Other combinations may be indicated. By adding a fourth column for those who resort to violence, the schematic possibilities are enlarged. Most theorists have been agitators in some measure, and it is often a matter of taste whether emphasis is to be placed on one or the other feature of their activity. Characteristically, the theorists have sought to appeal to the sentiments of their contemporaries by pamphleteering or by the direction of their speculative interests to immediate ends. Hobbes and Rousseau were shut off from oratory and organization, but no small part of their writing was intended to add fuel to the flames around them. Such men as Tom Paine are able to strike off excellent formulations of political theory in the heat of the fray. So far we have been discussing types which are distinguished according to some nuclear relation among a few variables. The characteristic mode of elaborating such a type is to imagine a host of situations in which the type may be found, and to describe the resulting picture. This is the elaboration of the preliminary sketch on the basis of sophisticated experience with social life. For these impressionist methods it is possible to substitute a more formal procedure. Having chosen a central primary relation, is it possible to find, by reference to specific instances, the relative frequency with which other traits are associated with the nuclear ones. The starting-point may be either institutional or functional, of course: those who are “judges,” “legislators,” or “bosses” may be investigated, or those who are “statesmen,” “conciliators,” or “administrators.” We have already had occasion to stress the point that institutional definitions and functional definitions do not precisely coincide, although either is a valid point of departure for research.