As my lecture in Westminister Abbey was delivered shortly after the publication of my “Introduction to the Science of Religion,” I ventured to take certain points which I had fully treated there as generally known. One of them is the exact value to be ascribed to canonical books in a scientific treatment of religion. When Mr. Lyall observes _in limine_, that inferences as to the nature and tendency of various existing religions which are drawn from study and exegetic comparison of their scriptures, must be qualified by actual observation of these religions and their popular form and working effects, he expresses an opinion which I hold as strongly as he holds it himself. After enumerating the books which are recognized as sacred or authoritative by large religious communities in India, books of such bulk and such difficulty that it seems almost impossible for any single scholar to master them in their entirety, I added (p. 111), “And even then our eyes would not have reached many of the sacred recesses in which the Hindu mind has taken refuge, either to meditate on the great problems of life, or to free itself from the temptations and fetters of worldly existence by penances and mortifications of the most exquisite cruelty. India has always been teeming with religious sects, and its religious life has been broken up into countless local centres which it required all the ingenuity and perseverance of a priestly caste to hold together with a semblance of dogmatic uniformity.” We must take care, however, in all scientific studies, not to render a task impossible by attaching to it conditions which, humanly speaking, cannot be fulfilled. It is desirable, no doubt, to study some of the local varieties of faith and worship in every religion, but it is impossible to do this with anything like completeness. Were we to wait till we had examined every Christian sect before trusting ourselves to form a general judgment of Christianity, not one of us could honestly say that he knew his own religion. It seems to me that in studying religions we must expect to meet with the same difficulties which we have to encounter in the comparative study of languages. It may, no doubt, be argued with great force that no one knows English who is ignorant of the spoken dialects, of the jargon of sailors and miners, or of the slang of public-houses and prisons. It is perfectly true that what we call the literary and classical language is never the really living language of a people, and that a foreigner may know Shakespeare, Milton, and Byron, and yet fail to understand, if not the debates in Parliament, at all events the wrangling of sellers and buyers in the markets of the city. Nevertheless, when we learn English, or German, or French, or any of the dead languages, such as Latin and Greek, we must depend on grammars, which grammars are founded on a few classical writers; and when we speak of these languages in general, when we