delivered the opinion of the court.
.Appellant is engaged in the manufacture, in Laurel in this state, of a building material known as Masonite boards. In this process second-growth pine is used, which is first chipped into small pieces, and these pieces are subjected to a high pressure of steam, which pressure is suddenly released, causing the pieces to explode into *162fiber. This fiber is put through a refining process and is mixed with water, so that it can- be pumped through the mill into the storage chest. Up to this point no chemicals of any kind are used, but in the storage chest a waterproofing or sizing material called petrolatum is added in quantity equal to about two per cent, of the weight of the fiber. Of this petrolatum or sizing material all, or practically all, is retained in the finished board, but there is nothing in the petrolatum which is poisonous. After the sizing has been added, the material flows over to a machine similar to a paper machine, where the excess water is drained away and the fiber remains in the form of a sheet, and then it is subjected to heat and pressure by the proper devices and thereby converted into boards suitable for building qrarposes.
The only other chemical used in the plant is caustic soda to clean the pressure plates, and there is a small amount of chromic acid which results as a waste or dripping from the iron plates which are coated with chromium. The amount of these chemicals or either of them which finds its way into the effluent of the plant is relatively negligible, and it is shown that in these quantities there is nothing harmful. There is suspended in the effluent a relatively small portion of the wood fiber, this suspension amounting to about one-half of one per cent, of the aggregate of the .effluent, and this effluent, in the total of about one million to one million three hundred thousand gallons of water per day, is discharged into Tallahala creek. There is also in this effluent various of the chemicals or acids which are naturally present in wood and which are to some extent released by the explosion of the pieces of wood into fiber, of which the only actively poisonous element is a small amount of methyl alcohol, later to be mentioned. When this effluent, with its suspension of wood fiber and the other natural elements which result from the explosion of the wood in the plant, reaches the stream into which it is discharged, it *163enters upon what may be called, to use a familiar term, the process of rotting; it turns the water dark, it gives off a more or less offensive odor, and this decomposition uses up the suspended oxygen in the stream so that all the fish die, not from poisons but from the absence of a sufficient quantity of suspended oxygen in the water. There is a large effluent of city sewerage cast into said stream, not only from the city of Laurel, and from all its dwellings, stores, hospitals, industrial enterprises of various kinds and the like, but also from the town of Ellisville; and there are other sewerage outfalls along the course of the stream, but we do not deem it necessary to detail all these. It is shown by the testimony that, when all these effluents, including the effluent from appellant’s plant, reach the stream, chemical changes take place, but what these changes are is not disclosed by any direct evidence.
The Tallahala creek into which the effluent from appellant’s plant is emptied flows in a southerly direction and into Leaf river in Perry county. In the latter county, at a point about forty or forty-five miles below Laurel, appellee, on a June day in 1929, went bathing in Tallahala creek, and soon, upon wading into the water, he experienced burning sensations in his feet and in those portions of his body touched by the water. He immediately came out and thoroughly bathed himself with pure water at his home, but he steadily grew worse, and finally was compelled to seek the services of a physician, who found him with eruptions on his feet and legs, which were swollen, in a high state of inflammation, and accompanied with fever. The physician diagnosed the trouble as being the result of a poisonous irritant of some kind applied to the skin. Subsequently, appellee brought an action against appellant for damages, alleging that he was injured by irritant poisons from the effluent from appellant’s plant, and prevailed in the trial court upon a course of evidence composed, to a considerable extent, of infer*164enee upon inference, as shall hereinafter be briefly mentioned.
Appellant contends that the proof in behalf of appellee, the plaintiff, as regards the ultimate fact essential to recovery by appellee, amounts only to a conjecture or a possibility, and does not possess the dependable character of an established probability. And appellant relies upon the statement found in many reported opinions, as, for instance, in Siemer v. Ry. Co., 180 Ky. 111, 118, 201 S. W. 469, that inferences drawn from other inferences are not sufficient upon which to determine the rights of litigants. See, also, 22 C. J. pp. 84, 85; 10 R. C. L. p. 870; 9 Ency. Ev. p. 880. If we were to give unqualified adherence to a rule so rigid as thus stated, there would be an end of this case without further elaboration, and upon a simple analysis of the long chain of inferences, one linked to the other, extending between the alleged wrongful injury and the original source thereof alleged to be a product of the effluent from appellant’s manufacturing plant.
Because appellee was in good health, so far as he knew, and had observed nothing unusual in that respect or indicative that he was about to experience any disorder, but when he went into the water he first felt the symptoms of the disorder which soon thereafter openly manifested itself, the first inference is that the water caused it; and, since pure water does not cause such an injury, the second inference is that some poisonous foreign substance in the water was the active cause. Because of the fact that, if others had in the past been similarly injured, the news of these injuries would likely have been current in that section of the country, the third inference is that the poison in the water had been lately injected therein; and, since there is a quality in flowing water by which it tends to rid itself of impurities as the current progresses down stream, the- fourth inference is that the poison was injected therein by some agency *165which precipitated a large quantity of foreign matter into the stream, and, since appellant’s plant, as shown by the evidence, is the one latest established and which injects a large effluent into the stream, the fifth inference is that appellant’s plant was the offending source.
But at this point, or at the last of the stated links of inference upon inference, appellee is met by the undisputed proof that, although the manufacturing plant of appellant discharges into this flowing stream about one million to one million three hundred thousand gallons of water per day, there is nothing in this effluent from appellant’s plant of an actively poisonous nature, whether taken internally or applied externally, except a small quantity of methyl alcohol, shown by the proof to be about one drop to a million gallons, and even then, as we know from the accepted chemical authorities, methyl alcohol in small quantities in dilution is not harmful by external application. Thus the chain of inference would appear to have been completely cut so far as it would connect appellant’s plant in the sequence of causation; but appellee replied by proof that the effluent from appellant’s plant, although not actively harmful to the human body when it was cast into the flowing stream, would in that stream undergo chemical changes, and, although the nature of these changes is not shown, or attempted to be shown, in the testimony, nothing directly shown therein that these changes are harmful changes, yet the sixth inference is sought to be raised that the changes are harmful changes, because, tracing back through the chain of inferences, there is one that appellee was injured because of some poisonous substance in the water. And it is at this sixth point — whatever may be admitted as to the strength of the previous links — that the judicial mind, in the impartial endeavor to secure to litigants a' determination of their rights only upon evidence which is cogently safe and dependable, must certainly pause *166and most seriously inquire whether probative probabilities have fallen behind and only possibilities are present.
While it is true that the rule is an established one that a presumption may not be based upon another presumption, Hinman v. Sabin, 147 Miss. 509, 514, 112 So. 871, and that following this many courts have said, and most of the texts announce, that an inference essential to establish a cause of action may not be based upon another inference, we think the rule holds good only as to legal presumptions, strictly speaking; but that, as to inferences deduced from the facts, it is not the unqualified rule that an inference may not be based upon another inference. Numerous cases of circumstantial evidence found in our books, and many trials in the everyday experience of our bench and bar, disclose that inference upon inference is availed of and is enforced. 1 Wigmore, Ev. (2 Ed.), pp. 258-260; 1 Jones, Com. Ev. (2 Ed.), pp. 625-637.
But the great weight of opinion, from so many of the able courts of the country, to the effect that an inference essential to the establishment of the cause of action may not be based upon another inference, admonishes us, first, that, in declining to recognize that rule, so stated, in this jurisdiction, we must, in allowing inference upon inference, do so with the firm limitation that the probabilities thereby permitted to be entertained are safe and dependable probabilities, measured by legal standards, for they involve more than the simple and generally unimportant affairs of everyday life; they involve in court procedure the liberty and property of others. Obviously every inference drawn from another inference produces a result wherein the quality of probability becomes weaker and soon or later a stage is reached when serious doubt arises whether, under legal standards, the ultimate inference in the chain of inferences is a legally safe and dependable probability or has become only a more or less *167strong possibility, and, when that stage is reached, the proof is insufficient, so far as concerns judgments at law.
The second admonition is that, in allowing inference upon inference, we should do so no further than the reasonable necessities of the case, in the interest of justice, require. Fortunately, in nearly all cases, particularly in civil cases, the necessities of the case do not require that the chain of inferences shall be lengthened, one link to another, through such an extent as to present as a practical question the problem whether the chain of inferences has become too long and therefore too weak to hold; for in nearly every case it will be found that one or more of the essential facts sought to be established by inference is or are capable of proof by direct, positive, or by demonstrative evidence, and thereupon there comes into play the rule that, where a party, who has the burden of proof, has the power to produce evidence of a more explicit, direct, and satisfactory character than that which he does introduce and relies on, he must introduce that more explicit, direct, and satisfactory proof, or else suffer the presumption that, if the more satisfactory evidence had been given, it would have been detrimental to him and would have laid open deficiencies in, and objections to, his case, which the more obscure and uncertain evidence did not disclose. 22 C. J. p. 115; 23 C. J. p. 40; 10 R. C. L. p. 885; 9 Ency. Ev. pp. 959, 960; 1 Jones, Com. Ev. (2 Ed.), pp. 185, 186; Wigmore, Ev. (2 Ed.), p. 585; Clifton v. United States, 4 How. 242, 11 L. Ed. 957.
And that is exactly the ease we have here. One of the absolutely essential links in the chain in this case is that the water contained poisonous elements which caused the 'particular injury, and this is sought to be established by inference. But this was a fact capable of direct and demonstrative proof by a chemical analysis of a sufficient sample of the water taken from the place where the alleged injury occurred and within a reasonable time thereafter. The state has in its employ chemical experts of *168the highest repute who, on request, in a case such as this, would make such an analysis, either free of charge or •for only a nominal cost, and would do so impartially and dependably. Such an analysis would demonstrably disclose whether or not the water at the place of injury contained the poisonous or injurious elements which appellee has charged against appellant, and, moreover, would disclose whether, and the manner by which, the nonpoisonous effluent of appellant’s plant had by subsequent changes become poisonous. Appellee, having the burden of proof, did not take this essential step, essential to the safe and satisfactory administration of justice, and, with a chain of inferences elongated and weakened by so many links as is the chain here, the presumption raised by that failure must be adjudged to turn the scales of judicial decision in favor of appellant. In other words, and to sum up what we have said, we shall allow in this jurisdiction the establishment of a case by inference upon inference so long as the ultimate inference, measured by legal standards, is, without too much doubt, a safe and dependable probability; but no such inference upon inference will be permitted to prevail when the fact sought to be established by such inference upon inference is capable of more satisfactory proof by direct, or positive or demonstrative, evidence, within the reasonable power of the party holding the burden to produce. In thus prescribing, we secure the administration of justice in a more dependable way, and at the same time reconcile our holding with the real weight of authority.
Reversed and dismissed.