We may admit all that is stated by those who admire its epics, or elucidate its philosophy; we may admire all this and much more besides, but we shall still miss the great elements of oratory and history, that connect the ancient languages of Greece and Italy with the thoughts, and feelings, and admiration of recent Europe. The same sort of reasoning applies to the Semitic languages. One element they have, in their grammatical representation, which gives them a value in philological philosophy, in the abstract, above all other languages--the _generality_ of the expression of their structure. This is _symbolic_, and its advantage is that it exhibits the naturally universal phenomena of their construction in a universal language. Yet neither this nor their historical value raises them to the level of the classical languages. Now, what has just been written has been written with a view towards a special inference, and as the preliminary to a practical deduction; and it would not have been written but for some such ulterior application. If these languages have so high a disciplinal value, how necessary it is that the expression of their philological phenomena should be accurate, scientific, and representative of their true growth and form? How essential that their grammars should exhibit nothing that may hereafter be unlearned? _Pace grammaticorum dixerim_, this is not the case. Bad {xii} as is Lindley Murray in English, Busby and Lilly are worse in Greek and Latin. This is the comparison of the men on the low rounds of the ladder. What do we find as we ascend? Is the grammatical science of even men like Mathiae and Zump _much_ above that of Wallis? Does Buttmann's Greek give so little to be unlearned as Grimm's German? By any one who has gone far in comparative philology, the answer will be given in the negative. This is not written in the spirit of a destructive criticism. If an opinion as to the fact is stated without reserve, it is accompanied by an explanation, and (partially, perhaps) by a justification. It is the business of a Greek and Latin grammarian to teach Greek and Latin _cito, tute, ac jucunde_,--_cito_, that is, between the years of twelve and twenty-four; _tute_, that is, in a way that quantities may be read truly, and hard passages translated accurately; _jucunde_, that is, as the taste and memory of the pupil may determine. With this view the grammar must be _artificial_. Granted. But then it should profess to be so. It should profess to address the memory only, not the understanding. Above all it should prefer to leave a point untaught, than to teach it in a way that must be unlearned.