Attempts have been made to discover (what has been assumed to exist) a form of words termed a test of causal relation, "which, put into the hands of jurors, can be used by them as a sort of legal yardstick to measure the evidence" and to determine, instantly and with mathematical exactness, whether or not the defendant's tort caused the plaintiff's damage. But a test possessing these qualities (a test with such potentiality) has never been found; "not because those who have searched for it have not been able and diligent, but because it does not exist."45

It is often difficult to determine whether the defendant's tort was the cause of the plaintiff's damage; but this difficulty arises "from the nature of the facts to be investigated." It is a practical difficulty "to be solved by the jury," and not a legal difficulty for the court.46

in such a predicament. As to the subject of "attempt" in criminal law, high authorities declare it impossible to give an exact definition or generalization in regard to the socalled "proximateness" of an act.

Sir J. F. Stephen in 2 History of Criminal Law of England, 224, says: "The law as to what amounts to an attempt is of necessity vague. It has been said in various forms that the act must be closely connected with the actual commission of the offense, but no distinct line upon the subject has been or as I should suppose can be drawn." The same author, in his Digest of Criminal Law, 1 Am. ed., Art. 49, after defining attempt as "an act done with intent," etc., "and forming part of a series of acts which," etc., says: "The point at which such a series of acts begins cannot be defined; but depends upon the circumstances of each particular case."

So Dr. Bishop in 1 New Criminal Law, § 759, paragraph 1, says: "An attempt may be too small a thing, or proceed not near enough to its accomplishment, for the law to notice. How great it must be, and how far progress, is matter not reducible to exact rule." See also 1 Bishop, New Criminal Law, § 762, paragraph 4; and compare Holmes, J., in Commonwealth v. Kennedy, 170 Mass. 18, 22, 48 N. E. 770, 771 (1897). And Dr. Kenny in Outlines of Criminal Law, Webb's Am. ed., 73, says: "It seems impossible to lay down any abstract test for determining whether an act is sufficiently proximate to be an 'attempt.""

For another instance of admitted vagueness, see Professor Maitland's comment on his own definition of "trust." He said: "It is a wide vague definition, but the best that I can make." Maitland, Lectures on Equity, 44.

Sir J. F. Stephen said of a certain statutory definition: "I do not think it happy, as it attempts to define what is essentially indefinite." 3 Stephen, History of Criminal Law, 316.

45 This phraseology is largely borrowed from Dr. Bishop's discussion as to "Test of Insanity," I Bishop, New Criminal Law, § 381.

46 See dissenting opinion of Doe, J., upon another topic, in State v. Pike, 49 N. H. 399, 438 (1870). See also New York, etc. R. v. Estill, 147 U. S. 591, 613, 13 Sup. Ct. 444, 453 (1892). The difficulty lies "in ascertaining the facts, and not in applying law to them." 3 Stephen, History of Criminal Law, 6.

"The question whether an item of loss is or is not a proximate consequence of the wrong is in each case a question of fact. Only general principles can be laid down, and in applying them much latitude must necessarily be left to the court and jury. If the case is a clear one, the court will direct the jury upon the question; but if the question is a doubtful one it will be left to the jury." (After giving instances where certain consequences were held proximate and others remote.) "An entirely harmonious course of decision on such a question is not expected. As the determination is really one of fact, under proper directions, and ordinarily for the jury, the decision may simply be the result of the court's upholding the right of the jury to decide one way or the other; and even if the court itself determine the question, as is not infrequent in practice, it is nevertheless natural to expect differences of opinion upon what are really close questions of fact."47

It may be urged in favor of laying down minute legal tests of causation that, in the absence of such tests, a defendant cannot foretell in advance the extent of the liability he is incurring. But what principle of justice requires the law to furnish him such information ?48 As to the standard of conduct, e. g. as to what conduct shall be deemed negligent, there is more room for the argument that a defendant ought to know what actions the law forbids, and thus be able to keep within legal bounds. But here, ex hypothesi, the defendant has been guilty of wrongful conduct; he has exceeded the legal limit, and the only remaining question is as to the extent of his liability, i. e. as to the existence of causal relation between the defendant's fault and the plaintiff's damage.49

Suppose that, in establishing a general rule upon this subject of causation in actions of tort, we are confined to a choice between two tests: First, the rule of liability for probable consequences only; second, the "General Rule" we have suggested ("defendant's

47 1 Sedgwick, Damages, 9 ed., §§ 116, 117.

48 See ante, 25 HARV. L. REV. 248-249; reply to argument founded on hardship. 49 France v. Gaudet, L. R. 6 Q. B. 199 (1871), was an action for conversion. At the time of the conversion, a special value was attached by special circumstances to the goods converted. Although these special circumstances were not known to the defendant, he was nevertheless held liable for the full actual value fixed by these circumstances. Mellor, J., said, on page 205, ". . . no notice to the wrongdoer could then affect the value, although it might affect his conduct; but upon what principle is a notice necessary to a man who ex hypothesi is a wrongdoer?"

tort must have been a substantial factor in subjecting plaintiff to the damage complained of").

If either of these tests is adopted as an exclusive general rule to be applied in all cases, it is likely that practical injustice will occasionally result. Under the first test recovery may sometimes be unjustly denied. Under the second test, recovery may sometimes be unjustly permitted. If injustice will appreciably result more frequently under one of these tests than under the other, then the test under which this happens should be discarded. Neither logic, nor legal symmetry, furnish conclusive reasons for adopting a rule of law. The decisive consideration is that furnished by experience the practical working of a rule.50

Our own impression is that practical injustice would result more frequently from the operation of the first test than from the second. Hence we should reject the first test; and, if we are compelled to choose one of the two, we should adopt the second test.

Suppose, however, that there is no difference between the two tests as to practical injustice. Suppose that the number of unjust results from the application of one test will be exactly equal to the number under the other test. Under the one test, there will be the risk that a recovery may occasionally be unjustly denied to the plaintiff. Under the other test, there will be the risk that, in an equal number of instances, a recovery may be unjustly allowed against the defendant. Whichever test then is adopted, one party or the other has got to run a risk of occasional injustice. Upon which of the two parties ought this risk to be imposed: Upon the innocent plaintiff or the tortious defendant? In the words of Judge Christiancy in Allison v. Chandler: 51

"... does not sound policy require that the risk should be thrown upon the wrongdoer instead of the injured party?" "The nature of the case is such as the wrongdoer has chosen to make it; and, upon every principle of justice, he is the party who should be made to sustain all the risk of loss which may arise from the uncertainty pertaining to the

50 "But as the law is a practical science, having to do with the affairs of life, any rule is unwise if, in its general application, it will not, as a usual result, serve the purposes of justice." Allen, J., in Spade v. Lynn & Boston R. Co., 168 Mass. 285, 288, 47 N. E. 88, 89 (1897). "The life of the law has not been logic: it has been experience." Holmes, The Common Law, 1. See 11 HARV. L. REV. 439; 14 HARV.

L. REV. 195.

51 II Mich. 542, 553-556 (1863).

nature of the case, and the difficulty of accurately estimating the results of his own wrongful act. Upon what principle of right can courts of justice assume not simply to divide this risk, which would be thus far unjust but to relieve the wrongdoer from it entirely, and throw the whole upon the innocent and injured party ?" 52

CAMBRIDGE, MASS.

Jeremiah Smith.

APPENDIX.

There are two classes of cases which have occasioned great difficulty. It has sometimes been supposed that, if it is laid down as a general rule that the improbability of a result does not per se exonerate a wrongdoer, then these cases must constitute exceptions to such a rule. In the foregoing general discussion very little consideration has been given to these classes. It is proposed now to deal with them more fully.

These classes of cases are:

1. Where, although defendant was in fault, yet no damage would have resulted but for an occurrence in the natural world, which constituted an extraordinary departure from the usual course of nature.

2. Where defendant's conduct was wrongful, but no harm would have resulted had it not been for the unforeseeable intervention of an independent wrongdoer.

As to 1, there is a remarkable conflict of authority. Some courts would order a verdict for the defendant, and others for the plaintiff.

As to 2, the weight of authority is in favor of exonerating the defendant; but it may be found upon examination that some cases of this description present a question for the jury.

As to the first class, the arguments pro and con are brought out, and authorities are collected, in the conflicting decisions in two recent cases; namely, Green-Wheeler

52 The frequent references to Professor Terry's book, Leading Principles of AngloAmerican Law, and to Professor Bohlen's articles in legal periodicals, hardly represent the full extent of the present writer's obligations to these learned authors; neither of whom, of course, can be held responsible for the ultimate conclusions here reached.

Professor Terry's book, published in 1884, though cited by such jurists as Sir Frederick Pollock and Professor Wigmore, is not often referred to in the reports either by counsel or judges. The work contains a good deal of valuable matter not to be found elsewhere. The learned author never evades a difficulty. He seldom fails to state the crucial issues upon each topic discussed; though he sometimes frankly confesses his inability to arrive at a satisfactory solution. A reader can derive great benefit from this book, even if he does not agree with all the author's theories or adopt all his phraseology.

Professor Bohlen's essays on various legal subjects, in the American Law Register, the University of Pennsylvania Law Review, and the HARVARD LAW REVIEW, are of great value. He lets in light upon every subject which he discusses. It is hoped that a volume may soon be published containing a full collection of all Professor Bohlen's articles in legal periodicals.

Shoe Co. v. Chicago, R. I. & Pac. R. Co., 130 Ia. 123, 106 N. W. 498 (1906), and Rodgers v. Missouri Pac. Ry. Co., 75 Kan. 222, 88 Pac. 885 (1907).

Precise question presented by the Green-Wheeler case: A common carrier negligently delays goods in transitu. While he is so delaying and the goods are still in his custody, the goods are destroyed by a flood, which was greater than ever before known and could not have been reasonably anticipated (in legal phraseology," an act of God”). If the goods had been promptly carried to their destination, they would have escaped the flood. Is the carrier liable for the value of the goods?

The Rodgers case differs from the Green-Wheeler case as to one matter of fact. In the Rodgers case the goods were no longer in transit when overtaken by the flood. They were at the railroad terminus, but had not been delivered to the consignee. But for the original delay in starting the goods on their transit, the goods would have reached the terminus so early that they would have been delivered to the consignee before the flood, and would thus have escaped destruction.

On the above general question there is a remarkable conflict of authority. We are inclined to agree with the Iowa result, and to dissent from the Kansas result. The carrier is a wrongdoer, the only wrongdoer in the chain of antecedents. And, but for his wrong, the damage would not have happened. Both these circumstances combined do not, as matter of law, make out that his tort was the legal cause, or one of the concurring causes. But these two circumstances are competent to be considered on the question of fact, and go far to justify a finding of causal relation. It would seem that a jury might reasonably find that the effect of defendant's negligence appreciably continued down to the time of destruction, and that it constituted a substantial factor in subjecting plaintiff to the loss. It may be urged that "the carrier's delay did not produce the flood"; but this is true of the usual operations of nature, where a defendant is liable if his tort concurred therewith in bringing about damage. The workings of nature constitute the surroundings for all human acts; no man is excused simply because he did not create natural events, such as the blowing of the wind, or the flowing of the tide, in the Romney Marsh case, L. R. 5 Exch. 204 (1870); or the coldness of the atmosphere in Harrison v. Berkley, 1 Strob. L. (S. C.) 525 (1847); or in Fox v. B. & M. R. Co., 148 Mass. 220, 19 N. E. 222 (1889). See 1 Beven, Negligence, 3 ed., 80.

In the case now under consideration, while the defendant's negligent delay did not cause the flood, yet it did in fact subject plaintiff's property to the operation of the flood.

"This consideration that the peril of accidental destruction is enhanced by the negligent extension of time during which the goods must remain in the carrier's control and out of the control of the owner, and during which some casualty may overtake them, has not, we think, been given sufficient consideration in the cases in which the carrier has been held not responsible for a loss for which he is not primarily liable, but which has overtaken the goods as a consequence of the preceding delay in their transportation." McClain, C. J., in Green-Wheeler Shoe Company v. Chicago, R. I. & Pac. R. Co., 130 Ia. 123, 129, 106 N. W. 498, 500 (1906).

The right of a railroad company to transport explosives in its freight cars "does not include the right to subject persons along the route to dangers from explosions for a longer time. . . than is reasonably necessary to the performance of the carrier's duty.” . . . “If, therefore, the car was unnecessarily and unreasonably delayed at the place where it exploded, so as to subject plaintiff's property to such dangers for a longer time than would have attended a transportation made with reasonable dispatch, such keeping of the car at that place was a nuisance." Williams, J., in Fort Worth, etc. Ry. Co. v. Beauchamp, 95 Tex. 496, 500, 68 S. W. 502, 504 (1902). The danger