It remains for us, before concluding our first chapter, to establish a fact of great scientific and practical value. This is that, after the anthropological classification which I have maintained for some ten years past, all who have been devoting themselves to the subject of crime as regarded from a biological and social standpoint have recognised the need for a classification less simple than that of habitual and occasional criminals, and which will be more or less complex according to the criterion which may be adopted. <p 48> In the first place, the necessity is generally recognised of abandoning the old arbitrary and algebraic type in favour of a classification which shall correspond more accurately with the facts of the case. This classification, originating in observations made within the prison walls, I have extended in the domain of criminal sociology, wherein it is now established as a fundamental criterion of legislative measures which must be taken as a protection against criminals, as well as a criterion of their responsibility. Secondly, the classifications of criminals hitherto given are not essentially and integrally distinct. It has been seen, as a matter of fact, that all the classifications which have been set forth amount to a recognition of four types, the born, the insane, the occasional criminals, and the criminals of passion; and this again resolves itself into the simple and primitive distinction between occasional and instinctive criminals. The category of criminals by contracted habit would not be accepted by all observers, but it corresponds too closely to our daily experience to stand in need of further proof. And on the other hand I must frankly decline to accept the authority of those who put forward classifications more or less symmetrical without having made a direct study of criminals; for the experimental method does not admit systems based on mere imagination, or on vague recollections of criminal trials, or on argumentative constructions built up from the systems of others. As a matter of fact, apart from the differences <p 49>of nomenclature, it is evident that the partial discrepancies in this anthropological classification of criminals are due in some measure to the different points of view taken by observers. For instance, the classification of Lacassagne, Joly, Krauss, Badik, and Marro rest upon a purely descriptive criterion of the organic or psychological characteristics of criminals. The classifications of Liszt, Medem, and Minzloff, on the other hand, depend solely upon the curative and defensive influence of punishment; and those of Foehring and Starke upon certain special points of view, such as the assistance of released prisoners, on their tendency to relapse.