Such--and each class is capable of great expansion--are the cases where philology requires caution. Another matter now suggests itself.
To be valid a classification must be _real_; not _nominal_ or _verbal_--not a mere book-maker’s arrangement. Families must be in definite degrees of relationship. This, too, will bear illustration. A man wants a relation to leave his money to: he is an Englishman, and by relation means nothing more distant than a _third_ cousin. It is nothing to him if, in Scotland, a _fifth_ cousinship is recognised. He has not found the relation he wants; he has merely found a greater amount of latitude given to the term. Few oversights have done more harm than the neglect of this distinction. Twenty years ago the Sanskrit, Sclavonic, Greek-and-Latin, and Gothic languages formed a class. This class was called Indo-Germanic. Its western limits were in Germany; its eastern in Hindostan. The Celtic of Wales, Cornwall, Brittany, Ireland, Scotland, and the Isle of Man was not included in it. Neither was it included in any other group. It was anywhere or nowhere--in any degree of isolation. Dr. Prichard undertook to fix it. He did so--well and successfully. He showed that, so far from being isolated, it was connected with the Greek, German, and Sclavonic by a connexion with the Sanskrit, or (changing the expression) with the Sanskrit through the Sclavonic, German, and Greek--any or all. The mother-tongue from which all these broke was supposed to be in Asia. Dr. Prichard’s work was entitled the ‘Eastern Origin of the Celtic Nations.’ Did this make the Celtic Indo-Germanic? It was supposed to do so. Nay, more--it altered the name of the class; which was now called, as it has been since, Indo-European. Inconveniently. _A_ relationship was mistaken for _the_ relationship. The previous tongues were (say) second cousins. The Celtic was a fourth or fifth. What was the result? Not that a new second cousin was found, but that the family circle was enlarged.
What follows? Dr. Prichard’s fixation of the Celtic as a member of even the same _clan_ with the German, &c. was an addition to ethnographical philology that many inferior investigators strove to rival; and it came to be current belief--acted on if not avowed--that tongues as like the Celtic as the Celtic was to the German were Indo-European also. This bid fair to inundate the class--to make it prove too much--to render it no class at all. The Albanian, Basque, Etruscan, Lap, and others followed. The outlier of the group once created served as a nucleus for fresh accumulations. A strange language of Caucasus--the Irôn or Ossetic--was placed by Klaproth as Indo-Germanic; and that upon reasonable grounds, considering the unsettled state of criticism. Meanwhile, the Georgian, another tongue of those same mysterious mountains, wants placing. It has undoubted Ossetic--or Irôn--affinities. But the Ossetic--or Irôn--is Indo-European. So therefore is the Georgian. This is a great feat; since the Caucasian tongues and the Caucasian skulls now agree, both having their affinities with Europe--as they ought to have. But what if both the Irôn and Georgian are half Chinese, or Tibetan, _i. e._ are all but monosyllabic languages both in grammar and vocables? If such be the case, the term ‘Indo-European’ wants revising; and not only that--the principles on which terms are fixed and classes created want revising also. At the same time, the ‘Eastern Origin of the Celtic Nations’ contains the most definite addition to philology that the present century has produced; and the proper compliment to it is Mr. Garnett’s review of it in the ‘Quarterly;’ the first of a series of masterly and unsurpassed specimens of inductive philology applied to the investigation of the true nature of the inflexions of the Verb. But this is episodical.