Opinion by

Mr. Justice Potter,

We have here two appeals, in cases which were tried together in the court below, were argued together here, and will be disposed of in one opinion.

The principal questions here presented for consideration, are whether there was sufficient evidence of the cause of the fire, and of the negligence of the defendant company, to justify the submission of these questions to the jury. The points are fairly raised in the first and second assignments of error, which are respectively to the refusal by the court below of binding instructions for defendant, and to the refusal of defendant’s motion for judgment n. o. v.

The subject of the liability of railroad companies for *58fire occurring along their lines, has been repeatedly considered by this court, and the character of evidence required to establish liability has been frequently pointed out. In Shelly v. Philadelphia & Reading Ry. Co., 211 Pa. 160, we said (p. 161) : “In the absence of direct evidence of the origin of a fire that is imputed to the negligence of a railroad company in not providing spark arresters, evidence of the unusual throwing of sparks of the company’s engines and of other fires started by them at about the time of the fire in question is admissible. This rule obtains also where there is uncertainty as to the engine that caused the fire and it cannot be shown that the fire proceeded from a particular engine. This class of testimony is admissible because of the failure of direct prqof and of the necessity of resorting to the proof of .circumstances as the best evidence and the only evidence of which the case admits.” In Mellinger v. Penna. R. R. Co., 229 Pa. 122, it was said (p. 126) : “In Lackawanna & Bloomsburg R. R. Co. v. Doak, 52 Pa. 379, recovery was allowed ‘although there was no direct evidence that the building was fired by the engine, or that the sparks were emitted from it at the time’; and so im Erie Ry. Co. v. Decker, 78 Pa. 293. Of course, the burden is always upon the plaintiff to show that the sparks from the defendant’s locomotive caused the fire. But ‘the evidence may be wholly circumstantial; as, first, that it was possible for fire to reach plaintiff’s property from the defendant’s engines; and, second, facts tending to show that it probably originated from that cause and no other’; Byers v. Balto. & Ohio R. R. Co., 222 Pa. 517. While the mere fact that sparks are thrown from the stack of an engine is not in itself evidence of negligence, where, however, sparks of an unusually large size are emitted it may be inferred that the engine is not supplied with a sufficient spark arrester: Henderson, Hull & Co., Ltd., v. Philadelphia & Reading R. R. Co., 144 Pa. 461, and cases cited at p. 477 of the report. ‘It is not required that the fact (negligence in the construction or *59management of the engine) he established by direct or positive proof; like any other fact, it may be established by circumstantial evidence; and, on account of the great difficulty of proving negligence in such cases, any proper evidence from which the negligence can be inferred is sufficient to throw the burden on the defendant’: Henderson, Hull & Co., Ltd., v. Philadelphia & Reading R. R. Co., 144 Pa. 461, 480. ‘The evidence that the engine threw out sparks of the kind described was in itself enough to carry the case to the jury, upon the question of whether the spark arrester was in good condition’: Byers v. Balto. & Ohio R. R. Co., 222 Pa. 547.” In Oakdale Baking Co. v. Philadelphia & Reading Ry. Co., 244 Pa. 463, the present Chief Justice said (p. 467) : “In support of their averment that the fire had been caused by sparks emitted from an engine negligently operated by the defendant, the plaintiffs were not confined to affirmative proof of the negligent operation of the engine, but were permitted to show such operation by circumstantial evidence. It was, therefore, competent for them to show, in support of their allegation that the fire was caused by defendant’s negligence, that at or about the time of the fire, before and after — within reasonable latitude — its engines had been negligently operated.” In 3 Elliott on Railroads (2d Ed. 1907), the general rule is stated as follows: “As there are few, if any cases where persons see the fire directly communicated, proof of communication must necessarily be more or less circumstantial. But, even though circumstantial evidence is sufficient to establish liability, where the evidence is such that it is a mere conjecture as to whether or not the company set the fire, it is proper to nonsuit the plaintiff. Where, however, it is shown that there was no probable cause for the fire except the railway locomotives, it may be sufficient to fasten it upon the company. It is not necessary that the plaintiff should produce evidence to exclude every other possible cause of fire. The plaintiff is not always required to show that any particular en*60gine set out the fire, aud evidence that other engines of the company, similar in .general construction to that supposed to have set out the fire, set out fire about the time of the injury, has often been held admissible.” Under the rule established by our decisions, particularly the Henderson, Byers and Mellinger cases cited above, the plaintiffs here were entitled to rely upon circumstantial evidence and were entitled to have that evidence submitted to the jury if it tended to show that it was possible for the fire to have originated from sparks negligently permitted to escape from defendant’s engines, and further that the fire probably originated from that cause, and that there was no other probable cause of the fire. From the evidence in the present case, the possibility of the fire having reached plaintiffs’ property from sparks negligently escaping from defendant’s passing engines, was manifest. The real question to be determined was whether the fire was probably due to that cause, and to no other.

The testimony of a number of witnesses tended to show that the fire started on the outside of plaintiffs’ fence next to the railroad; that it was very small when first seen, and, under the influence of a strong northwest wind, which blew from the direction of defendant’s tracks towards plaintiffs’ premises, the fire spread over the entire plant; in the two and a half hours preceding the fire, over thirty passenger trains, in addition to freight trains and drill engines, passed the point where the fire occurred; there was a heavy grade at that point and defendant’s locomotives during the two weeks previous to the fire and two weeks afterward, were seen to throw out sparks, many of them of larger size than could pass through a spark arrester in proper condition and repair; these sparks were hot and burned holes in clothing, set fire to combustible things on which they lighted, burned the persons of two individuals on which they fell, and caused horses in plaintiffs’ yard to run away. At times when the wind was from the direction of the railroad, *61they would be blown over to and light upon plaintiffs’ buildings, and in the yard; about five minutes before the fire was discovered, a heavy freight train went up the grade past the premises, pulling hard and emitting a dense smoke; there were also two shifting engines at work on the railroad near plaintiffs’ plant and in a lumber yard across the railroad. No witness testified to seeing a locomotive actually throwing out sparks on the day of the fire, though numerous witnesses said they had seen them doing so within a few days, some of them on the day before.

There was also evidence tending to show that the fire did not originate in any other way. The witness McCullough, superintendent of the American Ice Company, testified that when he first saw the fire the flames rose over the edge of the roof of the shed along the railroad; he could see no fire inside the plant; there was no building on fire, and the nearest fire was in the forge; the building containing the forge was not on fire; the wind was blowing from the northwest, directly from the fence to the shop; the company’s stack was more than 180 feet away from the place where the fire broke out, and the wind that day would have carried sparks from the stack in a contrary direction; smoking was not allowed in the yard. The foreman of the coal yard also testified that he was in the yard when the fire started in the fence, and he then saw nothing in the yard that was cause for the fire. Careful examination of the testimony has not disclosed anything indicating any other probable cause for the fire, upon the day upon which it occurred, than sparks from a locomotive. There was evidence that at times rubbish or waste material was burned up by means of small fires upon the premises, but it was not pretended that any such small fires were started on the premises upon the day in question. The point at which the fire which destroyed the plant, started, was definitely known and fixed. It began on the outside of a wooden fence next to the line of the railroad,.

*62After careful examination of all the evidence, we are satisfied that it was sufficient to justify its submission to the jury, upon the question of the origin of the fire. The decision in the case of American Ice Company v. Penna. R. R. Co., 224 Pa. 439, is cited as bearing against this conclusion. The language there used, was appropriate to the facts of that case, but it should be limited to those facts, in order to be consistent with the trend of our other decisions. The later case of Oakdale Baking Co. v. Philadelphia & Reading Ry. Co., 244 Pa. 463, restates what must be regarded as the established rule.

In the third assignment of error, complaint is made of the admission in evidence of the testimony of James G. Corcoran, a detective employed by plaintiffs after the fire to investigate its cause, to the effect that on June 13, 1913, four days after the fire, at 3:12 p. m., he saw deféndant’s locomotive, No. 997, passing plaintiffs’ premises and throwing out smoke and large sparks, some of which fell in the ice plant, and that, on June 25, he saw the same locomotive in defendant’s shop at Jersey City, and saw holes which had been burned in the screen, one being as the witness said, a two inch, the other a three and a half inch hole. 01

Under' the decisions above cited, this was competent evidence. If a locomotive was seen throwing out sparks larger than would escape through a spark arrester properly constructed and in good repair, it was proper to show the cause of such emissions. If it was owing to the bad condition of the spark arrester, it might fairly be inferred that similar emissions from other locomotives were due to the same cause. A statement of the general principle applicable to such conditions is found in 33 Cyc. 1373, as follows: “Where the engine alleged to have caused the fire is not clearly or satisfactorily identified, evidence as to the general condition of other engines of defendant of the same general appearance and construction and under similar conditions, at about the same time and place, in respect to throwing sparks or coals *63capable of setting fire, is admissible as tending to show a negligent habit on the part of defendant as to the construction, equipment and management of its engines, and therefore as tending to show negligence in that respect in the particular case, and as tending to show a probability that the fire originated from an engine of defendant.” The same principle is stated in 3 Elliott on Railroads (2d Ed. 1907), Sec. 1243, where it is said (p: 565) : “If the particular engine cannot be identified evidence is admissible that other engines of the defendant similarly constructed and operated set fires or threw igniting sparks equally far at other times, within a reasonable period, and at other places in the vicinity along the line, and the great weight of authority appears to be to the effect that such evidence is admissible without proof on the part of the plaintiff that the engines were similarly constructed and operated and without confining it to the exact time or day of the fire in question.”

There is no merit in the fourth assignment, which is to a portion of the charge in which the trial judge commented on the testimony of a witness as to bonfires in the ice plant. She stated that the last one she saw was two days before the fire which destroyed the plant. The comment of the trial judge that “it would be immaterial if there were no fires in the plant on the day of the conflagration,” was a natural and obvious statement.

The fifth, sixth, eighth, ninth and tenth assignments allege error in the answers to points. The questions raised are covered by the discussion of the first and second assignments of error.

In the seventh assignment error is alleged in the affirmance of plaintiffs’ fifth point, which referred to the dropping of burning coals upon the roadbed. This was not relevant to the point at issue. It was not alleged that the fire, which destroyed plaintiffs’ plant, originated from coals dropped upon the roadbed. The charge was, that the fire was caused by sparks from the smoke stack of a passing locomotive. The point as presented had no *64application, and should not have been affirmed, but we do not see that its affirmance was prejudicial to defendant. In his supplemental charge to the jury, the trial judge said: “The question for you to determine is, do you believe the probable cause of this fire was a spark emitted from the stack of an engine of the defendant company any larger than would pass through a spark arrester in good order and condition, and with the locomotive operated in an ordinary and proper manner?”

The eleventh assignment is to the refusal of defendant’s sixth point, which was as follows: “If the jury find from the evidence that there were causes for which the defendant was not responsible, which might just as well have produced the result complained of, and there is an absence of direct proof as to the cause of the injury, there can be no recovery for the plaintiffs, and the verdict must be for the defendant.” This point might perhaps have been properly affirmed, but under the 'evidence we cannot say that it was reversible error for the court to refuse to affirm it as presented. The rule applying to such conditions was properly stated in Wagener v. Philadelphia & Reading Ry. Co., 235 Pa. 559, in an opinion affirmed by this court, in which it was said (p. 564) : “Where the evidence points to one of several, possible explanations of an occurrence as more probable than the others under the circumstances, the jury may adopt it as supported by the preponderance of the evidence. And the most stringent test of the sufficiency of circumstantial evidence to justify a given conclusion goes no further than to require that it point to the same as the only rational explanation of the occurrence founded in the evidence.” In the present case it is by no means clear that there was any evidence of other causes of the fire than that alleged by plaintiffs, which was sufficient to be submitted to the jury. Of the other causes suggested there was no testimony indicating that the fire might have come from forges in the wagon shop. G-asoline and alcohol torches were used in the repair shop to burn paint *65off wagons, but there was no evidence that any paint was burned off on the day of the fire. So with certain electrical defects. It was shown that they had been repaired previous to the fire, and there was no evidence tending to show that the fire was caused by any defect in the electrical apparatus, or that it could have been so caused, taking into consideration the point where it started. As to bonfires, there was no evidence that there was any in the yard on the day of the fire, and when bonfires were made, the testimony showed that they were so located that the wind on the day of the fire would have blown flames and sparks in a direction away from the place in which the fire started.

We see no merit in the twelfth and thirteenth assignments of error, in which complaint is made of the action of the trial judge in reading to the jury his notes of testimony, and further that he failed to state certain modifications favorable to defendant, of the evidence of plaintiffs’ witnesses brought out on cross-examination. It does not appear .that any modification of the testimony of the witnesses on cross-examination was brought to his attention. It is not claimed that the trial judge misstated the testimony in any’ way. If he did so, it was the duty of defendant’s counsel to call his attention thereto at the time. The judge said to the jury, “You have heard the testimony in detail. My notes necessarily are not as extensive as those of the official stenographer, nor are you to be bound entirely by my notes. The entire testimony is for you. I merely refresh your memory as far as I can upon the testimony.”

In the fifteenth, sixteenth and seventeenth assignments of error, objection is made to certain additional instructions to the jury on the ground that they tended to put undue pressure upon the jurors to agree upon a verdict. We cannot see in these instructions anything approaching coercion or undue pressure. The trial judge very properly reminded the jury that after the length of time which had been consumed in the trial, and the vast *66amount of testimony that had been heard, it would be unfortunate if they were not able to arrive at a verdict. After further effort to clarify the issue, he said to the jury: “I am going to let you go back and discuss the evidence and see if it is not possible for you to reconcile your views. I am not here to urge you one way or the other, but a case of this kind should not be allowed to fail and be retried without a very earnest effort on the part of the jury to arrive at some conclusion. Of course, every man is entitled to his opinion under the circumstances, and that is what you are here for. If you are not convinced there is a liability on the part of the defendant, of course it is your duty to say so. You will retire gentlemen and discuss this matter a little more fully.” In this statement we see nothing that can fairly be considered prejudicial to defendant.

The eighteenth assignment is to the refusal of the court below to grant a new trial on the ground of after-discovered evidence. This is a matter so clearly within the discretion of the court below that its action in this respect will not be reviewed in the absence of clear error: Shade v. Llewellyn, 250 Pa. 456; Mellinger v. Penna. R. R. Co., 229 Pa. 122.

The assignments of error are all overruled, and the judgment in each case is affirmed.