Dr. A. R. Bliss, while employed as professor of pharmacology in Emory University, a corporation, was injured by a gas explosion in the stock-room of his department. He sued the university for damages and recovered. The defendant’s motion for a new trial, containing the general grounds only, was overruled, and it excepted. The following is a general description of the surroundings as gleaned from the petition and the undisputed evidence: The plaintiff’s injuries were received in the stock-room. This room was situated a few feet from the laboratory room, in which the students made experiments and received instructions. The stock-room and the laboratory room, with a hall or corridor between, were both constructed with two floors, about four inches apart, the upper floor being of wood and the lower of concrete. The upper floor rested upon sills which were laid upon the concrete. In the space between the floors were pipes through which gas was supplied to the laboratory. These pipes extended through small iron boxes set upon the concrete at various places in the laboratory, and in each of these boxes was a pet-cock to control the outlet of gas. The pipes were not airtight in the holes through which they went into the boxes, and it was physically possible for some air to pass from the boxes into the open space between the floors and vice versa, carrying gas if any should be released from the pipes in either place. The holes were 15/16 of an inch in diameter, while the diameter of the pipes was only *75311/16 of an inch. Bach box had a movable lid, the top of which was flush with the wooden floor. When a student found it necessary to use gas he would remove the lid from the box and connect a rubber tube with the pipe at the cock, then turn on the gas which would be conveyed through the tube to the student’s table. The gas was thus supplied to lamps and burned as needed. When the students were “through using the lamps, they.were supposed to turn off the gas cock, detach the hose, and put the iron cover back” on the box. The gas was subject to use by the classes in the above-described manner several times weekly.. The explosion which caused the plaintiff’s injuries was between the double floors, apparently beginning in the laboratory room and extending into the stock-room. The plaintiff alleged, and sought to prove, that, as a result of a defect or defects in the pipes, gas escaped and . accumulated between the floors and in some way became ignited, and that the defendant was negligent in failing to maintain the pipes in a proper condition and in failing to inspect the same and to warn the plaintiff of the defective condition. The plaintiff testified that “at the time the explosion took place that morning, the students, under [his] instructions, had opened these iron plates [or boxes], put the gas tubes on, and connected them up to the burners and were actually engaged in lighting up lamps. They lit these lamps by striking a match.” The plaintiff, however, undertook to exclude the theory that the explosion resulted from any negligence on the part of the students. This involved proof to the effect that there was no gas in any of the iron boxes containing the pet-cocks at the time of the occurrence, but that the explosion was from gas confined somewhere else between the floors. The plaintiff’s proof of negligence depended entirely upon circumstantial evidence, including an application of the maxim res ipsa loquitur. The defendant contended that the explosion was attributable to some other cause as probable as the fact of its negligence, and sought to sustain the contention by evidence. The defendant submitted also direct evidence to show that it was not negligent. It will be seen from the above statement that if an inference of the defendant’s negligence was legally authorized under the circumstantial evidence and was not conclusively rebutted by the direct testimony, the verdict should stand, but that if either *754of these propositions should be decided in favor of the defendant, the verdict must be set aside.
In our view of the case it is unnecessary to discuss the evidence by which the plaintiff claims to have shown the defendant’s negligence. At most it was only circumstantial and subject to rebuttal, and in our opinion, if a prima facie case was ever made out, it was rebutted as a matter of law. “When the plaintiff’s right to recover depended upon the establishment of a particular fact, and the only proof offered for this purpose was circumstantial evidence from which the existence of such fact might be inferred, but which did not demand a finding to that effect, a recovery by the plaintiff was not lawful, when, by the positive and uncontradicted testimony of unimpeached witnesses, which was perfectly consistent with the circumstantial evidence relied on by the plaintiff, it was affirmatively shown that no such fact existed.” Frazier v. Ga. R. Co., 108 Ga. 807 (33 S. E. 993). “A fact can not be established by circumstantial evidence which is perfectly consistent with direct, uncontradicted, reasonable and unimpeached testimony that the fact does not exist.” Neill v. Hill, 32 Ga. App. 381 (2 5) (123 S. E. 30). If the direct evidence established, without dispute, that the defendant was not negligent, the verdict in the plaintiff’s, favor is unlawful although the defendant has not satisfactorily accounted for the occurrence. Palmer Bride Co. v. Chenall, 119 Ga. 837 (4) (47 S. E. 329). The defendant proved by Professor Peebles that soon after the explosion the wooden floors in both rooms were entirely removed, so that the pipes could be tested and examined in every part; that they were tested and were found to be everywhere free from defects, except inside three of the iron boxes described above, in the area of the explosion, “right in the center” of it. Prom all the testimony it appears that these defects did not exist before the explosion but were caused by it. It will not do to say that these three defects probably existed before the explosion. If the plaintiff ever excluded the theory of negligence on the part of the students or others than the defendant in allowing gas to escape through the pet-cocks in these boxes, he at the same time showed that there were no defects in the pipes or cocks within the boxes before the explosion. Tinder the situation and all the circumstances, the exclusion of one of these facts equally excluded the other. The plaintiff’s conten*755tion was that the gas was confined between the floors somewhere outside the boxes, and that no gas was in the boxes. If none was ever in them before the explosion, then the defects found after the explosion can be explained upon no other theory than that they were caused by the explosion. See, in this connection, Ga. R. Co. v. Roberts, 114 Ga. 387 (40 S. E. 264); Carroll v. Atlanta Paper Co., 7 Ga. App. 584 (67 S. E. 680); Ga. Ry. &c. Co. v. Harris, 1 Ga. App. 714 (57 S. E. 1076).
The explosion occurred in March, 1921. The trial was in September, 1925. Immediately after the explosion and the tearing up of the old floors new floors were laid, and from that date until the trial the pipes had been constantly used without having been repaired, except at the three places mentioned, and nothing had occurred in the meantime to show that the tests and examinations, as made by Professor Peebles, were not thorough and exhaustive. The testimony of this witness was consistent with the circumstantial evidence, and, not being subject to discredit on any ground, was a cpmplete refutation of any possible inference that might have been deducible from the circumstantial evidence. It follows that the verdict of the jury is illegal and that the court erred in refusing a new trial.
Judgment reversed.
Jenkins, P. J., and Stephens, J., concur.