at which certain languages are given off from their parent stocks, such must be the case.

Now, although this is a difficulty, it is no greater difficulty than the geologists must put up with. With them also there are the phenomena of transition, and such phenomena engender unavoidable complication. They do so, however, without overthrowing the principles of their classification.

The position of a language in respect to its stage of development is one thing, the position in respect to its allied tongues another.

Two languages may be in the same stage (and, as such, agree), yet be very distant from each other in respect to affiliation or affinity. Stage for stage the French is more closely connected with the English, than the English with the MosoGothic. In the way of affiliation, the converse is the case. Languages are allied (or, what is the same thing, bear evidence of their alliance), according to the number of forms that they have in common; since (subject to one exception) these common forms must have been taken from the com mon mother-tongue.

Two languages separated from the common mother-tongue, subsequent to the evolution of (say) a form for the dative case, are more allied than two languages similarly separated anterior to such an evolution.

Subject to one exception. This means, that it is possible that two languages may appear under certain circumstances more allied than they really are, and vice versa.

They may appear more allied than they really are, when, after separating from the common mother-tongue during the ante-inflexionary stage, they develop their inflexions out of the same principle, although independently. This case is more possible than proved.

They may appear less allied than they really are, when, although separated from the common mother-tongue after the evolution of a considerable amount of inflexion, each taking with it those inflexions, the one may retain them, whilst the other loses them in toto. This case also is more possible than proved.

VOL. XLVII. NO. XCIV.-OCTOBER 1849.

X

Each of these cases involves a complex question in philology: the one the phenomena connected with the rate of change; the other the uniformity of independent processes.

These questions are likely to affect future researches more than they had affected the researches hitherto established. Another question has affected the researches hitherto established more than it is likely to affect future ones. This is the question as to the fundamental unity, or non-unity of language. Upon this the present writer has expressed an opinion elsewhere. At present he suggests that the more the general unity of the human language is admitted, the clearer will be the way for those who work at the details of the different affiliations. As long as it is an open question, whether one class of languages is wholly unconnected with others, any connection engenders an inclination to arrange it under the group previously recognised. I believe that this determined the position of the Celtic in the Indo-European group. I have great doubts whether if some affinity had been recognised from the beginning, it would even have stood where it now does. The question, when Dr Prichard undertook his investigations, was not so much whether the Celtic was in the exact ratio to any or all of the then recognised European languages in which they were to each other, but whether it was in any relation at all. This being proved, it fell into the class at

once.

The present writer believes that the Celtic tongues were separated from their mother-tongue at a comparatively early period of the second stage; i. e., when but few inflexions had been evolved; whilst the Classic, Gothic, Lithuano-Slavonic (Sarmatian), and Indo-Persian (Iranian) were separated at comparatively late periods of the same stage, i. e., when many inflexions had been evolved.

Hence he believes that, in order to admit the Celtic, the meaning of the term Indo-European was extended.

Regretting this (at the same time admitting that the Celtic tongue is more Indo-European than any other), he believes that it is too late to go back to the older and more restricted use of the term; and suggests (as the next best

change), the propriety of considering the Indo-European class as divided into two divisions, the older containing the Celtic, the newer containing the Iranian, Classical, Sarmatian, and Gothic tongues. All further extensions of the term he believes to be prejudicial to future philology; believing also that all supposed additions to the Indo-European class have (with the exception, perhaps, of the Armenian) involved such farther extension.

Note.-After the statement of the preceding remarks, Mr Kingsford suggested the possibility of languages becoming wholly uninflexional, and, as such, reduced to a condition like the languages of the first period, in which case they might (as in a cycle) undergo a second series of similar developments de novo; and so on, ad infinitum. This the present writer believed to be a philological possibility; indeed, in his Inaugural lecture at University College, he had expressed a similar notion.

Note. Since this was written, a heavy loss has fallen upon the learned world, in the death of Dr Prichard. This induces me to insist more strongly than I should otherwise have done, upon the exception taken to his position of the Celtic being more verbal than real. High as I put his work upon the Physical History of Mankind (especially as it appeared in the first edition, where, though less learned, it was more critical, more original, and more in advance of contemporary thinkers, than in its final form), I put his Eastern Origin of the Celtic Nations equally high; and, as a definite addition to ethnographical philology, even higher.

On the Fall of Rivers, especially that of the Jordan, in Palestine; the Thames, Tweed, Clyde, and Dee, in Britain; and the Shannon, in Ireland.

Mr Augustus Petermann, the geographer, in an interesting paper, read to the Geographical Society of London, "On the Depression of the Dead Sea, and the Fall of the Jordan," com

municates the following observations on the fall of rivers in general, and on that of the Jordan, in Palestine, of the Thames, Tweed, Clyde, and Dee, in Britain, and the Shannon, in Ireland.

The data regarding the British rivers, Mr Petermann informs us, 66 are a part of the results of laborious researches into the physical and statistic geography of the British Isles, which he hopes ere long to lay before the public in a series of maps.

1. The Fall of Rivers. This interesting branch of physical geography presents, in its comparative results, such striking anomalies as perhaps have never before been anticipated. There are rivers in this country that are of the same aggregate size and descent, and the one forms in its course a series of considerable waterfalls, the heights of some of which approach nearly to 100 feet; whilst the other, equivalent, as before mentioned, in size and fall, presents not even a single waterfall or cataract, nay, not even one decided rapid.

To ascertain the rate of fall, there are two points, the accurate knowledge of which is necessary,-1st, The length of the river in question; 2d, Its elevation.

The length of the river, that is to say, the development of its course, is greatly influenced by the extent of its windings. These windings, in their true extent, can only be delineated with sufficient accuracy in maps of a very large scale, such as is adapted for national maps, as the windings which rivers generally exhibit disappear in reduced maps to such a degree that a great difference is produced in the calculations.

For the purpose of forming a judgment of the rate of the decrease of a river's course by constructing different maps in decreasing scales, I made the following inquiry regarding a river which is not uncommonly meandering in its course. here allude to a portion of the Severn, from its source to Shrewsbury, and I found the following numerical results:

I

On the Ordnance or National Map 1 mile to 1 inch 81.8

The first, the Ordnance Map, forms the basis of the other three. By comparing the Index Map with the Ordnance Map itself, the scale of which is ten times larger, the figures shew that the windings of the river in the Index Map disappear to such a degree as to give a decrease of 13.3 miles for a length of 81.8 miles, or 16.3 per cent.; and so a map on a scale 25 times smaller gives a decrease of 23.6 per cent., and a map on a scale 42 times smaller 29.1 per cent.,—that is, measuring off the length of a river in a map of a scale of 42 miles to 1 inch, the results, however carefully measured, are nearly one-third too small.

Thus with regard to the Jordan, although not a meandering river, and forming almost a straight line from the Lake of Tiberias to the Dead Sea, yet the few bendings and windings of the river, when taken into account, give it a length of 80 English miles,* which is 11 miles longer than Professor Robinson's statement of 60 geographical miles. The 14-3 feet of fall per mile which he thus calculates, will be reduced by the 80 English miles to 12-3 feet. And I have no doubt that when the course of the Jordan is thoroughly explored, it will exhibit a still greater development, and the rate of fall will be still more reduced.†

'The fall of a river influences in part the velocity or force

* According to Robinson's Map, in his Biblical Researches.

The Jordan, as appears from Lieut. Molyneux's account, is extremely winding in its course; the American expedition under Captain Lynch found that it serpentines 200 miles, and the fact confirms Mr Petermann's argument.-Editor of Journal of Geographical Society.