We will now imagine that we have before us two designs (or more if you wish), which seem equally preferable from all the considerations of context. We must therefore overcome the dead heat by assessing the relative values of their mechanical contents. And I suggest that we might systematically do this by comparing them in the light of some general principles.

The first of these is that a determinate design is preferable to an indeterminate one.

All mechanisms are excellent mathematicians. They scrupulously behave according to the principles of their construction. And the more accurate our mathematics about them is worked out, the better we shall be able to forecast how they will work when made. If the calculations are indeterminate the forecast will be indefinite. We shall only know exactly what will happen when the machine is made, and it may then be too late. The evolution of aeroplane design illustrates this. In early designs things fell off. To stop this the next generation of planes used two or more members to do any one job, feeling there was safety in numbers. This meant a host of redundant situations occurred. As creep and fatigue are not respecters of redundances things still fell off, the only difference being that you could no longer work out in advance the order in which they did it. Both the maths and the future were indeterminate.