; but accidental homicide may cause no alarm,
whereas the intentional and malignant homicide may cause any quantity of
alarm and shock to the general sense of security. In this way,
therefore, the legislator has again indirectly to take into account the
moral quality which is itself dependent upon utility.

I must, however, pass lightly over a very clear and interesting
discussion to reach a further point of primary importance to the
Utilitarian theory, as to the distinction between the moral and legal
spheres.[407] Bentham has now 'made an analysis of evil.' He has, that
is, classified the mischiefs produced by conduct, measured simply by
their effect upon pleasures or pains, independently of any consideration
as to virtue and vice. The next problem is: what conduct should be
criminal?--a subject which is virtually discussed in two chapters (xv.
and xix.) 'on cases unmeet for punishment' and on 'the limits between
Private Ethics and the act of legislation.' We must, of course, follow
the one clue to the labyrinth. We must count all the 'lots' of pain and
pleasure indifferently. It is clear, on the one hand, that the pains
suffered by criminals are far less than the pains which would be
suffered were no such sanctions applied. On the other hand, all
punishment is an evil, because punishment means pain, and it is
therefore only to be inflicted when it excludes greater pain. It must,
therefore, not be inflicted when it is 'groundless,' 'inefficacious,'
'unprofitable,' or 'needless.' 'Needless' includes all the cases in
which the end may be attained 'as effectually at a cheaper rate.'[408]
This applies to all 'dissemination of pernicious principles'; for in
this case reason and not force is the appropriate remedy. The sword
inflicts more pain, and is less efficient than the pen. The argument
raises the wider question, What are the true limits of legislative
interference? Bentham, in his last chapter, endeavours to answer this
problem. 'Private ethics,' he says, and 'legislation' aim at the same
end, namely, happiness, and the 'acts with which they are conversant are
_in great measure_ the same.' Why, then, should they have different
spheres? Simply because the acts 'are not _perfectly and throughout_ the
same.'[409] How, then, are we to draw the line? By following the
invariable clue of 'utility.' We simply have to apply an analysis to
determine the cases in which punishment does more harm than good. He
insists especially