CLARK, Circuit Judge.

This case arises from a unique and tragic accident. Plaintiff, a college graduate, 25 years of age, was a teacher of mathematics and general science in the Essex High School of Essex, New York. Late in 1936, he received a pamphlet from a teachers' college giving a list of free and low-cost educational materials for elementary and grammar grades obtainable from various sources and companies. From the list it appeared that there could be procured from the defendant, among other things, the following: "Sinclair Exhibit. A process oil display on request — to Teachers only." Accordingly he wrote the defendant and in due course received its display, consisting of six four-ounce bottles and two jars of petroleum products. Each separate bottle or jar had pasted on its outer surface a label, on which appeared the defendant's name and trademark and the name of the product supposedly contained within. Each bottle was filled and was corked and sealed with wax or paraffin. The plaintiff also received an acknowledgment of his order in a letter, in which the defendant said that in accordance with the plaintiff's recent request it had asked its East Chicago refinery to send him "a free set of oil samples," that the "samples show various forms of oil ranging from crude petroleum to the most highly refined products," that it trusted the exhibit would reach him in good condition, that it was sending also a copy of its oil book containing information about production, refining, and marketing of petroleum products, "which may be of value in class work," and that it greatly appreciated the opportunity "to assist you in the work of educating the young people of America."

On receipt of the display plaintiff opened the carton and placed it on a shelf in his laboratory cupboard. He showed it on one occasion to his class and at another time he opened the container marked "wax" and used some of the wax in it. On April 15, 1937, he opened a can of sodium metal which he had procured from sources other than the defendant. Sodium metal, or metallic sodium as it is often called, when combined with water, is highly explosive, and the plaintiff planned to conduct an experiment before his class the next day to demonstrate this fact, using a very small amount of the sodium he had in his possession for this purpose. On the other hand, in the field of chemistry, kerosene is a recognized preservative of sodium metal. In preparing for the experiment and to preserve the sodium until needed, the plaintiff opened the bottle in the oil display marked "kerosene" and, without smelling or otherwise testing it, poured its contents onto the sodium. There was a terrific explosion and the plaintiff received severe burns, which, among other things, caused the loss of one eye. It then appeared that the defendant had filled the two bottles marked "kerosene" and "gasoline" with water, apparently to make the exhibit mailable (although here express service was used), as well as to avoid the appearance of discoloration of pure kerosene and gasoline. Perhaps, too, it hoped to make the display safer for all recipients, as the defendant claimed, though there was not direct evidence on this point and there was evidence that kerosene was not considered unsafe, certainly not so much so as the crude oil, for which no substitute was provided. No warning of any kind that the contents of the two bottles were not as labeled had been given.

Plaintiff brought suit in the Supreme Court of the State of New York. On defendant's petition the action was removed to the federal court on the ground of diversity of citizenship of the parties. In a trial to the jury the plaintiff offered direct evidence of the accident and its consequences, and then introduced the testimony of experts to show the properties of the chemicals in question, a general policy recognized among chemists of careful labeling of products, the known qualities of kerosene as a preservative of sodium, and of water as explosive in combination with sodium, that chemists did not rely on the sense of smell to test chemical products, since this is not a reliable or proper test, and that there is little, if any, odor in refined kerosene. An expert who had had experience with a concern which supplied educational displays testified that, while companies tried to limit such exhibits to display purposes alone, it was recognized that this was impossible and it had to be contemplated that such exhibits might be used for any purposes for which they would be available. Defendant introduced no evidence, but made motions for dismissal of the complaint at the close of the plaintiff's evidence, and later for direction of the verdict, and then to set aside the verdict after it was rendered, relying on its claims, again pressed vigorously here, that there was no evidence of negligence on the part of the defendant to be submitted to the jury, and that the plaintiff's contributory negligence was established as a matter of law. The court denied all these motions, but submitted the case to the jury with an appropriate charge on each of these issues. The jury rendered its verdict for the plaintiff in the sum of $24,000. From the judgment on the verdict, defendant appeals.

This, therefore, is one of those cases where not unusual human conduct produces results so unexpected and tragic as to startle and amaze. Yet legal precedents and human experience combine to show that bizarre accidents are far from unlikely, and recovery cannot be denied because of the uniqueness of the happenings. Usually judicial rationalization is couched largely in terms of "foreseeability," but it is obvious that, if it is the exact accident which must be foreseen, then recovery must regularly be denied (as it is not). Perhaps in this class of cases, it is easier to reach decision than to explain it, for explanations tend towards abstract generalities, under which are lumped all sorts of accidents, from railroad and automobile collisions to chemical explosions. At any rate, given some culpability on the part of a defendant, i.e., some conduct involving a departure from the natural, and hence the reasonable, then the courts look more for the possibility of hazard of some form to some person than for an expectation of the particular chance that happened.

Now the result can be stated in general terms of negligence as involving unreasonable conduct, or departure from the standard of that great legal figure — the ordinary prudent man — or it can be particularized in terms of a duty resting on a relationship between the parties, or on a finding of proximate cause. Much discussion there has been of all these matters — by judges and lawyers, by scholars and professors. A recent article shows a refreshing sophistication in picking a course amid all the theories: Gregory, "Proximate Cause in Negligence — A Retreat from `Rationalization,'" 6 U. of Chi.L.Rev. 36 (1939). The author refers to these concepts as devices employed "to control litigation before a court and jury" and as conventional ways of talking about legal liability, and suggests that most cases in which the issue of proximate cause predominates might just as easily be decided under the duty issue or vice versa. He says (p. 51):

"It cannot be too frequently repeated that we must not ask too much of the concept of foreseeability, even after we have discounted it in light of the remarks just made. For we must always remember that more than the judge's actual experience and his knowledge of the world contribute to his choice of the factors on which a chain of reasonable probability or foreseeability may be based. What he thinks ought to be, what he wants to see happen — in other words, his values and his notions of sound and desirable social policy — are bound to play a large part in influencing his choice or repudiation of the factors upon which a claim of probability or foreseeability leading to liability may be created. And if a judge's function is to do justice, this, after all, is no more than we should expect."

His conclusion (p. 61) is to come back to Street's famous statement (Foundations of Legal Liability, Vol. I, p. 110) that no definite principle of "proximate" and "remote," meaning, recoverable and non-recoverable, damages, can be laid down, but that the question "is always to be determined on the facts of each case upon mixed considerations of logic, common sense, justice, policy, and precedent," and "the best use that can be made of the authorities on proximate cause is merely to furnish illustrations of situations which judicious men upon careful consideration have adjudged to be on one side of the line or the other."

This perhaps does not help to a conclusion, but it does clear away some underbrush. We must then turn to a consideration of the parties' acts, looked at in the light of natural and expected ways of doing things.

On the issue of the contributory negligence asserted against the plaintiff, we need not hesitate long. The conduct of the plaintiff, in making use of one of these samples of rather common products, delivered into his hands without admonition or restriction, and in failing to test it by smell or otherwise, seems a sufficiently natural, i.e., reasonable, act, so that the court could not properly have held the plaintiff negligent as a matter of law. The testimony of the experts as to the practice of chemists of course supports this conclusion.

The real question is whether the defendant's act, in substituting water for the kerosene without any warning of the substitution, is sufficiently unnatural to make an issue properly to be submitted to the jury on the question of reasonableness of its conduct. We think here, too, that most of us would feel that the defendant's conduct was culpable, even though the consequences were much more severe than we would have expected. The act of substituting water for the kerosene does not seem so strange in itself, but the mislabeling without warning of any kind, either inserted in the carton when it was transmitted or contained in the letter mailed at the same time, or otherwise, does seem unreasonable. The cartons were intended "for teachers only." It would have been so easy to have warned the recipient. One does not need profound chemical experience to know that water is combined with chemicals to produce various reactions, that kerosene has its uses in the laboratory, as well as in the home, and that kits sent to schools are subject to all the chances to which the varied population of a school may subject them. If we feel the defendant at least somewhat culpable in failing to take the simple step of a warning, then we see no reason to take the case from the jury when the consequences are serious, and, because serious, are unexpected. We think, therefore, that the issue of the defendant's negligence was properly left to the jury.

If it is desirable to rationalize this result in terms of the defendant's duty — since it argues that it was engaged in only a free educational venture — we think that not difficult. After all, the defendant expected to, and doubtless did, receive much benefit from distributing its products among schools in this fashion. At any rate, it offered to, and did, set in motion a chain of circumstances leading to plaintiff's injury. As Chief Judge Cardozo said in Moch Co. v. Rensselaer Water Co., 247 N.Y. 160, 167, 159 N.E. 896, 898, 62 A.L.R. 1199: "The hand once set to a task may not always be withdrawn with impunity though liability would fail if it had never been applied at all. * * * If conduct has gone forward to such a stage that inaction would commonly result, not negatively merely in withholding a benefit, but positively or actively in working an injury, there exists a relation out of which arises a duty to go forward." Here defendant had launched upon a course of conduct, had set the stage, in such a way that it should go the little distance further to give notice of what it had done. And Judge Cardozo's famous decision in Palsgraf v. Long Island R. Co., 248 N.Y. 339, 162 N.E. 99, 59 A.L.R. 1253, does not hold otherwise, for there the unusual accident was caused by a hazard which the defendant could not reasonably foresee and against which it could not reasonably be expected to guard. 6 U. of Chi.L.Rev. 45, 46, 48.

So, also, proximate cause may be found for the accident from the defendant's neglect. L. Hand, J., puts the matter succinctly in The Mars case, D.C., S.D.N.Y., 9 F.2d 183, 184, when he states as the only intelligible line of distinction, "What would an ordinary man expect under all the circumstances, to be the result?" and then goes on to say that not merely physical causation is sufficient, but "there must also be some mental element. It has got to be one of those consequences which is not entirely outside the range of expectation or probability, as ordinary men view it." But as pointed out in Tullgren v. Amoskeag Mfg. Co., 82 N.H. 268, 276, 133 A. 4, 8, 46 A.L.R. 380, "it is not necessary that damage as a more rather than less probable result should be anticipated. * * * Danger consists in the risk of harm, as well as the likelihood of it, and a danger calling for anticipation need not be of more probable occurrence than less. If there is some probability of harm sufficiently serious that ordinary men would take precautions to avoid it, then failure so to do is negligence. * * * The test is not of the balance of probabilities, but of the existence of some probability of sufficient moment to induce action to avoid it on the part of a reasonable mind."

Or as Judge Davis well says in New York Eskimo Pie Corp. v. Rataj, 3 Cir., 73 F.2d 184, 186: "That does not mean that in this case it must have been able to anticipate the exact details from the time of its negligent act to the injury of the child by the explosion." And there liability was sustained where the defendant's driver threw some dry ice in the gutter and a child put some in a bottle, with the resulting explosion from its rapid solidification and expansion (500 times its volume if unpacked), putting out the eye of the child's sister. Other cases which help to persuade to a decision may include Parnell v. Holland Furnace Co., 260 N.Y. 604, 184 N.E. 112, affirming 234 App. Div. 567, 256 N.Y.S. 323; Payne v. Manhattan Queens Traction Corp., 277 N.Y. 393, 14 N.E.2d 449, 115 A.L.R. 1495; Smith v. Peerless Glass Co., 259 N.Y. 292, 181 N.E. 576; Rosebrock v. General Elec. Co., 236 N.Y. 227, 140 N.E. 571; and MacPherson v. Buick Motor Co., 217 N.Y. 382, 111 N.E. 1050, Ann.Cas. 1916C, 440, L.R.A. 1916F, 696. All of these involve situations where some defect or default of the defendant led to rather surprising consequences, but liability was supported. We do not read the recent case of Morse v. Buffalo Tank Corp., 1939, 280 N.Y. 110, 19 N.E.2d 981, as being opposed to these cases. There a divided court did not find liability to an infant plaintiff injured by a fire from gasoline procured from a drip can on defendant's premises by trespassing boys. The court relied upon the New York refusal to follow the doctrine of attractive nuisance, but the chain of causation was attenuated at best. Compare Am.L.Inst., Torts, §§ 310, 388.

Hence we conclude that the conduct of both parties should go before the jury. The defendant asserts that such a course, with the corporate defendant here, means that a verdict for the plaintiff is certain to follow. Even if that is justifiable prophecy, it still does not mean that the jury is wrong or that a general community standard as to what should be the risks of the business undertaken by the defendant must be disregarded. "As the experience of one man usually differs from that of another, our law wisely says that what is `reasonable' is to be determined by the jury — that is, it is to be the resultant of the, to a certain extent varying, opinions of twelve different persons." Palles, C.B., in Sullivan v. Creed [1904], 2 Ir.R. 317, 330. And there we may well leave the decision.

Judgment affirmed.