that property may be destroyed by occurrences which would constitute an extraordinary departure from the usual course of nature is sometimes regarded as such an appreciable risk that it is insured against.

Somebody has got to bear the loss in this case. Should it be the innocent plaintiff, or the negligent defendant whose fault is potentially operative down to the time of loss?

That defendant could not have anticipated such a flood would be decisive in his favor, if plaintiff were relying on defendant's failure to take special precautions against the flood, as proof of his negligence. But here defendant's negligence is established on other grounds; namely, his delay; and the question is not whether his failure to anticipate the flood constitutes negligence, but whether his liability for the actual consequences of his admitted negligence should be restricted to such of those consequences as were foreseeable by him.

Suppose a carrier puts a covering over goods, sufficient to protect them from any rain that could reasonably be anticipated; and that, while he is carrying the goods with reasonable promptness, an entirely unprecedented rain occurs whereby the goods are ruined. The defendant is not liable for failing to provide a better covering. There is no tort on his part, no fault of his anywhere in the chain of antecedents. Hence, there is no question of causation. But how would it be, if the defendant was guilty of negligent delay in the carrying, and if such negligent delay exposed the goods to this extraordinary rain, which they would otherwise have escaped? Here the defendant is clearly in fault. The only question is one of causal relation. Is he to be exonerated from liability for the consequence of an admitted fault, because that consequence was not foreseeable by him?

Is there any limit to the enforcement of claims of this class? Can a consignee recover against a negligently delaying carrier for damage happening to goods a year after their delivery, upon the allegation that, but for the detention, the goods would have been sold and would not have been exposed to a cyclone twelve months later? Probably not. In general, a jury could not reasonably find that the effect of the delay appreciably continued so long and that the delay was a substantial factor in subjecting plaintiff to the loss. In cases not so extreme, the evidence may sometimes justify the submission of the question of fact to a jury; but no mathematical line can be drawn. The nearer the happening of the damage comes to the time of the defendant's delay, the more apt will the jury be to find a causative relation under our test. They are still more likely to find causative relation when the loss occurs during the delay and while the goods are still in the defendant's custody; but we do not regard either of these elements as legally essential to liability. (See, however, 1 Sedgwick, Damages, 9 ed., §§ 119a to 119e.)

Compare two decisions of the Massachusetts court: Denny v. New York Central R. Co., 13 Gray 481 (1859); and Stevens v. B. & M. R. Co., 1 Gray 277 (1854).

The Denny case is somewhat like the Rodgers case; and the defendant was held not liable.

In the Stevens case, the goods had arrived at the place of destination, and were in the railroad freight depot. The consignee sent a teamster to remove the goods. The railroad freight agent negligently believed that the goods were not there, and negligently told the teamster so. Hence the teamster went away without the goods. The goods, remaining in the freight depot, were destroyed by an accidental fire during the following night. Under the law of Massachusetts, a fire occurring without fault of the railroad would not per se make the railroad liable. Norway Plains Co. v. B. & M. R. Co., 1 Gray (Mass.) 260 (1854). Held, that the railroad was liable for the value of the goods. Shaw, C. J., said, at pp. 281-282 “... we think that, as the negligence

of the agent of the corporation in this case prevented the plaintiffs from getting their goods into their own possession on Monday afternoon, by means whereof they remained in the depot and were burnt, the loss was so directly the consequence of this default on the part of the corporation that the value of the goods. . . is the just rule of damages."

The Stevens case has come up (substantially) in several jurisdictions, and the weight of authority is with the decision in 1 Gray 277. See Railroad v. Kelly, 91 Tenn. 699, 20 S. W. 312 (1892), and Central Trust Co. v. East Tennessee, etc. Ry. Co., 70 Fed. 764 (1895).

How can we reconcile the Stevens case with the Denny case or the Rodgers case? The negligent misstatement (and the leaving the goods in the depot) had no tendency to cause the fire, but merely exposed the goods to the fire should it occur. There seems to have been no reason for anticipating that the fire would occur. Again, the defendant's conduct in both cases was simply negligent, not wilfully wrong. The delay in transportation in the Rodgers case was in one sense passive conduct, and the mistaken statement in the Stevens case was in one sense affirmative conduct; but the gist of the tort in each case is negligence.

It might be said that the existence of causal relation is a question of fact, and a question of degree, not determinable by definite rules; that in the Stevens case the defendant's fault is nearer in point of time to the happening of the damage; and that the defendant's fault is more obviously a substantial factor (potentially operative) in the damaging result. But if the results in these cases cannot be reconciled, we think it does not follow that the Stevens case is wrong. We prefer the other alternative, that the Rodgers case is wrong.

In 18 Yale L. J. 340-342, Mr. Larremore, if we understand him rightly, concedes that, under the established legal doctrine (“under existing abstract rules”) as to causation, a decision like that in the Rodgers case is logically sound. But he believes the result to be practically unjust; and hence thinks the court should establish, for such cases," a direct exception" to the established "theory of proximate and remote causes." If Mr. Larremore would follow Mr. Beven and Professor Bohlen, he would find a better way. The Rodgers case and its fellows all proceed upon the idea that there is an arbitrary rule of law, that a tortfeasor is not liable for improbable consequences. Mr. Beven and Professor Bohlen deny the existence of such a rule; and we think that they are right.

The foregoing reasoning would tend to establish the liability of a carrier, who was not "a common carrier." If the defendant is "a common carrier," the argument against him is still stronger. Such a defendant, in setting up the plea that the damage was due to the act of God, is claiming the benefit of an exception to the stringent general rule as to a common carrier's liability. But the benefit of this exception should be allowed only to those common carriers who are personally free from fault. It ought not to be allowed where the carrier's tortious delay exposed plaintiff's property to destruction by an extraordinary departure from the usual course of nature. See Brown, J., in Bibb Broom Corn Co. v. Atchison, etc. R. Co., 94 Minn. 269, 275-276, 102 N. W. 709, 710-711 (1905).

As to the second class of cases, which it has sometimes been supposed must constitute an exception to a general rule that the improbability of a result does not per se exonerate a wrongdoer:

Class 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.

We have seen (ante, 25 HARV. L. REV. 118-119 et seq.) that there was formerly a tendency to hold that an earlier wrongdoer was never liable, though the intervention of the later wrongdoer was foreseeable. While this view prevailed, it would have been hopeless to contend that the earlier wrongdoer could ever be liable in case of nonforeseeable intervention. But now that this former view is generally abandoned, the question as to the possible liability of an earlier wrongdoer in case of non-foreseeable intervention is entitled to judicial consideration.

Suppose that, under our suggested rule of causation, the earlier tort is found as a fact to be in part the cause of the commission of the later tort, and thus in part the cause of the damage to the plaintiff which followed immediately upon the commission of the later tort. But suppose also that these consequences were not foreseeable (as probable) at the time of committing the earlier tort. Would the earlier tortfeasor be exonerated on that ground alone? We think not. The vital question is, whether the earlier, as well as the later, acts are "traceable by their substantial effects to the ultimate result which constitutes the injury." It may be that "such an entirely new form has been imparted by the later act to the abnormal conditions created by the earlier act, that it would be unjust to hold the author of the earlier act responsible for the final injury. But if no such metamorphosis has taken place, and if the injury is physically an actual result of a coöperation between the abnormal conditions created by both acts, either of the authors of those acts must, upon any rational principles, be regarded as responsible for a part of the injury" (though in measuring the amount of recovery "the law will not usually undertake to apportion the share of each, but will hold each liable for the whole "), "and it is idle to ask whether the later act was one which might have been anticipated. . . . The law should concern itself, not with the time at which an act is done, but with the question whether the act is still potentially operative for harm at the time the injury itself was inflicted." See Mr. Labatt, 33 Can. L. J. 720, 721. And these principles should govern (apply) even though the commission of the later tort was not induced directly or indirectly by defendant's earlier tort.

It has been said: ". . . if two distinct causes are successive and unrelated in their operation they cannot be concurring. One of them must then be the proximate, and the other the remote, cause." Williams, J., in Kerr v. City of Lebanon, 149 Pa. St. 222, 226-227, 24 Atl. 207, 208 (1892).

If "proximate" cause is here used in the sense of "legal" cause, the statement is erroneous, as implying that contiguity in space or nearness in time are legal tests of the existence of causal relation and that the antecedent which is nearest in space or time is invariably to be regarded as the sole legal cause. See ante, 25 HARV. L. REV. 107-108, comments on Bacon's Maxim.

Moreover, there are cases where "the original act of negligence, the primary causation, may be in its nature so continuous that the concurrent wrongful act precipitating the disaster will in law be regarded not as independent, but as conjoining with the original act to create the disastrous result." Henshaw, J., in Merrill v. Los Angeles Gas & Electric Co., 158 Cal. 499, 505, 111 Pac. 534, 537 (1910), citing as one instance Pastene v. Adams, 49 Cal. 87 (1874).

If two causes are operating together and each is a substantial factor in producing the damage, "it is not necessary that the beginning of their operation should be simultaneous." See Bishop, Non-Contract Law, § 450. "So long as the act of the defendant still concurs with the new human act it remains a proximate cause of any further loss." I Sedgwick, Damages, 9 ed., § 126.

It has been said: ". . . directly the natural course of events is . . . accelerated ... by any other impelling agency, that agency becomes the causa proxima and the

natural consequences of the original impelling agency are held to cease," unless the original wrongdoer ought to have known of the new cause and ought to have foreseen the probable result. Pigott, Torts, 166. But Professor Bohlen rightly says that the intervening agency, in order to be regarded in law as breaking the causal connection between the original wrong and the damage, "must divert, and not merely hasten, the natural effect of the wrong, . . ."; 40 Am. L. Reg. N. s. 163, citing Elder v. Lykens Valley Coal Co., 157 Pa. St. 490, 27 Atl. 545 (1893). And see Coleman, J., in Thompson v. L. & N. R. Co., 91 Ala. 496, 499-500, 8 So. 406, 407-408 (1890).

A part of the following passage from the opinion of Miller, J., in Insurance Co. v. Tweed, 7 Wall. (U. S.) 44, 52 (1868), has often been quoted in reference to this subject: "One of the most valuable of the criteria furnished us by these authorities is to ascertain whether any new cause has intervened between the fact accomplished and the alleged cause. If a new force or power has intervened of itself sufficient to stand as the cause of the misfortune, the other must be considered as too remote. In the present case we think there is no such new cause."

These dicta related to the construction of a clause in a contract of insurance, not to causal relation in actions of tort. "Some obscurity . . . has arisen from the attempt to draw analogies from insurance cases for application to cases that are governed by quite different principles." Terry, Leading Principles of Anglo-American Law, § 549. In actions on contracts of insurance there has been a tendency to hold that, for certain purposes as between the insured and the insurer, the cause nearest to and immediately preceding the loss is alone to be regarded. See Knowlton, C. J., in Lynn Gas & Electric Co. v. Meriden Ins. Co., 158 Mass. 570, 576, 33 N. E. 690, 691 (1893). But "consequences may be proximate in actions for tort which would not be so if the question were of an underwriter's liability to pay for losses." Terry, Leading Principles of AngloAmerican Law, § 549. Thus a shipmaster's negligence "may be a proximate cause of the loss in an action against him" (by his employer) "for neglect of duty and a remote cause in an action against the underwriter on the policy." Terry, Leading Principles of Anglo-American Law, § 543. "In an action on a policy the causa proxima is alone considered in ascertaining the cause of loss; but in cases of other contracts and in questions of torts the causa causans is by no means disregarded." Lord Lindley, in Fenton v. Thorley & Co., [1903] A. C. 443, 454.

The phrase "of itself sufficient to stand as the cause of the misfortune" is ambiguous. See Terry, Leading Principles of Anglo-American Law, §§ 549, 558. It may mean - a new force which would have come into operation just the same and which would have produced the same final damage at just the same time, even though the defendant's earlier tortious conduct had never taken place. If the premise is construed in this way, the conclusion is not likely to be disputed. But the phrase may mean a new force, sufficient when, and only when, added to, or operating in connection with, the effect already existing as a result of defendant's tort. If this is the correct interpretation, then we submit that the proposition is not universally true. There are cases where the earlier tortious act may have such continuous efficacy (may continue to be so potentially operative) that it must be regarded as a substantial factor in subjecting plaintiff to the damage. "We have been cited to no authority in a suit for the recovery of damages, where it was shown that, if the 'result' was the necessary and inevitable effect of a first cause, and a new independent force intervened sufficient of itself to produce the effect, and only hastened the result, the first cause was held to be too remote. In such cases both causes necessarily contribute to the result." Coleman, J., in Thompson v. L. & N. R. Co., 91 Ala. 496, 500, 8 So. 406, 408 (1890). Because the later or intervening tortfeasor can be held liable, it does not necessarily follow that the earlier tortfeasor is exonerated. The contrary view seems

to have been taken by Strong, J., in Milwaukee & St. Paul Ry. Co. v. Kellogg, 94 U. S. 469, 475 (1876), in the following passage:

"We do not say that even the natural and probable consequences of a wrongful act or omission are in all cases to be chargeable to the misfeasance or non-feasance. They are not when there is a sufficient and independent cause operating between the wrong and the injury. In such a case the resort of the sufferer must be to the originator of the intermediate cause." But this is in effect the same fallacy involved in the view of the trial judge in Vicars v. Wilcocks, 8 East 1 (1806), namely, that the fact that the plaintiff has a remedy against the "later" wrongdoer constitutes per se a sufficient reason for denying him a remedy against the "earlier" wrongdoer. Mr. Bower's answer to that view has already been quoted in an earlier part of this article:

"Where it is a question whether A. has been injured by B., it is wholly immaterial whether he has or has not an additional or alternative remedy against C., and it can never lie in the mouth of a wrongdoer, if he is a wrongdoer, to set this up." Bower, Code of the Law of Actionable Defamation, 315. Compare Watson, Damages for Personal Injuries, § 74.

Of course, it is not contended that the earlier of two "successive" wrongdoers would always be liable. On the contrary, in a large proportion of cases it would be found as matter of fact, and rightly found, that the earlier tort was not potentially operative at the time of committing the later tort (or, at all events, not so at the time of the damage) and that the later wrongdoer was the sole substantial human factor in bringing about the damaging result.

"There is no doubt that a man sometimes may be liable in tort, notwithstanding the fact that the damage was attributable in part to the concurrent or subsequent intervening misconduct of a third person. . . . But the general tendency has been to look no further back than the last wrongdoer, especially when he has complete and intelligent control of the consequences of the earlier wrongful act." Holmes, J., in Clifford v. Atlantic Cotton Mills, 146 Mass. 47, 48, 49, 15 N. E. 84, 86, 87 (1888).

Moreover, if the only fault sought to be imputed to the earlier conduct was that of negligence, it might be held that there was no negligence if it was improbable that the earlier conduct would tend to induce the commission of a subsequent tort.

Even if the foregoing general views are deemed sound, it may still be contended that the wilful tort of a later wrongdoer, if not foreseeable as probable, will always and necessarily break causal connection, and thus prevent holding the earlier wrongdoer as a part of the cause of the damage. It cannot be denied that such an idea has been entertained. See Holmes, L. J., in Sullivan v. Creed, [1904] 2 I. R. 317, 356. But we are inclined to question its universal applicability to all conceivable situations. There are wide diversities of fact as to the nature of the tort of the earlier wrongdoer, and as to the extent of its continuing efficacy. Those diversities (as to continuing efficacy) are differences of degree, not reducible to rule. All that we contend for is, that the conduct of the earlier tortfeasor may, in some cases, be of such a nature and have such a continuing effect that a jury can find it to be a substantial factor in subjecting the plaintiff to the damage, i. e. find it to be a part of the cause, and not merely an antecedent fact whose effect ceased when the later wrongdoer began to commit a wilful tort. In Fottler v. Mosely, 185 Mass. 563, 70 N. E. 1040 (1904), the defendant, the earlier of two "independent" wilful tortfeasors, was held liable in an action of deceit, although he did not foresee the commission of the later tort; and although the earlier tort had no tendency to induce the commission of the later tort, but merely exposed the plaintiff to a risk of loss in case the second tort should be committed.