ive imposition of such liability was arbitrary and
unreasonable in itself and hence unconstitutional. On that theory,
the New York Court of Appeals held workmen's compensation
unconstitutional, and a minority of the Supreme Court of the United
States recently announced the same proposition.

Because of its implications for constitutional law, in view of the
increasing frequency of legislation imposing responsibility at one's
peril in certain enterprises, in the case of certain dangerous
agencies and in situations where it is felt that the loss should be
borne by all of us rather than by the luckless individual who chances
to be hurt, the basis of tort liability has become a question of
moment beyond the immediate law of torts. It is a practical question
of the first importance, as well as a theoretical question of
interest, whether we are to generalize our whole system of tort
liability by means of one principle of liability for fault and for
fault only, as the French sought to do and as we later sought to do
largely under their influence, or, on the other hand, are to admit
another source of delictal liability alongside of fault, as the French
law does in fact and is coming to do in theory, and as our law has
always done in fact. For in our law as it stands one may perceive
readily three types of delictual liability: (1) Liability for
intentional harm, (2) liability for unintentional culpable harm, (3)
liability in certain cases for unintended non-culpable harm. The first
two comport with the doctrine of no liability without fault. The third
cannot be fitted thereto. We must either brand cases of the third type
as historical anomalies, of which we are gradually to rid ourselves,
or else revise our notions of tort liability. Let us remember that the
nineteenth century was well advanced before we understood the subject
of negligence and that before we had convinced ourselves that no
liability without fault was orthodox common law, the highest court of
England had given absolute liability a new field by the decision in
_Rylands_ v. _Fletcher_. We are not questioning a long-established
dogma in Anglo-American administration of justice, therefore, when we
ask whether the orthodox theory of the last generation is adequate as
an analytical statement of the law that is, or as a philosophical
theory of the law that ought to be. My own belief is that it is
neither.

Suppose that instead of beginning with the individual free wi