DENISON, Circuit Judge.
This action was brought by Bell, hereafter called the “plaintiff,” to recover damages for the destruction of his tobacco factory by a fire set by sparks from a passing engine of the railroad company, hereafter called the “defendant.” The plaintiff recovered a verdict and defendant brings this writ of error.
It was plaintiff’s theory that the spark, which set the fire, entered .an open second-story window of the factory facing north toward the railroad, and distant about 225 feet from the track. (This distance, and all others hereafter stated, are measured on the angle of the northwest*397erly wind then blowing.) This upper story was full of dry tobacco, hanging in frames, and very combustible. All the windows were open, or removed, as is customary in drying tobacco. It was defendant’s theory that the fire caught from a defective furnace flue in the southerly part of the building. This main issue was clearly submitted to the jury, which found in favor of the plaintiff. Defendant insisted that a verdict for it should be directed, and this insistence raises the questions of chief importance.
[1] No one saw the spark enter the window and no one saw the kindling and first moments of the flames. Plaintiff depends wholly upon circumstantial evidence. This court has considered that two things are essential to plaintiff’s right to recover: First, that the fire was set by such a spark; and, second, that the spark escaped through defendant’s negligence. Garrett v. Railroad, 101 Fed. 102, 41 C. C. A. 237, 49 L. R. A. 645; Railway v. South Fork Co., 139 Fed. 528, 537, 71 C. C. A. 316, 1 L. R. A. (N. S.) 533. There is no statute in Tennessee declaring prima facie liability in such case, or changing the common-law rules of liability or of proof.
[2] Upon the first subject — whether the fire was set by a spark from the engine— -plaintiff's evidence fairly tended to show these things, in addition to those already stated: Before any alarm had been given, and while all other parts of the factory were free from smoke, several people noticed smoke coming from this window and from the roof just above it. Then, looking through the window, they saw a small flame in the hanging tobacco, just inside the window. There was not then, and there had not been, any fire maintained in that room or in any part of the building from which smoke could have reached this point, which was upon the side toward the wind. The fire occurred immediately after the noon hour, while most of the hands were away. Employes had been, not long before, in the room in question, and there was then no fire there, and no one had entered the room after that time. The wind was blowing 30 miles an hour from the track toward the factory. An engine and freight train passed, going upgrade, and laboring hard, just before the fire; the interval between the passing of the engine and the first observation of the fire being variously stated at from 5 to 15 minutes. As the engine passed, a shower of cinders was heard to fall upon the roof of a wing of the factory, which roof, on the slope towards the witness, was from 100 to 200 feet from the track. A pedestrian on the adjacent highway, who was waiting for this train to pass, at a distance from the track which he is unable to state, and which, from his testimony, might have been anywhere from 75 to 150 feet (he thinks it was this maximum), and who was in the line between the engine and the factory, observed the laboring engine and heavy smoke and that a shower of cinders fell on him, and that some of them were alive so that they burned his hat.
On the other hand, it was claimed that an engine equipped as this one was cannot throw out a spark which can have sufficient vitality to set out fire at anything like the distance here involved. Tf this was established in any clear way, it might be difficult for circumstantial *398probabilities.to prevail against it; but we do not think it was so established. It is true that plaintiff called witnesses of some practical experience who testified that, in their opinion, if the spark arrester was in good condition, a live spark big enough to set anything afire could not be carried more than 20 or 30 feet from the track. While this is a matter of opinion and the plaintiff might not have been absolutely bound thereby, yet, if the record stopped here, plaintiff would have had difficulty in supporting recovery upon the theory that apparatus which may have been in good condition was negligently operated. Plaintiff is not confined to this theory. His declaration alleged both the imperfect apparatus and negligent operation of apparatus assumed to be perfect. The case was eventually submitted to the jury upon both theories. Plaintiff apparently preferred the former theory, and offered testimony supporting that theory and inconsistent with the other. Defendant was not satisfied to leave the record in this shape. It put on expert witnesses showing qualifications which may,have impressed the jury more favorably than the qualifications of plaintiff’s witnesses, and who testified in effect that, if the arrester was in good condition and the engine properly operated, the danger zone ought not to extend more than 100 feet though even this limit would be elastic, but said in effect that if the engine was improperly operated, even though in good condition, the danger that live sparks would be carried further would be increased and they could not undertake to fix any absolute limit. It was conceded that the force of the wind would be an element operating not only to carry the cinders further, but tending to keep them alive at a greater distance from the engine, and that how long a spark would live would depend on its size, on the force of exhaust and the force of the wind, and on the extent of the combustion which had taken place when it came into the air. Taking all the expert and opinion evidence together, it cannot be said that it operates conclusively to prevent any inference otherwise proper to be drawn from the facts in evidence.
We do not overlook defendant’s testimony tending to show that fire first broke out in another part of the building, where it had several times before been set on fire by a defective furnace flue, and that the fire was burning before the train passed; but those considerations were for the jury. On the motion to direct a verdict for tlie defendant, it must be assumed that -plaintiff’s testimony is true, and he must have the benefit of every fair inference therefrom.
[3] We have recently had occasion to examine and reaffirm the rule that while merely from equally balanced uncertainties the jury may not infer defendant’s causal relation to plaintiff’s injury, yet plaintiff’s evidence need not exclude every other possible source of injury; it is enough if the inference of defendant’s liability is fairly and reasonably probable and distinctly more probable than other suggested explanations. Railway v. Jones, 192 Fed. 769, 773-775, 113 C. C. A. 55; Railway v. Scherer, 205 Fed. 356, decided May 6, 1913. Judged by this principle and on the whole record, we are satisfied that plaintiff was .entitled to have submitted to the jury the issue whether the spark set the fire. Plaintiff’s testimony shows a continual shower of *399cinders going toward the factory just before the fire, falling on parts of the building, and some of them alive at a distance half or two-thirds the distance from the track to the supposed point of ignition; it shows the fire originating in material so inflammable as to be very easily set on fire in the strong wind then blowing; and it suggests no other plausible cause for the fire. On all the evidence, a presumption that a live spark was carried 225 feet cannot be said tO' be “contrary to reason, or opposed to natural and physical laws.” Railroad v. O’Neill (C. C. A. 6) 186 Fed. 13, 15, 108 C. C. A. 115. For some instances of verdicts sustained under more or less analogous circumstances, see Greenfield v. Railway, 83 Iowa, 270, 49 N. W. 95; Hammond v. Railroad, 211 Mass. 549, 98 N. E. 582; Railroad v. Burley Society, 147 Ky. 22, 143 S. W. 1040.
This leaves the question whether there was sufficient evidence of negligence. Concededly, if the spark guard was out of order because of a broken netting or some such defect which would permit large burning cinders to escape, its use in that condition would be negligence. Plaintiff seems to maintain that as defendant claimed the use of a screen in proper condition would surely prevent the escape of sparks which could live for 225 feet, and as plaintiff proved that this fire was set by a spark from this engine, thereby it would be established that the screen was defective. Defendant attacks this position as reasoning in a circle, or as an inference upon an inference, and says that since the first conclusion, viz., that the spark set the fire, is only an inference drawn from the circumstances, wc cannot base thereon the further inference that the spark guard was defective, appealing to the well-known prohibitory rule of evidence. U. S. v. Ross, 92 U. S. 281, 283, 23 L. Ed. 707. We do not need to decide whether a conclusion that the spark guard was defective would be a forbidden inference based upon an inference, or whether, under these circumstances, it would be a composite and unitary inference based upon all the’ facts and circumstances in their mutual relations to each other; and this is unnecessary because the refusal to direct a verdict may well rest upon evidence of other negligence.
There was proof fairly tending to show that this engine, in the matter of throwing sparks, was not operated with the care and prudence which the situation required. Although there was an upgrade, the engine was not pulling anything like its capacity; if it had been carrying the proper amount of steam, it would have pulled the train without effort or labor. It did in fact labor, with the wheels slipping, and was making a great noise with the exhaust from the stack; a new fire had been put in just before starting upgrade; the furnace doors were open below the grates, instead of being there closed and being open above the fire. With the doors in that condition and the exhaust turned into the stack and with the ports working rapidly, as when the wheels are slipping, a very powerful draft would be created up through the fire, which would greatly increase the normal proportion of cinders going through the spark guard and correspondingly increase the number if not the proportion of live cinders among them, and would tend to keep these alive for a longer time than with normal operation. If we *400add to these'circumstances the positive proof that showers of cinder's and sparks were thrown by this engine at this time, we have sufficient indications that the enginemen, having failed to maintain a proper fire and proper steam to come up this grade easily, were at this point forcing the draft in a way which, considering the. force of the wind and the inflammable character of the surroundings, the jury was authorized to think imprudent and negligent.
It follows that the request for an instructed, verdict was properly denied.
[4] Defendant submitted six special requests which the court refused to give, and error is assigned thereon. One of these requests was directed to the burden of proof,- and We think it was fairly covered by the general charge. Of the others, request No. 4 is typical. It is:
“The mere fact that Watson heard cinders fall on the tin roof is -no evidence that these cinders were alive or hot enough to set fire either to the tin roof or other more inflammable substance, as boards or tobacco.”
It is the settled rule that requests like this, which specify one particular bit of testimony and attempt to characterize it as evidence or as no evidence, are as apt to be misleading as to be helpful to the jury, whose duty it is to consider all the facts in their mutual relations and not in any isolated way, and that it is within the discretion .of the trial court to deny such requests. Railroad v. Ives, 144 U. S. 408, 433, 12 Sup. Ct. 679, 36 L. Ed. 485; Watson Co. v. Berberich (C. C. A. 8) 94 Fed. 329, 333, 36 C. C. A. 364.
[5] There was nothing in the action of the court in denying a new trial to make it an abuse of discretion or to take the case out of the familiar practice by which we refuse to review the decision of such motions.
We agree with the District Judge that the excluded evidence of a .statement, made during or just after the fire, to plaintiff’s superintendent and of his reply, was not admissible, either as an admission by plaintiff or as a part of the res gestse. We think it not necessary to state more fully the circumstances' upon this point. The ruling would not be useful elsewhere.
The judgment must be affirmed, with costs.