MacIntyre, J.
The petition of John H. Gentle, administrator of the estate of Albert Evans, deceased, against the Western & Atlantic Railroad made substantially the following case: “On August 9, 1936, and between one and two o’clock a. m. . . Albert Evans (who will be hereafter referred to as the decedent) was employed by the defendant as a brakeman on one of its freight-trains which was operating on this occasion from Atlanta, Georgia, to Chattanooga, Tennessee.” The defendant in operating this train, and the deceased in performing his duties in connection therewith, were engaged in interstate commerce. “Said freight-train upon which decedent was working as a brakeman was what is known as a pick-up freight, placing cars at various points between Atlanta and Chattanooga, and picking cars up at various points between said cities and carrying freight and cars between said cities.” When this train reached a point somewhat less than two miles north of Cartersville and near a place known as Junta, about the time stated above, it “stopped on the main line in order to place a car on a siding, and in order to take a car from the siding and place’ it in the train.” At this point the railroad runs generally north and south, and the direction toward Chattanooga is referred to as north and the direction toward Cartersville is referred to as south. “When said freight-train stopped, . . it was broken at the south end of the car to be left upon the siding. After *284the train was broken at this point, the north section of the train, consisting of the engine and two cars, proceeded north up the track and passed the switch which led to the siding upon which the car was to be left. The north section of the train, after said switch was thrown, backed into the siding and coupled the car to be left on the siding to the north end of the car which was to be removed from the siding. The north section of the train then pulled out on to the track on which the rest of the freight-train was standing, and, after the switch was thrown, backed down toward the south section of the train which was standing with the brakes on. The north section of the train backed down upon the south section, and the two sections came together with the south end of the car which had been removed from the siding contacting and running against and into the coupling on the first car of the south section with sufficient force to have caused the cars to couple by impact.” The track at this point is slightly down grade toward the north. “After said two sections of said train had come together and made contact, the train was broken at the south end of the car to be left upon the siding and at the north end of the car which had just been removed from the siding and placed in the train.” The north section of the train, consisting of the engine and two cars, then proceeded back up the track past the switch which led into the siding. The decedent threw the switch,- and the north section started to back into the siding. At this time the decedent was riding upon the south car, and was in a position and place in which his duties required him to be. Before the south ear cleared the main line, the car which had been removed from the siding and placed in the train on the main line moved down from the south section of the train and rolled down upon the north section, and in and because of said collision decedent was so crushed and mangled that he died a short time thereafter. “Said car which became separated from the south section of said train, and which rolled' down, upon, and against the forward section of said train, was equipped with automatic couplers, which when in proper condition operate by impact and couple when the coupler of one car comes in contact with the coupler of another car. The automatic coupler on the south end of said car which became detached and which had been removed from the siding was defective in that it'failed to couple by impact when the two see*285tions of said train came together after said car had been removed from the siding. . . The failure of said car which became detached to couple by impact resulted from a defective coupling appliance, and said failure of said coupler to operate by impact was a violation of the Federal safety-appliance act and directly and proximately contributed to and caused the injury and death of the decedent.”
The petition contained two counts. The second count made substantially the case as above narrated, except that it charged that the automatic coupling of the front car of the south section of the train, and the car which was backed into the same and which separated therefrom, causing the injury and death of decedent, were both defective, and would not couple automatically by impact. The allegations concerning the manner in which the deceased met his death and the previous switching movements of the train were fully sustained by the evidence. Dudley Barton, the engineer of the train, testified as a witness for the plaintiff, in substance, that the couplers on the cars in question were what were known as automatic couplers; that his movement backward to couple the car taken from the side-track (coal-car) to the north ear of the rest.of the train “was the ordinary movement in making a .coupling;” that he could tell when the impact was made — -could feel the vibration of it; that before pulling the coal-car from the sidetrack the deceased went to the south end of it, “and I suppose fixed the coupling where he thought it would couple.” He further testified: “Even if couplers are in good condition, they do not always couple on the first impact. It happens pretty often that you have to make one or two or three trials before they couple. I don’t guess a brakeman even can tell how hard to hit to make a coupling, except to try it. It is just merely, with the engineer, it is just a guess, to try it and see if it makes, and then if it don’t make to try it again, either hit it harder or easier — lighter, whatever they signal you to do back there. After you try once and it don’t make, the brakeman will signal you either to hit it harder or to hit easier, until the coupling finally makes. . . I have frequently had them where they don’t couple the first time, plenty of times. When I told Mr. Gambrell [attorney for the plaintiff] I came back in the usual manner and speed, I meant by that that I came back just as Mr. Evans (the deceased) . . signaled me *286to come back up there. . . I couldn’t tell whether I had hit it hard enough or too hard or not hard enough, I couldn’t tell at all.”
Gus Thomas, a former railroad employee who had had experience in switching trains, testified, as a witness for the plaintiff, that when the automatic couplers are in proper position to couple, and the cars come together at a speed of from two to six miles per hour, “they are supposed to couple automatically by impact. If they are in condition they will couple automatically. If they do not couple automatically, that would indicate they are out of order. . . The ordinary speed that a car is backed into another car for the purpose of coupling is about from two to six miles an hour. . . If it failed to couple one time, that is an indication that it is out of order. If it is in order, it will couple; if a coupling is made at too great a speed, from ten to fifteen miles per hour, they might not couple; if they did not come together with sufficient force they would not couple. . . ’“I did not work out, while I was in service, any formula or any definite rate of speed that I had to go to make couplings. Some cars, in coupling, are harder to couple than others. Some you would hit them harder and some you wouldn’t. That would be indicative that some would be in good condition and some wouldn’t. . . In my six years of experience in switching for the Southern Railway, I have had plenty of couplers to fail to couple. All of them that failed were due to a defect. . . In switching in the yards, there are a multiplicity of different situations that arise in connection with everyday switching. Under one situation it is necessary, in order to make a coupling, to bring the cars together by impact with more force than under another situation.” Q. “And the switchman, be he ever so perfect, he never does know what force is necessary to make a coupling, does he?” A. “No, sir. . . In my six years experience as a switchman in the Southern yards in Atlanta, I have failed to bring two cars together by impact with sufficient force to couple them. I have had that experience. . . In that case I would back up and come back again, and they would come together. . . I would come back with a lot more force, and they would couple the next time. I have had the experience, in my switching in the yards of Atlanta, of two cars coming together by impact with such severe force that they did not couple. That *287would be true even though the couplers were in perfect condition. . . Every time that I failed to make a coupling I did not make an investigation to see why that ear did not couple. A lot of times we would just back up and hit the car again, and go ahead, and it would couple the second time. I couldn’t say if it was a defective coupler. I don’t think I said yesterday, if two cars didn’t couple on the first trial, that that was due to a defect in the coupler; I said they should couple by the first trial. It is a defective coupler when they don’t couple. . . If a coupler is defective, that defect can be discovered upon examination of the coupler. If two cars come together by impact, and they fail to couple, and a man who knows his business goes- and examines that coupler, he can tell whether or not anything is wrong with the coupler.”
J. E. Head, the fireman on the train, testified, as a witness for the defendant, concerning the coupling of the coal-car to that part of the train standing on the main line, as follows: “His [engineer’s] speed in backing up was the ordinary speed for making a coupling. The cars hit the train. I know that. I felt it and heard the jar. I heard the sound and felt the jar.” He further testified, concerning automatic couplers: “Sometimes they will couple by impact and sometimes they won’t. . . I felt the coal-car and the north end of the train come together. I don’t undertake to say how hard or how light it hit. I don’t know whether there was enough force used to make the coupling or not. I don’t' know whether too much force was used or not. It is a fact that, with couplers in perfect condition, if not enough force is used they don’t couple, or if too much force is used they don’t couple. I had no way of knowing that night whether or not the proper amount of force was used to make the coupling. It was dark that night; it is not possible on a dark night like that for a person on the engine to know whether or not he has used the proper amount 'of force to make a coupling, without making a test. . . It is- a fact that, with couplers in perfect condition, fifteen or twenty per cent, of the time they don’t couple on the first trial. You often have to make a second or third trial, even with the couplers in good condition. . . Out of my experience I know that couplers in good condition often don’t couple on the first trial. . . After the accident was over and we got ready to leave Junta, the engine was coupled into this car that stood on the north end of the train. *288I suppose that coupler worked all right then, I don’t remember of making no test nor nothing. We didn’t have any trouble with that coupler on our trip that I know anything about. . . The fact that a coupling doesn’t make on the first trial, that is no indication that it is out of condition; we frequently have to try two or three times. Couplers won’t couple always on the first trial, and them be in first-class condition.”
The defendant introduced several experienced railroad men who testified substantially that it frequently occurs that with couplers in perfect condition and in proper position they do not couple upon first impact. As one of these witnesses explained, as to coupling couplers in good condition: “I would make the same coupling once that would not make the second time in the same manner. In other words, I may hit the car with the same speed and I would make it, and with the same speed that would couple the first time on the second time would not couple. I have had experience of couplers coming together by impact and failing to couple, and immediately thereafter made an examination of that coupler and found it to be in perfect condition.” It further appears from the testimony of all the witnesses that no definite speed has ever been determined at which automatic couplers will couple under all conditions. The remaining material facts will hereinafter appear.
This action was brought under the terms of the Federal employer’s liability act, which provides, in so far as material here, that “Every common carrier by railroad while engaged in commerce between any of the several States . . shall be liable in damages to any person suffering injury while he is employed by such carrier in such commerce, or, in case of the death of such employee, to his or her personal representative, for the benefit of the surviving widow . . and children of such employee, for such injury or death resulting in whole or in part . . by reason of any defect or insufficiency, due to its negligence, in its cars, engines, appliances, machinery, track, road-bed, works, boats, wharves, or other equipment.” Title 45, c. 2, § 51 (Apr. 22, 1908, c. 149, § 1, 35 Stat. 65). The petition alleges a violation by the defendant of the safety-appliance act. The section upon which this violation is based is in terms as follows: “It. shall be unlawful for any common carrier engaged in interstate commerce *289by railroad to haul or permit to be hauled or used on its line any car used in moving interstate traffic, not equipped with couplers coupling automatically by impact, and which can be uncoupled without the necessity of men being between the ends of the cars.” 45 U. S. C. A. c. 1, § 3 (March 3, 1893, c. 196, § 2, 27 Stat. 531). A violation of this section constitutes negligence per se under the provisions of the Federal employer’s liability act above set out. San Antonio &c. Ry. Co. v. Wagner, 241 U. S. 476 (36 Sup. Ct. 626, 60 L. ed. 1110). Concerning the purpose and scope of this section it has been said: '“It is very plain that the evils against which these provisions are directed are those which attend the old-fashioned link and pin couplings,, where it was necessary for men to go between the ends of the cars to couple and uncouple them, and where the cars, when coupled into a train, sometimes separated by reason of the insecurity of the coupling. In Johnson v. Southern Pac. Co. (Utah 1904), 196 U. S. 1 (25 Sup. Ct. 158, 49 L. ed. 363, 370, 17 Am. Neg. R. 412), this court said of the provisions for automatic couplers that 'the risk in coupling and uncoupling was the evil sought to be remedied.’” St. Louis &c. R. Co. v. Conarty, 238 U. S. 243 (35 Sup. Ct. 785, 59 L. ed. 1290). See also Buschalewski v. N. Y. Central Ry. Co., 105 Misc. 541 (173 N. Y. Supp. 506); U. S. v. Philadelphia &c. R. Co., 223 Fed. 215; San Antonio &c. Ry. Co. v. Wagner (Tex. Civ. App., 1914), 166 S. W. 24; affirmed, 241 U. S. 476, supra. This act is designed for the protection not only of employees who are injured while going between the cars to operate a coupler which is not automatic, or an automatic coupler in a defective condition, but to any employee who suffers injury proximately resulting from a failure to obey this section of the safety-appliance law. Louisville & Nashville R. Co. v. Layton (1917), 243 U. S. 617 (37 Sup. Ct. 456, 61 L. ed. 931), affirming the decision of the Supreme Court of this State, 145 Ga. 886 (90 S. E. 53). It follows that if the evidence shows that the deceased was killed by reason of a violation of this act by the defendant, a recovery should be allowed.
What is the duty placed on a railroad by this section? The authorities are that by this section it is made unlawful for any common carrier engaged in interstate commerce by railroad to haul on its line any car not equipped with automatic couplers capable of being coupled and uncoupled without the necessity of *290a man going between the ends of the ears. The liability for failure to obey this provision is absolute, and not dependent on lack of reasonable care. St. Louis &c. Ry. Co. v. Taylor, 210 U. S. 281 (28 Sup. Ct. 616, 52 L. ed. 1061); Texas &c. Ry. Co. v. Rigsby, 241 U. S. 33 (36 Sup. Ct. 482, 60 L. ed. 874); Great Northern R. Co. v. Otos, 239 U. S. 349 (36 Sup. Ct. 124, 60 L. ed. 322). Thus, if a railroad should fail to equip its cars with “couplers coupling automatically upon impact,” as described in the statute, or, after having done so, should fail to keep them in proper working order at all times, this would be a plain violation of the statute. The failure of couplers to couple is not per se a violation of the act. That on a given occasion couplers failed to couple upon impact does not disrobe them of their character of “couplers coupling automatically upon impact,” as we construe those terms to be used in this act. By the use of the above term, we do not apprehend that it was the intention of Congress to require railroads to equip their cars with couplers designed to couple automatically upon impact, without the necessity of men going between the ends of the cars, which would, upon all occasions and under all circumstances, actually couple, without fail, upon impact. The railroad lias complied with the law when it equips its cars with couplers designed to couple automatically upon impact without the necessity of men going between the ends of the cars, which couplers are of general use among well-regulated railroads; and unless the failure of the couplers to couple on a given occasion results from some defect therein, the railroad can not be held responsible for injuries resulting proximately therefrom. We think that this is apparent from the mass of adjudications under this section. To give the act any other construction would result in absurdities plainly not within the contemplation of Congress in the passage of this act. Literal interpretation of statutes producing absurd consequences has often been condemned by the Supreme Court. See Sorrells v. U. S., 287 U. S. 435, 446 (53 Sup. Ct. 210, 77 L. ed. 413, 86 A. L. R. 249). In this connection, it was said in St. Louis Southwestern Ry. Co. v. Bounds (Texas Civ. App. 1922), 244 S. W. 1099: “Neither the language nor the purpose of the statute requires the equipment of cars with appliances that will operate with unfailing precision on every occasion. Such a degree of perfection is not essential to the safety of the employee *291for whose protection the equipment is required. It may be expected from the very construction of such devices that in the course of time, and under the unavoidable varying conditions incident to railway traffic there will be occasions when more than one impact may be required in order to effect a coupling. . . It would manifestly be unfair to hold that the carrier had violated the statute until the inefficiency of the device had been disclosed by some reasonable test that would justify the conclusion that it was defective.” See also Chesapeake & Ohio Ry. Co. v. Charlton, 247 Fed. 23. It appears without dispute that the couplers in question were what is known as American Eailway Association D-type couplers, which are standard automatic couplers in general use by well-regulated railroads throughout the United States. This type coupler was the result of the efforts of a committee composed of experts representing different coupler manufacturers, appointed by the American Eailway Association, to study and recommend a standard coupler, combining the best features of all couplers manufactured and used before that time.
There is no question that the defendant had equipped the cars in question with the type of couplers required by the above section of the safety-appliance act; and the only question that remains to be answered is, were they defective? The burden of proof in this regard rested upon the plaintiff. In many of the adjudicated cases we find it stated that "the failure of a coupler to work at any time sustains a charge that the act has been violated.” See Philadelphia &c. Ry. Co. v. Auchenbach, 16 Fed. (3d) 550, and cit. We seriously question the general applicability of this rule to all cases, especially to a case of the present character, where it appears from all the evidence that inferences other than that the couplers were defective are authorized from their failure to couple upon first impact. In no case called to our attention by the very exhaustive and able briefs filed by counsel for both parties, or any that our research has revealed, has it been expressly ruled that the mere showing that upon one occasion couplers did actually fail to couple authorized a finding that the couplers were defective, though it may be inferable therefrom that such be the case. Each of the cases examined (and we have diligently tried to examine all of them) presents a stronger state of facts than those above stated. In San Antonio &c. Ry. Co. v. Wagner, supra, it appeared from *292the evidence that the drawbar on the engine was out of line, that the coupling-pin on the box-car failed to drop as it should have done at the first impact, that it required manipulation in preparation for the second impact, and that the drawbar on the, engine was so far out of line as to require adjustment in preparation for the second impact. The Supreme Court held that this, together with the opinion evidence that the coupler was defective, was sufficient to sustain a finding that the equipment was defective. It is at once apparent that this case is materially different from the one at bar. In that case the Supreme Court said: “We do not in this case determine, what was conceded in Chicago R. I. & Pac. Ry. v. Brown, 229 U. S. 317, 320 (33 Sup. Ct. 840, 57 L. ed. 1204), that the failure of a coupler to work at any time sustains a charge that the act has been violated.” In the Brown case, there referred to, the only question determined by the court was whether or not a switchman who reached in between slowly-moving cars to remove a coupling-pin after repeated unsuccessful attempts to operate the automatic coupler, was, as a matter of law, guilty of contributory negligence such as would defeat his right to recover; it being there conceded by counsel for the railroad that in “the Taylor case, 210 U. S. 281, 52 L. ed. 1061, 28 Sup. Ct. Rep. 616 and in Chicago, B. & O. Ry. Co. v. U. S., 220 U. S. 559, 55 L. ed. 582, 31 Sup. Ct. Rep. 612, this court settled that the failure of a coupler to work at any time sustains a charge of negligence in this- respect, no matter how slight the pull on the coupling lever.” It is to be observed that this concession was made in a case where a coupling-lever failed to .work, and not where merely the couplers failed to couple.
Neither the Taylor case nor the Chicago &c. Ry. Co. case, supra, is similar to the present ease. The Taylor case was a suit for the death of an employee who was crushed between two cars while attempting, in the discharge of his duty, to couple them. The present act was not in force at that time. The suit was based on a violation of an act requiring that all ears be equipped with draw-bars of uniform height. The court in construing that act said that, it “requires that the center of the drawbars of freight-cars iised on stand ard-guage railroads shall be, when the cars are empty, 34-1/2 inches above the level of the tops of the rails; that it permits, when a car is partly or fully loaded, a variation in the *293height downward, in no case to exceed 3 inches. . . If, when unloaded, its drawbars are of greater or less height than the standard prescribed by law, or if, when wholly or partially loaded, its drawbars are lowered more than the maximum variation permitted, the car does not comply with the requirements of the law.” It appeared from the evidence of the plaintiff that the automatic coupler on one of the cars was about four inches lower than the link and pin coupler, which, if true, disclosed a violation of the act as there construed. The court held that it was for the jury to determine between this evidence and the evidence for the defendant showing an allowable variation. In the Chicago &c. Ry. Co. case, which was a suit for the recovery of penalties for a violation of the safety-appliance act, including the one now in question, the court merely considered the question whether the act prescribed an absolute duty, or whether the act was complied with if the defendant, having equipped its train with proper appliances, exercised reasonable care and diligence in keeping them in repair. It seems to have been there conceded, as in the circuit court, that “there was evidence tending to prove the defective condition of each of the four ears as charged.” This evidence was not set out or discussed. In the Auchenbach case, supra, it appeared, that two sections of the train moved together and did not couple. The brakeman discovered this, and went between the cars. He worked the coupler in and out for three minutes, and the pin would not drop. The section failed to couple on second impact. Similar facts appeared in the Brown case, supra. In Yazoo &c. Ry. Co. v. Cockerham, 134 Miss. 887 (99 So. 14), it appeared that the decedent was attempting to couple a ear to the engine. A nail was used as a cotter pin. After failing to couple,' the decedent went between the cars to fix the coupler, and when the engine backed against the car he was killed. All the evidence disclosed that a bent nail was used instead of a cotter pin. There was evidence that the coupling could not be made in the ordinary way. Several efforts were made. There was also evidence that the bent nail protruded, touching the drawhead so that the pin could not fall. In Chicago &c. Ry. Co. v. Linehan, 66 Fed. (2d) 373, it appeared that the plaintiff was switching foreman. There were four cars on a house track, not coupled. The purpose was to couple the engine to car 1, then to car 2, then to car 3, then to *294car 4, and then remove all four from this track. The plaintiff adjusted the coupler on car 4, and then on car 3. When the engine and cars 1 and 2 moved against car 3, the cars did not couple. He then noticed that the knuckle which he had opened on car 3 was closed by the impact. He attempted to work the knuckle with the lever extending to the edge of the car, provided for this purpose, and it failed to work. He thereupon went between the cars to adjust the same, and was injured.
We have examined the following cases, and find them to be distinguishable from the facts of the present case: In re Sorenson Drainage Ditch, 27 S. D. 342 (131 N. W. 300); Montgomery v. Carolina &c. Ry. Co., 163 N. C. 597 (80 S. E. 83); Parker v. Atlantic City Ry. Co., 87 N. J. L. 148 (93 Atl. 574); Northcutt v. Davis, 113 Kansas, 444 (214 Pac. 1113); Holtz v. Chicago &c. Ry. Co., 176 Minn. 575 (224 N. W. 241); O’Donnell v. B. & O. Ry. Co. 324 Mo. 1097 (26 S. W. (2d) 929); Stewart v. Wabash Ry. Co., 105 Neb. 812 (182 N. W. 496); Gillum v. Pacific Coast Ry. Co., 152 Wash. 657 (279 Pac. 114); Fletcher v. S. D. Cent. Ry. Co., 36 S. D. 401 (155 N. W. 3); St. Louis &c. Ry. Co. v. McWhirter, 145 Ky. 427 (140 S. W. 672); Clark v. Erie Ry. Co., 230 Fed. 478; Smith v. Atlantic Coast Line R. Co., 210 Fed. 761; Pa. Ry. Co. v. Jones, 300 Fed. 525; Overstreet v. Norfolk &c. Ry. Co., 238 Fed. 565; Chesapeake & Ohio Ry. Co. v. Charlton, 247 Fed. 34; Philadelphia &c. Ry. Co. v. McKibbin, 259 Fed. 476; Didinger v. Pa. Ry. Co., 39 Fed. (2d) 798. The cases fall generally within one of three categories: (1) where there was direct proof that some part of the coupler failed to properly function, such as the lever extending to the edge of a car, for the purpose of placing the coupler in position to couple; (2) where there was direct proof of some defect in the coupler; and (3) where the couplers separated while the train was in motion. We are in doubt whether such cases are analogous to the facts of the present case. It is the general rule that where proved facts give equal support to each of two inconsistent inferences, neither of them is established, and that judgment must as a matter of law go against the party upon whom rests the necessity of sustaining one of these inferences as against the other before he is entitled to recover. This is true because “where a plaintiff in a civil case supports his action solely by circumstantial evidence, before he is authorized *295to have a verdict in his favor the testimony must be such as to reasonably establish the theory relied upon, and to preponderate to that theory, rather than to any other reasonable hypothesis.” Ga. Ry. & El. Co. v. Harris, 1 Ga. App. 714 (57 S. E. 1076); Penn. Ry. Co. v. Chamberlain, 288 U. S. 333 (53 Sup. Ct. 391, 77 L. ed. 819), and cit.; Smith v. First National Bank, 99 Mass. 605, 611-612 (97 Am. D. 59). The evidence in the present case, as a whole, we believe shows that several inferences may arise from the failure of a car to couple to another: (1) that the couplers were defective; (2) that the impact was not sufficient under the particular circumstances; or (3) that the impact was too great under the particular circumstances. It must be conceded that such a showing does not demand a finding that the couplers were defective. It, at best, merely raises an inference of this fact. New Orleans &c. Ry. Co. v. Scarlet, 249 U. S. 528 (39 Sup. Ct. 369, 63 L. ed. 752); Minneapolis &c. Ry. Co. v. Gotsehall, 244 U. S. 66 (37 Sup. Ct. 598, 61 L. ed. 995). Whether or not the inference that the couplers were defective so preponderates as to make out a prima facie ease upon such showing we will not decide, for the reason that we are of the opinion that, conceding this to be true, the evidence produced by the defendant as to the condition of these couplers a short time after the time in question demands a finding that the couplers were not defective.
It appears without contradiction in the evidence that the coupler on the north end of the car to which the coal-car failed to couple, up to the point in question, worked properly. Immediately after the accident the engine was coupled thereto without difficulty. On the same morning (Sunday) this car was placed on a siding at Kingston and was taken to Rome on the following day (Monday). On Tuesday morning the coupler in question was given a thorough inspection, and it was found to be in good working condition. It was also shown that between the point of the accident and Rome there was no place that the coupler, if defective, could have been repaired, and it is purely conjectural that it could have been repaired on the track by an employee of the defendant. On the same morning of the accident an inspector went to the scene of the accident and found the Rhode Island car (the car that was to be placed on the siding) and the coal-car coupled together on the side-track. He gave the coupler on the south end of the coal-*296car (the one in question) a thorough inspection, and found it to be in good working order. The next morning a switch-engine was coupled to this end of the coal-car, and these cars were moved to a delivery-track thirty or forty car-lengths distant. The coupler worked perfectly, both in coupling and uncoupling to and from the engine. The next day the general car foreman of the defendant came from Atlanta, and he and the inspector above mentioned again examined 'the coupler on the coal-car thoroughly. They took it apart and examined every part, and found it in good working order. What amount of evidence could be more conclusive than this, that the couplers were not defective? Such showing, as against a mere inference upon which the plaintiff’s case precariously hangs, to our minds authorizes no other finding but that the couplers in question were not defective. The testimony of Gus Thomas, already set out in part, does not, when taken as a whole, conflict with the testimony of the witnesses that the couplers in question were not defective. While in several parts of his testimony he stated that where couplers failed to couple upon impact, this indicated that they were defective, for “if they are in condition they will couple automatically,” he further testified that because of the multiplicity of different situations which arise in coupling cars it is necessary at times to bring cars together with greater or less force than at other times, and that there is no way to determine, under given situations, as to how much force should be used in order for the couplers to work automatically. In no part of his testimony did he attempt to testify that, under the situation and circumstances of the attempted coupling of the cars, immediately before the death of the deceased, the failure of the couplers to couple by the impact there made showed that the couplers were out of order. Moreover, his testimony was but an opinion which we think must yield to the proved facts; that is, that the couplers failed upon this occasion to couple, and that upon inspection they were found not to be defective. The witness did not inspect the couplers in question, and was not testifying that in his opinion, after such inspection, he believed them to be defective. Atlanta & Charlotte Air-Line Ry. v. Gravitt, 93 Ga. 369 (20 S. E. 550, 26 L. R. A. 553, 44 Am. St. R. 145).
In Neill v. Hill, 32 Ga. App. 381 (123 S. E. 30), it was said: “A fact can not be established by circumstantial evidence which *297is perfectly consistent with direct, uncontradicted, reasonable, and unimpeached testimony that the fact does not exist. Frazier v. Ga. R. &c. Co., 108 Ga. 807 (33 S. E. 996); Hendon v. State, 10 Ga. App. 78 (72 S. E. 522); Ga. Ry. & El. Co. v. Harris, 1 Ga. App. 714, 717 (57 S. E. 1076).” In Frazier v. Ga. R. &c. Co., supra, it was said: “When a plaintiffs 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.” See also Emory University v. Bliss, 35 Ga. App. 752 (134 S. E. 637); Palmer Bride Co. v. Chenall, 119 Ga. 837 (47 S. E. 329); Penn. Ry. Co. v. Chamberlain, supra. Compare Gainesville, Jefferson &c. Ry. Co. v. Edmondson, 101 Ga. 747 (2) (29 S. E. 213); So. Ry. Co. v. Myers, 108 Ga. 165 (33 S. E. 917). It is to be said that it is not unthinkable that a coupler may be defective, though upon inspection no defect can be discovered. A coupler may fail to couple so many times, or so frequently, as to authorize no other inference, even in the face of evidence that no defect was discoverable therein upon inspection. Such, however, is not the present ease. The plaintiff proves only one failure. The defendant proves success before and after the unfortunate occurrence. This court is always reluctant to upset the verdict of a jury upon the general grounds; but we are convinced, after a careful consideration of the evidence and the. law applicable thereto, that a verdict for the defendant was demanded. The judge erred in overruling the motion for new trial.
Judgment reversed.
Guerry, J., concurs. Broyles, G. J., disqualified. Stephens, P. J., dissents.