The fact also, on which I then strongly insisted, that the Uraon Koles, and Rajmahal Koles, might be Koles in blood, but certainly not in language, their language being, like that of the Gonds, Dravidian, is now no longer disputed. But beyond this, all is still as hypothetical as it was twenty years ago, simply because we can get no grammars of the Munda dialects. Why do not the German missionaries at Ranchi, who have done such excellent work among the Koles, publish a grammatical analysis of that interesting cluster of dialects? Only a week ago, one of them, Mr. Jellinghaus, gave me a grammatical sketch of the Mundári language, and even this, short as it is, was quite sufficient to show that the supposed relationship between the Munda dialects and the Khasia language, of which we have a grammar, is untenable. The similarities pointed out by Mason between the Munda dialects and the Talaing of Pegu, are certainly startling, but equally startling are the divergences; and here again no real result will be obtained without a comparison of the grammatical structure of the two languages. The other classes of Indian languages, the Taic, the Gangetic, subdivided into Trans-Himalayan and Sub-Himalayan, the Lohitic, and Tamulic, are still retained, though some of their names have been changed. Without wishing to defend the names which I had chosen for these classes, I must say that I look upon the constant introduction of new technical terms as an unmixed evil. Every classificatory term is imperfect. Aryan, Semitic, Hamitic, Turanian, all are imperfect, but, if they are but rightly defined, they can do no harm, whereas a new term, however superior at first sight, always makes confusion worse confounded. The chemists do not hesitate to call sugar an acid rather than part with an old established term; why should not we in the science of language follow their good example?