In Italy the _Inflexible_ system (which has met in France with the fate we have just seen) obtained temporary favor, and was adopted in the _Duilio_ and the _Dandolo_, two very large ships, of 11,000 tons each, of a speed exceeding fifteen knots, and each carrying four 100-ton guns in turrets. Although these ships are 340 feet in length, even the armored belt amidships (if “belt” in any sense so short a strip of armor may be called[22]) is but 107 feet long, leaving therefore 233 feet of the ship at the ends wholly devoid of water-line protection. As the author of the “citadel system,” I cannot regard such an arrangement as this as a fair and reasonable embodiment of it, the discrepancy between the armored and unarmored portions being greater in these two ships than even in the _Ajax_ and _Agamemnon_, which are perhaps the worst examples of the abuse of the citadel system in the British navy. It is to the credit of the Italian government that ships of this type were not repeated in their navy; and it is but right to point out that there were excuses (which probably ranked in the minds of the designers as _reasons_) for a more extreme proportionate limitation of the citadels being adopted in the _Duilio_ and _Dandolo_ than in the _Ajax_ and _Agamemnon_. Among these were the possession by the Italian ships of heavier armaments, and of far greater steam-power and speed than the British ships possessed—a matter to which further reference will be made hereafter—and probably, also, the adoption of somewhat finer water-lines as a means of attaining the superior speed. In this connection it may be well to observe that the question of leaving so-called armored line-of-battle ships without armor at the extremities is first one of principle, and afterwards one of degree. The principle (which should be observed in the design of every armored vessel which is intended for the line of battle, or for those close and severe contests of ship with ship which will probably supersede in a great degree the system of fighting in lines of battle) is this: the proportion which the armored citadel bears to the unarmored ends must always be such as to enable the ship to keep afloat all the time the armor itself holds out against the attack of the enemy; so that injuries to the unarmored ends, however great or multiplied, shall not alone suffice to destroy the ship. Whatever may occur in the future to interfere with the application of this principle—and I do not deny that such interferences may arise under certain perfectly conceivable circumstances—nothing has yet happened to justify its abandonment, or to even justify the remotest chance of its being violated.