+31.+ Why is it then that the Romans conquer? And what is it that brings disaster on those who employ the phalanx? Why, just because war is full of uncertainties both as to time and place; whereas there is but one time and one kind of ground in which a phalanx can fully work. If, then, there were anything to compel the enemy to accommodate himself to the time and place of the phalanx, when about to fight a general engagement, it would be but natural to expect that those who employed the phalanx would always carry off the victory. But if the enemy finds it possible, and even easy, to avoid its attack, what becomes of its formidable character? Again, no one denies that for its employment it is indispensable to have a country flat, bare, and without such impediments as ditches, cavities, depressions, steep banks, or beds of rivers: for all such obstacles are sufficient to hinder and dislocate this particular formation. And that it is, I may say, impossible, or at any rate exceedingly rare to find a piece of country of twenty stades, or sometimes of even greater extent, without any such obstacles, every one will also admit. However, let us suppose that such a district has been found. If the enemy decline to come down into it, but traverse the country sacking the towns and territories of the allies, what use will the phalanx be? For if it remains on the ground suited to itself, it will not only fail to benefit its friends, but will be incapable even of preserving itself; for the carriage of provisions will be easily stopped by the enemy, seeing that they are in undisputed possession of the country: while if it quits its proper ground, from the wish to strike a blow, it will be an easy prey to the enemy. Nay, if a general does descend into the plain, and yet does not risk his whole army upon one charge of the phalanx or upon one chance, but manœuvres for a time to avoid coming to close quarters in the engagement, it is easy to learn what will be the result from what the Romans are now actually doing. +32.+ For no speculation is any longer required to test the accuracy of what I am now saying: that can be done by referring to accomplished facts. [Sidenote: Flexibility of the Roman order.]