joined by Justices Smedley, Brewster and Wilson, dissenting.
This dissent rests on my conviction that there is no evidence to support the finding of causal connection between the defendant’s violation of the ordinance and the explosion which killed Mr. Wilson, so that the judgment of the trial court in favor of the defendant should accordingly be affirmed.
The majority statement, “What is proximate cause is a question for the jury to determine from all the facts and circumstances proved,” is not, I’m sure, intended to mean that courts cannot in a proper case decide the question as a matter of law. We have held proximate causation to be established as a matter of law. International-Great Northern R. Co. v. Hawthorne, 131 Texas 622, 116 S.W. 2d 1056. We have also held as a matter of law that it was not established. Western Telephone Corp. v. McCann, 128 Texas 582, 99 S.W. 2d 895; Talley v. Bass-Jones Lumber Co., Tex. Civ. App., 173 S.W. 2d 276, er. ref.; Bowles v. Bourdon, 148 Texas 1, 219 S.W. 2d 779, 13 A.L.R. 2d 1. When the question of “no evidence” is properly raised, as here, it is as much within our jurisdiction — and duty— to scrutinize the reasonableness of jury inferences of causal connection as it is to examine jury inferences of negligence, contributory negligence or any other matter of fact. The domain of inference or deduction from circumstantial evidence is one in which courts are perhaps more than usually justified in reviewing jury action, because the task there involved is more often one of reasoning from undisputed facts than of determining the truth or falsity of disputed facts. Indeed, in the very case before us the Court of Civil Appeals has ignored as unsupported by evidence the jury finding that the intoxication of Mr. Wilson was a proximate cause of the explosion, and the majority evidently does not quarrel with that holding.
*286I do not understand the majority to hold that in negligence per se cases like the present, the plaintiff’s burden to establish proximate causation is satisfied by showing merely that the event which produced the injury was of the general type the ordinance or statute in question was designed to forestall. Waterman Lbr. Co. v. Beatty, 110 Texas 225, 218 S. W. 363, clearly recognizes the rule to be otherwise. The holding in that case that evidence of plaintiff minor’s employment to work with dangerous machinery (in violation of a statute), plus evidence of his injury from such machinery in the course of working with it, was sufficient to make a fact issue of liability, does not affect the instant case. In the Beatty case, clearly but for the illegal hiring of the minor to work with dangerous machinery, he would not have worked with it and accordingly would not have been injured. But in the present case, can it be said with any-think like equal probability, that, but for the illegal use of the flexible hose, the explosion would not have occurred?
The majority opinion seems to suggest, though not to state in so many words, that the jury might have reasonably considered the flexible hose to be in some manner defective as a hose, though the Court of Civil Appeals admits that the evidence was all the other way. I can imagine no other- explanation for the statement that “The rubber hose was not produced in court and the plaintiffs never at any time had access to the hose to make an inspection.” It was not produced in court, and not unnaturally so, since the plaintiffs did not even plead that the hose was defective, asked for no issue on the point, nowhere complained of the failure to submit such an issue and, so far as the record shows, at' no time gave any indication of not having themselves inspected it, much less of not having been able to inspect it. The defendant testified without contradiction or equivocation that he kept the hose in a readily available spot for people to inspect it following the accident and that “after the test was made by the fireman” somebody came out and asked to see the hose or to test it. From the testimony produced by the plaintiffs themselves it appears that the Austin Fire Department investigated and made a report on the explosion, and that a special investigation was also made by a private investigator by the name of K. R. Herbert, who himself testified for the plaintiffs and was presumably acting on their behalf. If there was lack of information about the hose on the part of the plaintiffs or any evidence that it was a defective hose, surely some indication thereof would have appeared in the record. But even the special investigator said not a word on either subject. The overwhelming inference from the record, is that the plain*287tiffs did have “access” to the hose and for good reason chose not to raise the issue of its condition. The defendant testified positively that the hose was in good condition and that the firemen tested it at the scene of the accident and found it airtight. In the absence of evidence contrary to this, and the burden of proof resting on the plaintiffs, did the defendant have to conduct a test of the hose in open court — and prove by disinterested witnesses that the hose thus tested was the same hose in question — in order to escape an inference of defective condition, when the latter was not even alleged or otherwise relied upon by the plaintiffs ? Hardly.
Now the majority, like the Court of Civil Appeals, is right to the extent of recognizing as a legitimate factual inference, that the wall valve was open at some time prior to the explosion on the morning of March 23rd. The event and its character as a gas explosion were undeniable facts as was also the circumstances that gas could not possibly have entered the room except through the wall valve. This inference involves a minimum of speculation. Not so with the further conclusion, which the majority seems to reach, that the valve was opened prior to 9 P.M. of March 22nd, when Mr. Wilson returned to the tourist court and entered his cabin in an intoxicated condition. That there was gas in the cabin at 7 A.M. on March 23rd is certainly no evidence that the valve was open at or before 9 P.M. of March 22nd, or that gas was entering the room at that time. Neither is such evidence to be found in the statement of the defendant that he saw the hose lying disconnected from the wall valve shortly after the explosion. Possible it is found in the unsworn statement attributed to Mr. Wilson by plaintiff Mrs. Kerley and one of her “in-laws”, but this statement was rightly excluded by the trial court as hereinafter discussed. Otherwise there is not an iota of proof — direct or circumstantial — that the wall valve was open or gas escaping prior to 9 P.M. of March 22nd. There is indeed uncontradicated and quite credible evidence that gas was not escaping at that hour or at the earlier hour of about 6 P.M., when Mr. Wilson was first shown into his room. Even could the jury lawfully disregard such evidence as coming largely from the defendant himself, they certainly could not, by disbelieving it, treat it as the opposite of what it was. True, the jury did find that Mr. Wilson did not himself disconnect the hose from the wall jet, as will be later discussed. It also found that his intoxication caused the accident. And conspicuously there is no finding, express or implied, that the wall jet was turned on or that the gas was escaping at or prior to *2889 P.M. of March 22nd, after which Mr. Wilson remained in the room alone and in an intoxicated condition.
So all the evidence whereby the defendant’s use of the hose is to be connected with the explosion is this: there was a dangerous amount of gas in the room at 7 A.M. of March 23rd; the gas came through a pipe in the wall, with a jet, the valve of which was open at whatever time it was that the gas entered; Mr. Wilson turned on an electric light switch, and the explosion followed; a short while after the explosion, and after several people had entered the room, the defendant saw the hose was not connected to the wall jet; the temperature during the ten or so hours just before the explosion was about 65°.
Upon this evidence to say “therefore, but for the flexible hose, there would have been no gas and no explosion” seems obviously a flight into speculation. But this is not all the evidence to be considered.'It must be taken as fact that Mr. Wilson was the only person in the room during the ten hours in question, that he was intoxicated during that time and since he last entered the room at about 9 P.M. of the 22nd and that, while in Austin in evidently good health and equipped with funds and an automobile, he had strangely failed to return to his not distant Waco home, as he said he would, and had just been reported by his wife to the “missing persons” bureau of the Waco police. With these added facts, the speculation that the hose caused the explosion is still less justifiable.
The theory of the majority seems to be that, where a more or less mysterious event gives rise to a variety of speculations as to the cause, and one of them will connect the event with the fault of the defendant, the jury should be left free to adopt that one, even though it be merely a speculation. The Court of Civil Appeals evidently took a view less harsh to the defendant, as it sought forthrightly to justify in detail the reasonableness of the inference, which it considered the jury might properly draw. In so doing, it held void the finding of causal connection between the explosion and Mr. Wilson’s intoxication, yet leaned heavily oii the finding that he did not disconnect the hose at the wall jet. Its theory, restated in more elaborate form, is this: The hose was seen disconnected from the wall jet shortly after the explosion. The jury found (upon sufficient evidence) that Mr. Wilson did not disconnect it. So it may be inferred to have been blown loose by the explosion! From the inference that it was blown loose, one may in turn further infer that it came loose at its “weakest” point (i.e. the connection at the wall rather than the *289one at the stove). From this inference that the wall connection was the “weakest”, that is to say, less strong than the connection at the stove, it may be still further inferred that the “weakest” connection was loose enough to allow gas to escape backwards between the interior portion of the rubber connection tip (as to the character of which there is not a word of evidence except that it was a “rubber tip”) and the exterior of the wall jet, over which (presumably) the tip was designed to fit tightly. From the inference that the tip was thus loose enough for gas to escape, it is further inferred that the gas did so escape. And since such a thing would probably not have happened with a rigid metal connection, it is finally inferred that the use of the hose caused the explosion.
r This line of reasoning errs at the outset in assuming the jury finding that Mr. Wilson did not disconnect the hose to mean that there was evidence that he did not. The issue was a “defendant’s issue”, and there was no proof that Mr. Wilson did disconnect the hose, so the answer had to be in the negative. But neither was there, a shred of evidence that be did not disconnect it. Intoxicated as he undoubtedly was, one may quite as logically speculate one way as the other. The rest of the reasoning, while certainly not unintelligent, is nevertheless pure speculation by the process of piling inference on inference, each step adding its own element of speculative fact until the final conclusion rests on a mass of speculative fact. The jury may not pile inference upon inference. Federal Underwriters Exchange v. Hightower, Tex. Civ. App., 161 S. W. 2d 338, er. ref. The very ingenuity of the theory reflects its essential weakhess. It is indeed one possible explanation of the event, if one is willing to leave quite, unsolved the riddle of when and how the wall valve came to be turned on, so the gas could enter the room. Whatever the. range "of jury freedom to pick and choose between this and that item of evidence from the same witness, theré can be no doubt whatever that Mr. Wilson entered the cabin twice before the explosion — once about 6 P.M. and again about 9 P.M. of March 22nd. Surely the jury could not disbelieve the defendant’s statement that when he let Mr. Wilson in at 9 P.M., he turned on the light. Since it believed his statement that he found Mr. Wilson to be intoxicated at this hour, how could it disbelieve that he did the altogether natural and possible thing of lighting up a room that must have been in darkness? If the wall valve was then open and gas escaping, why was there no explosion at that time? And how could the jury, believing the defendant’s testimony about Mr. Wilson’s condition, at the same time disbelieve his further statement that *290he entered the cabin with Wilson and smelled no gas? It seems almost a certainty on the record that the valve was shut at and for some time prior to 9 P.M. of March 22nd. Unless it be the fact of the explosion ten hours later, there is no evidence at all to the contrary. Under these circumstances, even if we assume the altogether unproved fact that Mr. Wilson did not disconnect the hose, is it not quite as likely that he himself somehow opened the wall valve and otherwise tampered with the gas installations as that gas was leaking into the room from the hose connection when he entered and continued to do so until the explosion the next day? The recognized presumptions of normal conduct obviously do not apply in the case of men sojourning intoxicated in tourist cabins within easy reach of their own homes and families, while their wives anxiously seek them through the “missing persons” bureau. In such cases it is almost the normal thing for something abnormal to happen.
Neither Bock v. Fellman Dry Goods Co., Tex. Com. App., 212 S. W. 635, which originated part of the language quoted by the majority from Burlington-Rock Island Ry. Co. v. Ellison, 140 Texas 353, 167 S. W. 2d 723, nor the latter case itself is closely in point here on the facts. The inferential process in those cases was much less involved and speculative than in the instant case. As to the language used by way of stating a rule of law, if it means that where the facts are equally consistent with the existence of proximate cause and with the absence of it, the jury is at liberty to find proximate cause to exist, it is evidently contrary to our holding in refusing a writ of error in Talley v. Bass-Jones Lumber Co., supra, (a later decision than the Ellison case). In the Talley case the alleged negligence was that of attaching a smoke stack guy wire to an anchor or base which was of wood and inflammable. In a fire the anchor burned, the guy wire was loosened and the smoke stack fell, causing the death of a fireman. The opinion of the Court of Civil Appeals, which we approved, contains this language:
“So it is equally as probable that the falling of stack No. 1 and the consequent death of Talley was caused by the severing of the guy wires by the fire as that the occurrence was produced by a guy wire being attached to the inflammable 2x12 timber at the west end or northwest corner of the fuel house. In such circumstances, the jury may not be permitted to speculate as to which of the two causes was the proximate cause of Talley’s death.” 173 S. W. 2d 276, 279.
The rule thus stated does not seem to require proof of proximate cause beyond reasonable doubt as suggested in the Bock *291and Ellison opinions. It only requires that there be a preponderance of evidence in favor of that version or theory of the event which connects the legal wrong with the injury in suit. See also Western Telephone Corp. v. McCann, supra. If the balance of probability is equal between the theory that entails liability and one that does not, the party asserting liability has simply failed to prove his case.
Viewing the matter in a slightly different way, there is no proof of causation, where the evidence discloses simply a possibility of it, even though it does not appear that other possibilities are essentially more likely. In Bowles v. Bourdon, supra, the principles of which are not necessarily limited to the malpractice type of suit, the physician bound up a child’s arm in a manner which could cause a contracture, and a contracture followed, though it was shown that the disease could also have Gome from causes other than the binding. In holding that there was no evidence of proximate causation, this court said:
“All it (the evidence) shows is that what respondent (defendant) did was not a probable but only a possible cause of the contracture; that it was only one of several things that could have caused the injuries complained of.”
and further:
" “And if the plaintiff would rest upon inferences rather than upon direct testimony, he meets the same rule. ‘The proof must establish causal connection beyond the point of conjecture. It must show more than a possibility. Verdicts must rest upon reasonable certainty of proof. When the proof discloses that a given result may have occurred by reason of more than one proximate cause, and the jury can do no more than guess or speculate as to which was, in fact, the efficient cause, the submission of such choice to the jury has been consistently condemned by this court and by other courts.’ Ramberg v. Morgan, 209 Iowa 474, 218 N. W. 492, 498.” 219 S. W. 2d 779, 785, 13 A.L.R. 2d 1.
No doubt if the hose theory were the only possible theory one could imagine as explaining the explosion, it would have to be accepted, despite all the speculation it involves. It would in such event become a “probability”. See Bowles v. Bourdon, supra. But, as before suggested, under the peculiar conditions prevailing here most anything might have happened, including even attempted suicide of Mr. Wilson. On the record it is almost a certainty that he turned on the wall valve after 9 P.M. on *292March 22nd. He may well have done something else besides. ■ Considered separately, neither the hose theory nor any other can well be called a probability. Considered together, no one is so much more likely to be true than the others as to be called a relative probability. The record simply does not give a “probable” or satisfactory explanation of the gas leakage. Such a situation, with the burden of proof resting on the plaintiff, does not entitled the jury arbitrarily to select an explanation that imposes liability.
The alleged statement of Mr. Wilson proffered by Mrs. Kerley and her relative-in-law and excluded by the trial court was relevant and much more favorable to the plaintiffs than the admitted ones which Mr. Wilson made earlier to the police. But it was righly excluded. Doubtless no one of the several circumstances working against admissibility under the rule of the res' gestae, such as the lapse of some four hours time, the apparently, continuous consciousness of the declarant, the fact that the state- ■ ment came in reply to questions, and the fact that it followed earlier and apparently different statements on the same subject and conversation on other subjects, would necessarily exclude application of the rule. However, taking all these conditions together, they clearly outweigh those of Mr. Wilson’s continuous suffering and state of shock, in determining whether the statement had the requisite degree of spontaneity in relation to the explosion. See City of Houston v. Quinones, 142 Texas 282, 177 S. W. 2d 259; Pacific Mutual Life Ins. Co. v. Schlakzug, 143 Texas 264, 183 S. W. 2d 709. Texas Law of Evidence, McCormick and Ray, sec. 426 et seq.
Justices Smedley, Brewster and Wilson concur with the conclusions here expressed.
Opinion delivered May 9, 1951.
Rehearing overruled June 13, 1951.