general, corrupted by a large bribe, did reveal the secret and shortly afterwards, when entering a church, was consumed by fire which fell from heaven upon him. The story is obviously legendary. The venal general is as unreal as the fire from heaven; he is merely introduced to us as “an awful example,” and we cannot endow him with reality by rejecting the fire. The claim of the Marquess Carabbas to reality is not established by denying the existence of Puss-in-Boots. Had the secret been divulged the sea-fire would have been used against the Greeks, and no mixture that can be identified with it ever was. A saltpetre mixture, then, would not, in all probability, have fulfilled the first condition, nor would it have fulfilled the second. There is no conceivable connection between saltpetre and the sea, or water in general. A saltpetre mixture (of suitable proportions) would have proved a much better incendiary than Greek fire, but it would have acted as effectively from a fort as from a ship. Indeed, if we consider the ill effect of the moist sea air on the impure saltpetre of early times, we are justified in saying that the action of such mixtures on land would have been better, in general, than at sea. A saltpetre mixture would have fulfilled the third condition by burning with much noise and smoke, which we may suppose to be the essential meaning of the Emperor Leo’s phrase, “thunder and smoke.”[70] We cannot reasonably attach greater significance to one of the commonest of all metaphors, thunder, which has been applied times unnumbered to the human voice, to the bursting of a child’s cracker,[71] and to the whirring of a dart. “Never burst such peals from the thunder-cloud,” says Vergil, as were produced by the javelin thrown by Æneas.[72] As regards the fourth condition, the above statement of the Emperor Constantine about sea-fire and siphons[73] completely justifies us in concluding that there was some necessary connection between the two things. Now, there was no necessary connection between saltpetre mixtures, even when explosive, and siphons. Small quantities of such mixtures could have been, and eventually were, thrown by hand, in grenades, like Greek fire. Saltpetre mixtures, therefore, would not have fulfilled the fourth condition. The result of the foregoing inquiry is, that a saltpetre mixture would have only fulfilled one, the third, of the four conditions to which the sea-fire was subject; and we have now to cast about for some mixture of known substances, not hitherto combined together for warlike purposes, which would have fulfilled them all.