The penal agricultural colony, in lands which need clearing, is the best for adults, passing from the least to the most healthy according to the categories of criminals--born, habitual, occasional--and according to the gravity of the crimes committed. To this may be added, for convicts less capable of restoration to social life, labour in mines, especially when the mines are State property. What I have said of malaria I say of fire-damp: it is much better that these should kill off criminals, than honest workmen.

The penal agricultural colony in lands already cultivated is best for children and young people.

This is the ideal and the typical form of segregation for criminals, against whom it would not be sufficient to exact strict reparation of damage, on the principles already set forth.

Wherever there is a crowding of humanity, there is human fermentation and putrefaction. Only labour in the open air will secure physical and moral health. And if agricultural work would be less fitted for criminals from the towns, there is no reason why an agricultural colony should not make itself as far as possible self-sufficing by means of workshops where prisoners could ply the trade to which they were accustomed when at liberty. For town convicts without a trade, such as vagabonds, beggars, and the like, on the ground of their muscular incapacity for hard and regular work, an agricultural colony is still the most fit, for it provides light and varied occupations, as the agricultural colonies of Holland, Belgium, and Austria bear witness.

The same evolution will take place in regard to the segregation of criminals as in regard to the seclusion of the insane; first, hospitals and prisons, with a terrible communion of corruption in both cases; then barrack life, in asylums or penitentiaries, vast and isolated; lastly, for the insane, a system of so-called village asylums, and even a free colony for harmless idiots who can be put to agricultural work and minor trades, as at Gheel in Belgium. Similarly for criminals, the sanitary ``elbow room'' of agricultural colonies will be substituted for the infectious barrack-life of the great prisons.

As for habitual criminals, their anthropological characteristics remind us that we must distinguish between the two crises of their criminal activity, and, as a consequence, between the methods of defence against them. That is to say, we must distinguish between the initial moment at which they commit their first crime and the subsequent period in which they become habitual offenders, recidivists, and even incorrigible.

Thus it is clear that at the initial moment of their criminal career they ought to be subjected to the measures which I am about to indicate for occasional criminals; whereas, when from occasional they have become, partly by their imprisonment, habitual offenders, they must be subjected to the measures already indicated for born criminals. The latter are incorrigible through congenital tendency to degenerate, and the former are incorrigible through acquired tendency; but they end in the same degree of anti-sociality and brutalisation. There is, however, this difference, that habitual offenders nearly always commit less serious crimes, such as theft, swindling, forgery, indecent assault, whilst the born criminals, though they may be petty thieves, or not very formidable swindlers, are more frequently murderers, footpads, guilty of arson, or the like. Thus the discipline of their segregation must vary accordingly.

For occasional criminals, social defence must have a character of prevention rather than of repression, so as to save them from being driven, by a mistaken prison organisation, to become recidivists, and therefore habitual and incorrigible criminals.