root, I apprehend, is Sansc. pat, Lat. pateo, pando, &c., to spread, and hence, I take it, the name Bander, of several small bays on the S.W. coast of Asia, of Bantry Bay in Ireland, and of Boderia, the name given by Ptolemy to the Firth of Forth.

In the Sansc. parth, to spread or extend, we may perhaps find the origin of the following. Can the name of the Parthians be hence derived, in reference to their well-known mode of fighting?

In the sense of "that which spreads" I am inclined to bring in the root ta, tav, tan, tam. While in the Gaelic we find tain, and the Obs. ta, water, taif, sea—in the Welsh we have the verbs taenu and tafu, to expand or spread. The latter, I think, must contain the root-meaning; and the appellatives must rather signify water of a spreading character. In this sense we find the words to, tû, tau, in the Hungarian dialects signifying a lake. The Sansc. has tan, to extend, but we must presume a simpler form ta, corresponding with the above Obs. Gael. word for water. Mone explains tab, as in Tabuda (the Scheldt), as "a broad river, especially one with a broad mouth." This sense no doubt obtains in many of the names of this group, for, as well as the Scheldt; the Tay, Taw, Teign, and Tamar, all have this character in a more or less notable degree. In other cases the sense may be that of comparative broadness—thus the Timavus, though little more than a mile long, is 50 yards broad close to its source. So the characteristic of the Dane, as noticed by the county topographers, is that it is "broad and shallow." And the feature which strikes the topographer is of course that which would naturally give the name. There are, however, some other roots which might intermix, as Sansc. tan, resonare, Lat. tono, Germ. tönen, &c. Also Gael. and Ir. taam, to pour; Gael. and Ir. tom, to bathe, Welsh and Ir. ton, unda.

From the root tan, to extend, we may probably also derive the word tang found in Hung. tenger, sea, Ostiakic (an Ugric dialect of the Finnic class) tangat, river, and in the Dan. tang, sea-weed, which probably contains a trace of an older sense.

CHAPTER VI.

CHARACTER OF COURSE.

In the inscription of Pul found at Nineveh, as deciphered in the Proceedings of the Asiatic Society, vol. 19, pt. 2, the Euphrates is called the Irat, which is conjectured by the translator to have been a local name. It seems to be from the Sansc. irat (=Latin errans, Eng. errant), from the verb ir, Lat. erro, to wander. The same word seems to be found in the Irati of Spain—perhaps also in the Orontes (=Irantes=Irates), of Syria. Possibly also in the Erid-anus or Po, though I am rather inclined to agree with Latham that the word contained therein is only ridan.[57] Perhaps then the form Irt or Urt in river-names may be a contracted form of irat, as we find it in the Germ. irrthum, a mistake.