By invitation of the driver, named Keller, plaintiff, Fish, was riding with Keller on the seat of a five-ton automobile truck as it approached a crossing of the Pennsylvania Railroad. Plaintiff’s nine year old hoy was with him. When the head of the truck was 15 feet from the track, Keller stopped, and he and Fish looked in both directions and listened. Neither seeing nor hearing any approaching train, they started on, but when about halfway to the track the engine stalled. Plaintiff got off, went forward, and cranked the engine, looked up and down the track, and signaled to come ahead, and, as the front of the truck passed where he was standing, stepped into the seat. Before the truck had finished crossing the track, it was struck by a train from the south. Keller was killed, plaintiff badly hurt, and the boy somewhat injured. Fish thereupon brought this action in the court below, and it was twice tried without permanent result. At the close of plaintiff’s evidence on the third trial, the court instructed a verdict for defendant, and to review this instruction, Fish brings this writ of error.

It was assumed by the court below, and is conceded by both parties, that the evidence tended to show negligence by the railroad. Taken most strongly for plaintiff, the proof was that the train was running from 60 to 70 miles an hour, and gave no warning of its approach to this crossing. The instruction was based upon plaintiff’s contributory negligence.

The truck was going east. Shortly south of the crossing, the railroad, as it comes north, curves to the right. The westerly bow of this curve was occupied by a train standing on a switch track east of the main track, so that one standing upon the highway could see south along the track only as far as this curve, and, as he moved to the east, and approached the track, the closer he came, the shorter his range of view *202down the main track. The instruction given to the jury was based upon a supposed mathematical demonstration that when Fish left his place of comparative safety upon the ground, and stepped upon the moving truck, the train was in plain sight, and the danger to the truck was imminent. This result is reached by computing the extreme distance which the truck could have traveled after Fish mounted it, and before the collision, its rate of speed, and thus the time consumed, the speed of the train, and hence the greatest possible distance which can be attributed to the train at the critical moment. From these comparisons, it is said that the train could not have been over 500 or 600 feet away, and must have been plainly visible.

We are not satisfied that any definite and conclusive computation leading to this result can be based upon this record. There are too many elements of uncertainty, and a minute variation in any one of these elements destroys the demonstration. We first observe that the computations rest upon the inerrancy of the sole witness, the nine year old boy, who testified at this trial two years later. It is right to assume that he was an honest witness, giving his best recollection of everything; but the jury is not bound to accept from such a witness his estimates of time and distance as perfectly accurate, particularly, when the inferences therefrom tend to show that two capable men were guilty of suicidal carelessness. Tf the distance from the track-to where the truck first stopped was 20 or 25 feet, instead of 15 feet, or if Fish walked two or three steps toward the truck before mounting it at the west rail of the track, the computation is upset; and neither of these suppositions would be .so substantially variant from what the witness said, or would so impeach his testimony, that the jury might not have adopted one or the other as the most reasonable explanation of the event. The margin is too close to allow for the trifling errors in recollection or mistakes in estimates commonly found in the testimony of even the best witnesses. This situation illustrates the difference between the instant case and our opinion in Hurlburt v. Erie Co., 221 Fed. 911, 137 C. C. A. 481. There, there was a clear vision of the track for a quarter of a mile, and the estimates of the witness as to the speed of the horse and buggy, or the place they began to watch out' for the train, could be greatly modified and varied without affecting the result.

We next observe it is assumed that the truck was running not less than 2 miles per hour. There is no testimony fixing the speed, but this rate was thought to be the reasonable minimum; on the contrary, we think it is not uncommonly to be observed that a heavy truck, starting up from a standing position to cross over a railrpad track, sometimes moves very slowly — say as slowly as 1 mile an hour — for the first few feet; and i,f this assumption is permissible, the computation is again upset, for not more than five or six seconds need have intervened between the appearance of the engine around the curve and the collision. Much is made of the statement that the truck “slowed down” to allow plaintiff to mount it; from this it is inferred that the truck was moving rapidly enough to permit the speed to be reduced; but still without stopping. ■ However true this would be of a truck under full headway, it can hardly be accepted as an infallible descrip*203tion of precisely what happened when the truck was just getting started. When this boy is speaking of an occurrence only 7 feet ahead of where the truck started, the distinction between slowing down and going slowly is not enough to support an instructed verdict.

Still further, it is assumed that after Fish had gone upon the track and signaled to come ahead, and as he stood on or near the track with his back to the south, it was his duty to turn around and look again to the south before he mounted the truck. This might be true as matter of law, if a substantial interval had elapsed after the first look; but, if the time intervening was only two or three seconds, it cannot be imperatively concluded that a reasonably prudent man would have turned around and looked again. Such interval only covers the time necessarily consumed in preparing to mount and mounting the truck.

Again, it is not at all inconsistent with the testimony that the place where the truck stalled and stopped was so close to the rail that it would seem to Keller and Fish to be in danger from a passing train, even though measurements and reflection would have shown the lack of danger. It' follows that the attention of both might well have been concentrated on the task of getting the truck across the track, and hence that their duty of looking out for an approaching train, cannot be judged by the same strict standard as if there had not appeared to be an emergency.

This history of the case indicates that the questions of fact involved are close, and the inferences necessary to support the right of action may not appeal to all as the probable ones; but we are not convinced that all reasonable men must agree in concluding that Fish did not exercise ordinary care under the circumstances.

The judgment must be reversed, and the case remanded for a new trial.