The writer who asserts that the oldest work in any language is of such antiquity as to be separated from the next oldest by any very long interval--by an interval which leaves a wide chasm between the first and second specimens of the literature which no fragments and no traces of any lost compositions are found to fill up--makes an assertion which he is bound to support by evidence of the most cogent kind. For it is not always enough to shew that no intrinsic objections lie against the antiquity of the work in question. It may be so short, or so general in respect to its subject as to leave no room for contradictory and impossible sentences or expressions. It is not enough to shew that there were no reasons against such a literature being developed; since it is difficult to say what conditions absolutely forbid the production of a work stamped by no very definite characteristics. Nor yet will it suffice to say that the preservation of such a work is probable. All that can be got from all this is a presumption in its favour. The great fact of a work existing without giving this impulse to the production of others like it, and the fact of the same means of preservation being wholly neglected in other instances, still stand over. They are not conclusive against certain positions; but they are circumstances which must be fairly met; circumstances which if one writer overlook, others will not; circumstances which the critic will insist on; and circumstances which, if the dazzle of a paradox, or the appeal to the innate and universal sympathy for antiquity keep them in the background for a while, will, sooner or later, rise against the author who overlooked them. Neither are arguments from the antiquity of language conclusive. When two works differ from each other in respect to the signs of antiquity exhibited in their phraseology, the inference that the oldest in point of speech is _proportionably_ old in point of time is not the only one. It is an easy thing to say that in the Latin literature the language of Ennius represents a date a hundred years earlier than that of Cicero, and that of Cicero a date 400 earlier than the time of Boethius, and that when we meet elsewhere compositions which differ from each other as the Latin of Ennius does from that of Boethius, there is 500 years difference between them. It is by no means certain that any two languages alter at the same rate. But an average may be struck, and it may be said that greater antiquity of expression is _primâ facie_ evidence of a greater antiquity of date. It is: but is only so when we are quite sure that the _dialects_ of the two specimens are the same. There are works printed this very year in Iceland which, if their dates were unknown, would pass for being a hundred years older than the Swedish of the eleventh century.