Although this differentiation of types was based ostensibly upon displacement or tonnage, in reality it was formed on a more scientific basis. Admiral Sir George Elliot demonstrated, in 1867, that the real basis was not a rule of size, but a _law of safety_, similar to that which operates in the natural world; a law so important that it should under no circumstances be disregarded. He showed that sailing ships conformed to this law. He showed that the reduction of a vessel’s size, for instance, endowed her with smaller draught and an increased speed; that the dispensing with one quality automatically gave another in compensation; and that thus the weakly armed vessel always possessed the means, if not to fight, to escape from capture.[173] With the coming of steam and armour, all this was changed. Size had now no inherent disability; on the contrary, the larger the ship the greater the horse-power which could be carried in her, the greater her probable speed and sea endurance. The small ship had no advantages. The old classification had clearly broken down. The first ironclads, the _Warrior_ and her successors, although of frigate form, belonged to no particular class; they were of a special type intended to cope with the most powerful ships afloat or projected; and subsequent ships were designed with the same end in view. These ships being faster as well as more powerful than those of a smaller size, there was no object in attempting to build others of a frigate class for the purpose of outsailing them. As material developed, and as the warship became more and more obviously a compromise between conflicting qualities, differentiation of types was once more seen to be necessary. Attempts were made to classify on the bases of displacement, material, defensive and motive power, service, system of armament. In the end British construction divided itself into two categories: armoured and unarmoured vessels. And each of these categories was subdivided into classes of ships analogous to those of the old sailing ships.