delivered the opinion of the court.
This is an action brought by Rufus S. Tanner, plaintiff below, against The Chesapeake and Ohio Railway Company, to recover damages for personal injuries resulting *408from the partial derailment of an east-bound coal train of the defendant company at Big Island, Virginia, on the morning of January 11, 1932. There were three trials of the case in the court below, the first two of which resulted in a disagreement of the jury. At the third trial the jury rendered a verdict in favor of the plaintiff for the sum of $25,000, and the court having entered judgment thereon, the same has been brought before this court for review.
The defendant’s main line track in the vicinity of Big Island runs in a general east and west direction on the south bank of the James river. On a narrow strip of land between the defendant’s right of way and the river, the Bedford Pulp and Paper Company maintains a paper mill consisting of a number of buildings, some of which are connected. Approaching the scene of the accident from the west, the defendant’s main line track is straight as it passes the new Big Island depot. About three hundred feet east of the depot building the track crosses Hunting creek, and a short distance from the creek begins a three degree twenty-two minute curve to the right. A little further east from Hunting creek is located the switch to a side track leading out on the north side of the main line which serves the paper mill. After passing the switch to the mill siding, the main track again becomes straight or tangent and remains so for over a thousand feet.
At the time of the derailment there were four box cars on the mill siding, and the plaintiff, an employee of the paper company, was inside one of these cars supervising the unloading of a shipment of pulp wood. The train involved in the accident, consisting of a locomotive and 128 cars loaded with coal with a caboose on the rear, was nearly a mile long. After passing the switch point referred to, and while the front of the train was passing the mill buildings, which extend along the river for eight hundred feet or more, the twelfth car from the locomotive completely derailed approximately six hundred feet east of the switch, breaking loose from the cars ahead and plunging to the north of the main line about opposite the *409cars standing on the mill siding. The twenty-nine cars in the train immediately following the twelfth car also completely derailed, in consequence of which the cars on the mill siding were struck by the derailed cars, and the plaintiff was thrown out of the car in which he was working and severely injured. Two other men in the car were also injured at the same time and considerable damage was done to the mill buildings. The main track was also torn up for some distance east of where the thirty cars left the rails. The train locomotive and first eleven cars were not derailed but proceeded along the track until brought to a stop by the automatic application of air brakes. All the other cars in the train behind the thirty derailed cars likewise remained on the track and were stopped by the same means; several of such cars having passed the switch in question.
None of the train crew had any knowledge of the derailment until the respective parts of the train in which they were riding had been stopped by the automatic brakes; and it appears that no one actually witnessed the initial derailment—that is, when the twelfth car first left the rails. Therefore, the evidence on this subject is entirely circumstantial.
It is the contention of the plaintiff below that the derailment of the cars and plaintiff’s injuries were the proximate result of the railroad company’s negligence in the operation of the train and the construction and maintenance of its track, and he alleges numerous particulars in which he claims the track was defective and the train negligently operated. For a clear understanding of the plaintiff’s case it is necessary to take up separately each of the circumstances upon which the plaintiff relies to show negligence on the part of the defendant.
(a). Plaintiff produced evidence to show that there was a straight rail eighteen feet long located in the track on the outside of the curve above referred to, the eastern end of which rail is seventeen and one-half feet west of the mill switch, and that the proper curvature of the rail *410should have been nine-thirty-seconds of an inch at its middle ordinate.
There is a conflict in testimony among plaintiff’s own witnesses as to whether the rail in question was straight or slightly curved, but all of them testify that the only effect of such a straight rail would be to cause slight stress and strain on the track structure. None of these witnesses expressed the opinion that the short rail caused the derailment of the train, or that it was dangerous to operate the train over it.
To the contrary, it is shown by undisputed evidence that this rail had been in the curve of the track for several years prior to the accident and had remained in the same position from the time of the accident until the last trial of the case, nearly two years afterwards, without causing any trouble in the operation of the trains over the same.
(b) . It is claimed that the end of the switch point in question was one-sixteenth of an inch higher at the point than the stock rail, and was level with the stock rail for two-thirds of the length of it, but plaintiff’s witnesses also failed to agree as to this particular. A close examination of the plaintiff’s evidence on this point, discarding the testimony introduced in behalf of the defendant, fails to show that the alleged elevation of one-sixteenth of an inch of the switch point could have caused the derailment or that any real significance was to be attached thereto. On the other hand, it was shown by disinterested expert witnesses of the defendant that the only effect that would result from an elevation of one-sixteenth of an inch in the switch point was that a train traveling east might either chip the point or mash it down.
(c) . It is. claimed that there was looseness, or play, in the switch which might cause a split switch and produce a derailment. On this point also the evidence is in direct conflict, but it was agreed by all the expert witnesses on both sides that if there had been a split switch it would have caused an immediate derailment of the train and the cars to pile up at and near the switch. There was, *411however, not only no evidence produced that there was a split switch on the occasion in question, but the fact that the complete derailment of the twelfth car took place six hundred feet east of the switch and all cars behind it safely passed over the switch conclusively proves the contrary. This being the case, even if there was looseness in the switch point, that circumstance could not have been the cause of the accident.
(d) . It is testified by a witness for the plaintiff that he saw water and mud forced up by the work train in spots on the south side of the switch and east of the frog when the train was cleaning up the wreck, but it was testified by plaintiff’s chief expert witness, Mr. Barnard, that he did not think that the condition of the ballast had anything to do with the derailment one way or the other, and that the ballast was ideal in its condition. It also appears from undisputed testimony that the track at the switch point was well drained and that the ballast was thirty-three inches deep.
(e) . It is next claimed that there was rotten ties near the switch. Plaintiff’s witness, Foster, testified that he found a decayed tie fifty-five feet west of the switch when he examined the track in February, 1932, and Mr. Barnard testified that when he examined the track about a month after the accident, he found a number of ties in a bad state of preservation, some badly warped, and some cracked and open at the ends, and showed signs of decay. There is not, however, any evidenec in the case that the condition of the ties anywhere in the vicinity of the accident rendered the track unsafe, hut there is affirmative and uncontradicted evidence to the contrary. It is shown by the defendant that none of these ties had been replaced from the time of the accident until the time of the last trial of the case, nearly two years afterwards, and during such time defendant’s trains had been safely operated over same, which would seem to show conclusively that the condition of the ties, in no event, could have been remotely connected with the derailment of the train.
*412(f) . It is next claimed that the switch in question was dangerous because it was located on the outside of the curve. Mr. King, one of plaintiff’s expert witnesses, testified that “when it is necessary, as perhaps it was in this case,” to put a switch on the outside of a curve then, from the standpoint of safety and good engineering, he would say that everything about that switch should always be kept in the best of condition and carefully inspected.
That it was necessary to locate the switch in question on the outside of the curve in order to serve the plant of the paper mill seems to be admitted. It also seems to be conceded that the only danger existing from the location of such a switch would be that which might result from an open or split switch, of which there is no evidence whatever in this case, but on the other hand, all the evidence is to the contrary. These facts would seem to make further discussion of this point purely academic.
(g) . As previously stated, the curve in the track, herein-before referred to, was a slight curve of three degrees and twenty-two minutes, and it is shown by the plaintiff’s evidence that the super-elevation of the' outside or north rail of the track was two and one-sixteenth inches. It is earnestly contended by the plaintiff that the speed of the train on the occasion in question in rounding this curve was dangerous, and an act of negligence on the part of the defendant company, which caused or contributed to the derailment of the train.
The evidence as to the speed of the train at the time of the accident is in direct conflict. It was testified by several witnesses of the plaintiff that in their opinion the train was running at a speed of from forty-five to fifty miles an hour. It is testified by the train crew and several disinterested witnesses that the speed of the train was between twenty-five and thirty miles per hour, and the preponderance of the evidence seems to sustain the defendant on that point. Assuming, however, that the jury had a right to find that the train was being operated at a speed of from forty-five to fifty miles an hour, and the super-eleva*413tion of the outside rail on the curve was only two and one-sixteenth inches as the plaintiff contends, the question is whether or not the evidence shows that such a rate of speed on said curve was so dangerous as to constitute negligence on the part of the railroad company, and if so, whether such negligence was the cause of plaintiff’s injuries. Much expert and technical evidence was introduced by both sides on this point which we cannot undertake to discuss in detail.
It appears from the American Railway Engineering Association Manual, which was offered in evidence by the plaintiff, that “in order to throw the center of gravity of the train in the center of the track and produce what is called equilibrium, or balanced speed” the defendant’s train should not have traveled more than thirty miles an hour around the curve in question. The same manual, however, also refers to the report of a special committee of the American Railway Engineering Association, which report was duly adopted by the association, and which defines, with speed tables, not only “equilibrium speed,” but also what is called “comfortable speed,” “safe speed” and “theoretical overturning speed.”
“Comfortable speed” is defined as a speed which has been developed by experience and observation to mean that speed at which you can run a train around a curve, and the passengers' will not feel any uncomfortable or unpleasant lurch in going around the curve. This speed for a curve such as has been described is stated to be forty-eight miles per hour.
“Safe speed” is defined as a speed at which the combined forces of the weight of the car and the centrifugal force due to the curve motion will throw the weight of the train two-thirds on the outside rail of the curve and one-third on the inside rail. This speed is stated to be sixty-two miles per hour; and “theoretical overturning speed” is stated to be ninety-eight miles per hour. It therefore seems from the published reports of the American Railway Engineering Association, which are admitted by both *414plaintiff and defendant to be reliable and authentic, that it was perfectly safe to take the curve in question at the speed of between forty-five and fifty miles per hour, at which the plaintiff’s witnesses claim the train was running at the time of the derailment.
But aside from all the technical evidence on the subject found in the record, it seems clear from the undisputed physical evidence that the speed of the train, whatever it may have been, did not cause the derailment for the reason that it is shown, as already stated, that the locomotive and eleven cars in the train passed around the curve in safety, and that the twenty-nine cars following the twelfth car did not derail until they were thrown off the track by the derailment of the twelfth car. This would seem uncontrovertible evidence that the twelfth car in the train, which caused the whole trouble, would not have left the track unless there had been something wrong with the running gear of that car.
In other words, considering all the evidence relied on by the plaintiff, we are unable to find anything to show that, whatever the condition of the track and the speed of the train may have been, the derailment was caused by any of the defects in the track or the operation of the train, as claimed by the plaintiff.
The defendant, however, undertook to show that the derailment was caused by a broken arch bar on the north or left-hand side of the rear truck of the twelfth car. The undisputed evidence is that the effect of a broken arch bar would be a gradual lowering of the end of the truck on that side until the column bolts by which the' arch bars are held in place came in contact with the cross-ties or any other object near the rail. These column bolts lie normally in the truck a little over eight inches outside of the rail, and from two to three and a fraction inches above the top level of the rail.
It is testified by the defendant’s witness, O. D. Day, the tool car foreman of the defendant company at Clifton Forge, that it was his duty to investigate derailments and *415determine the cause whenever possible, and clean up the wreck and make a detailed report of it; that upon being notified of the wreck, he got his men and equipment and immediately went down to Big Island; that after reaching the scene of the wreck he found a portion of a certain arch bar truck, consisting of a top arch bar, bottom arch bar, part of a tie strap, two column castings, two journal boxes and portions of two column bolts, all of which were hanging together. The bottom arch bar was found broken in two pieces, the break being between the column bolt hole and the bend extending to the journal box. Included in the break was an old break on the inside top of the arch bar extending one and one-half inches across the top and one-half inch into the thickness of the bar. The remainder of the break was clearly a fresh break. The two attached column bolts had been sheared off at their bottom ends. Later on they found at the wreckage the balance of this truck, including the wheels, truck bolster, and the other set of arch bars, column bolts, and column castings, all attached to each other.
Upon matching up these two findings, it was definitely ascertained that this truck was the truck of car C. & O. 22508 which, according to the undisputed evidence, was the twelfth car in the train. An examination of the truck by this witness and C. W. Maddox, chief car inspector of the C. & O. system, and J. S. Williams, master mechanic of the Clifton Forge division of the railroad, disclosed that there was a broken arch bar on the rear truck of the car C. & 0. 22508 and that there was no other broken arch bar in the wreckage.
In corroboration of the defendant’s contention that the broken arch bar of the car 22508 was the cause of the derailment of that car and the derailment of the next twenty-nine cars in the train in consequence thereof, it was shown by witnesses on both sides that at a station known as Major, which is a mile and a half west of Big Island, there is a spur track leading out on the north side of the main line, and a few minutes after the derailment there were *416found fresh metal drag marks on the frog of the spur track at about the same distance as the column bolts extended outside the north main line rail.
It is testified by the station agent, who was in the depot at the time the train passed, that the train was not scheduled to stop at Big Island, but as the front of the train went by the depot he was standing within ten feet of the train and his attention was attracted by an unusual noise, something striking against iron, and this noise frightened him to such an extent that he jumped back from the front of his office and ran out of the depot as soon as possible; that he was unable to see what caused the noise, but looking towards the front of the train he saw some car truck springs fly out from under it on the south side of the track. Another employee, who was in the freight room of the depot, also testified that he heard a loud noise of something releasing and hitting as the front of the train passed; that on coming out to see what caused the train to stop, and walking up the highway on the south side of the track, he found a coil truck spring and part of a spring plate which had not been at the spot when he walked by there about twenty minutes before. A farmer who was in the depot, when he heard the train approaching Big Island, went outside to hold his horses and as the front of the train passed by he also heard what he described as a powerful knocking as if something were dragging from the train. In contradiction of this evidence several witnesses placed on the stand by the plaintiff, who were near the depot, testified that they did not hear the noise referred to and saw nothing the matter with the train when it passed the depot. This evidence, however, can hardly be termed contradictory, as it is only negative in character.
On either side of the Big Island depot is a private road crossing at each of which were crossing boards laid parallel with the railroad track, the outside boards being almost level with the rails. It was testified by J. T. Reynolds, defendant company’s track man; O. D. Day, William Putney, regular track walker for the defendant of *417that section; R. G. .McGehee, assistant division engineer, and J. E. King, engineer of maintenance of way of the defendant company, and several disinterested witnesses that there were fresh metal drag marks tearing the wood fiber in the wood crossing boards longitudinally on the north side at each of these crossings. These marks being also about eight inches outside the north rail of the main line about the same distance that the column bolts would drag the boards in event that the arch bar on that side had been broken. It is further testified by Reynolds, the first track man to get to the scene of the derailment, that he had found a portion of the stock rail of the mill siding was pulled up towards the main line and was canted or tilted towards the main track.
It is conceded by both parties that the first indication of the wheels of any car off the track was about forty feet east of the switch to the mill siding, the marks of the flanges of the wheels being plainly shown on the cross-ties from that point to the point of the derailment on the north side of both rails of the main line track. It is shown by witnesses on both sides that there were marks on the outside of the stock rail, which looked as if metal had dragged against metal along the outer edge of the ball of the rail, extending eastwardly to a point where the marks on the stock rail showed that metal had dragged over it at the point where the wheel marks on the ties began.
Putting all the foregoing physical facts together, the theory of the defendant is that the arch bar broke somewhere west of Major station, and the rear end of the car had begun to drop down when the column bolts left the mark on the spur track at Major, dragging over the crossing boards on each side of Big Island depot, and as the car passed over the mill switch the column bolt nuts had sunk low enough to hang the ball of the north turn-out or stock rail of the mill siding, opposite the heel of the switch, and ran along that rail pulling upon and across the north stock rail until, with the two forces working on *418the car wheels, the column bolt ends were sheared off just above the nuts, and the rear wheels of the twelfth car, which had been pulled up and away from the main line by the leverage action of the bolt nuts and stock rail, flopped back on the ties instead of the rails.
Reynolds and Day, two of the first men to arrive on the scene, found in the vicinity of the stock rail of the mill siding the bottom end of a column bolt, and a similar piece of column! bolt with a nut attached which had been sheared off from the end of another column bolt. These bolt ends were fitted up by Maddox, chief car inspector, and J. S. Williams, master mechanic, and found to match a portion of the bolts that were in the truck of the broken arch bar.
In contradiction of the foregoing evidence, the plaintiff introduced several witnesses who had seen the crossing boards at Rig Island station at different periods of time after the accident, and who testified that they did not “notice” or “observe” any fresh marks or scars on the crossing boards. Two witnesses also testified that they were at the scene of the wreck on the day it occurred, and that the stock rail of the mill switch was not torn up and bent or canted over to the north rail of the main line track. These witnesses, however, were unable to fix the exact time that they visited the scene of the wreck, and it appears from the plaintiff’s evidence that Reynolds replaced the stock rail of the mill siding back into its proper position sometime in the early afternoon of January 11, and it is entirely probable from the evidence that this had been done before the two witnesses, above referred to, visited the scene of the accident.
It is contended by counsel for the plaintiff that the testimony of Day, Williams, and Maddox, in regard to the broken arch bar and the sheared off ends of the column bolts found near the scene of the derailment is unworthy of belief, although their evidence in regard to those matters is entirely uncontradicted. The ground upon which counsel base their statement is that the cross-examination *419of these witnesses shows that they had changed their testimony in some respects between the first and third trials of the case. We have critically read the cross-examination of these witnesses by plaintiff’s counsel. While it may be true that there is a slight variance on some points, according to the statement of counsel as to what their evidence was on the first trial of the case, we find only such immaterial changes in their, testimony as might appear in the cross-examination of any witness under the same circumstances. There is nothing in their cross-.examination which renders their testimony unworthy of belief.
It is also claimed by counsel for the plaintiff that the evidence of these witnesses was “manufactured” and should not be believed because the defendant failed to produce the broken arch bar and the bolt ends and nuts at the trial. It appears when the broken arch bar and bolt ends were found, they were taken to the shop at Clifton Forge, along with all the rear truck of the car C. & O. 22508, there assembled, examined by the master mechanic, J. S. Williams, and then, as ordered, sent to the office of the chief car inspector, in Richmond, Virginia. After holding the parts ninety days, no further instructions having been given with respect thereto, they were junked in accordance with the established practice of the defendant railroad.
It was further argued that the failure of the defendant company to produce this physical evidence at the trial shows that it could not sustain its theory as to the cause of the derailment. This argument might have some force but for the undisputed facts, already stated, that the engine and first eleven cars in the train passed safely over the switch and that portion of the track which the plaintiff claims was defective; that there were marks of only two wheels on the ties from a point forty feet east of the switch to the place where the cars piled up, six hundred feet east of the switch; and that forty-five cars behind the twelfth car in the train also safely passed over the same *420switch and track, fifteen of which were standing on the track alleged to be defective after the train had been stopped by the derailment.
Ihe plaintiff did not claim in the lower court that the doctrine of res ipsa loquitur applied, but he undertakes to do so here. The doctrine of res ipsa loquitur is an evidential presumption, not to be invoked to overcome evidence, but to be applied in its absence. In C. & O. Ry. Co. v. Baker, 150 Va. 647, 143 S. E. 299, 300, Mr. Justice Holt, referring to the discussion of the doctrine in Hines v. Beard, 130 Va. 286, 107 S. E. 717, says: “But the court asserts in no uncertain terms the now generally approved doctrine that when the defendant endeavors to rebut the presumption of negligence and introduces evidence tending to free itself of the charge of negligence, then the burden of proof resting upon the plaintiff to establish the negligence of the defendant inheres in the case in all its stages.”
In Norfolk-So. Ry. v. Tomlinson, 116 Va. 153, 81 S. E. 89, 93, a derailment case, the court; said:
“In order to rebut the presumption of negligence arising from the derailment, the defendant, we do not think was bound to satisfactorily account for the cause of the •accident. Sometimes accidents occur which are inexplicable, and to hold that a carrier of passengers must under all circumstances show what caused the accident in order to rebut the presumption of negligence arising from "the accident itself would in such cases impose an impossibility upon the carrier, and render it practically an insurer of the safety of the passenger injured.”
In C.& O.Ry.Co. v. Baker, supra, where a passenger was injured by a derailment of the defendant company’s cars, the court again refers to the case' of Hines, Director General, etc. v. Beard, 130 Va. 286, 107 S. E. 717, as follows:
“In conclusion Of the discussion of this question, the court holds that the plaintiff, having based his right of recovery on the negligence of the defendant, must show it by a preponderance of the evidence, and at no stage of *421the case can he escape this responsibility; that when the plaintiff has proved that he was a passenger, the fact of derailment and the resulting injury, then the doctrine of res ipsa loquitur, upon which the presumption of negligence rests, may dispense with further proof on his part, until it is made to appear in some way, either by his evidence or that introduced by the defendant, that it is at least doubtful if the derailment was the result of the defendant’s negligence. ‘Unless—says the court-—the evidence as a whole preponderates in favor of the plaintiff on the question of the defendant’s negligence, the plaintiff cannot recover. A mere equipoise will not entitle the plaintiff to a verdict. If it is just as probable that the derailment is due to some other cause as to the negligence of the defendant, there can be no recovery by the plaintiff. This, we think, is the proper application of the maxim res ipsa loquitur to derailment cases, and the one sustained by the best considered cases.’ In Riggsby v. Tritton, 143 Va. 903, 129 S. E. 493, 45 A. L. R. 280, the court followed the principles thus enunciated, as also in Tidewater Stevedore Co. v. Lindsay, 136 Va. 88, 116 S. E. 377. For a discussion of the cases generally upon this subject, see Hughes v. Atlantic City & Shore R. Co., 85 N. J. Law 212, 89 Atl. 769, L. R. A. 1916A, page 927, and note, page 930.” The doctrine cannot be relied upon to overbalance a conflict in evidence.
If the foregoing principles are applicable in the case of a passenger to whom the highest degree of care is owed by the carrier, a fortiori, it applies in the instant case where the defendant owed the plaintiff a lesser degree of care—the ordinary care due an invitee.
We are of the opinion that the plaintiff failed to show that the derailment of the train was due to any of the acts of negligence charged against the railroad company, but on the other hand the undisputed evidence of the defendant, at the least, tends to show that the derailment was caused by a broken arch bar on the rear north end of the twelfth car from the engine. Viewing the whole *422evidence in the light most favorable to the plaintiff, the most that can be said is that it is just as probable that the derailment was due to the broken arch bar as to the negligence the plaintiff charges against the defendant, and such being the case, the plaintiff cannot recover. Hines, Director General, etc. v. Beard, supra; C. & O. Ry. Co. v. Baker, supra, and cases cited.
One other contention on the part of the plaintiff remains to be noticed. It is contended that if the derailment was caused by a broken arch bar, it could have been avoided if the railroad company had made proper inspection of the car. The undisputed testimony is that this car had been inspected during the preceding twénty-four hours prior to the derailment at Hinton, West Virginia, and also at Clifton Forge, Virginia, without discovery of any defect. It is shown that the inspection was regular Class A train yard inspection that is commonly approved and practiced by all standard railroads; and that the crack in the arch bar was hidden from view by a flange of the column casting, and could not be discovered by any means short of dismantling the car.
It is argued that the car had been in use so long that it should have been previously shopped by the railroad, in which case the flaw would have been discovered. The evidence shows that the inspection given the car was in accordance with general and ordinary usage, and there is nothing in the evidence tending to show that as to employees the usage was not reasonably safe or adequate for its purpose and occasion, and nothing as to strangers tending to show that the usage did not afford as high protection as would result from any other known and practical methods of the business. Such being the case, the usage itself is conclusive evidence of the exercise of ordinary care on the part of the railroad company in the performance of its duty of inspection. Jeffress v. Virginia Ry. & P. Co., 127 Va. 694, 104 S. E. 393.
Since the methods employed by the defendant company constituted ordinary care, we do not think such care re*423quired the defendant to shop and dismantle the car without any evidence of the defect in the arch bar, or until it should have been done according to the usage in such cases.
In view1 of the conclusions reached, we are of the opinion that the defendant company cannot be held legally responsible for the regrettable accident involved, and that the evidence requires that final judgment should be entered here for the plaintiff in error.
Reversed.
Campbell, C. J. and Gregory, J., dissenting.