Plaintiffs allege that their crate and basket factory was burned by the emission of sparks from defendant’s engine, the result of defective construction or negligent management. Eor the purpose of showing that the engine used by the defendant on the day of the fire, emitted sparks, plaintiffs introduced testimony to the effect that defendant used the same engine on its road from Warsaw to Clinton, several days before, and after, and on the day of the fire. They thereupon introduced R. B. Faison, who testified, after objection by defendant, that he was at Turkey, a station between Warsaw and Clinton, the distance between the two points being about twelve miles, on the. day of the fire and next day thereafter. That on the last named day he came to Clinton on the train. Thought it was the same train which went to Clinton on the day of the fire. Did not see the engine on the day of the fire, nor the day before. When train reached the “Y” it stopped. There was a car load of cotton seed hulls attached to the train, or making a part thereof— *583second car from the engine; the hulls were on fire. Employees were carrying water from the engine to put on the fire. Saw smoke coming out of the top of the car. Defendant insists that, in the absence of any evidence tending to show that the hulls were fired by the engine, the testimony was irrelevant and incompetent. The plaintiffs contend that it is competent for them to show that the same engine, shortly before or after the fire in question, emitted sparks. In this we concur. The proposition is well stated and sustained by abundant authority, being entirely consistent with the reason of the thing, in 11 Am. & Eng. Enc. (2 Ed.), 512. The decision of this court in Ice Co. v. Railroad Co., 126 N. C., 797, is not in conflict with this principle. In that case, it' was held incompetent to show that engines, other than the one which set fire to the property, emitted sparks. In Cheek v. Lumber Co., 134 N. C., 225, it was proposed to show that the engine alleged to have set fire to the wood, twelve months before the fire in question, and at another place, emitted sparks. This was held to be irrelevant. The defendant’s exception, however, is based upon the contention that, assuming the fact testified to by Faison to be true, that on the day after the burning of plaintiffs’ factory, a car load of hulls attached to the engine, which it was alleged set fire to the factory, was seen on fire, it did not tend to prove the fact in issue, — that the engine, by the emission of sparks, set fire to plaintiffs’ factory. The witness does not say that the hulls were set on fire by the engine or that the engine emitted sparks. The evidence relied upon by plaintiffs to show that defendant’s engine set fire to their factory is circumstantial. No witness says that he saw the fire communicated to the factory. There is evidence other than that of Faison, both competent and relevant to be considered by the jury, tending to sustain plaintiffs’ contention. Was the testimony of Faison relevant? That is, did it tend to prove the plaintiffs’ allegation? If the witness had testified that the cotton seed hulls were fired *584by sparks from the engine, or that, the engine emitted sparks at or about the time that they were found to be on fire, such condition would have been relevant upon the question whether the engine emitted sparks at the time of the fire. The question, therefore, resolves itself into this. Does the condition dsecribed by the witness Faison reasonably tend to show that the fire was communicated to the hulls by the engine? If suit had been brought by the owner of the hulls, charging that they were burned by the negligence of defendant, he would, in the absence of any explanation in regard to the origin of the fire, have been entitled to recover; not, however, because any inference would have been drawn that the engine communicated the fire, by emitting sparks, but be.cause the carrier was an insurer and could only escape liability by showing that the fire was caused by the act of God or the public enemy. The principle upon which the relevancy of proposed testimony depends, has been frequently announced by this court and the authoritative writers on the law of evidence. The difficulty is frequently found in its application. Pearson, J., in Bottoms v. Kent, 48 N. C., 154, approving the language of Best on Evidence, says: “The rule, that evidence which is too remote is inadmissible, may be stated thus: that as a condition precedent to the admissibility of evidence, either direct or circumstantial, the law requires an open and visible connection between the principal and the evidentiary facts, whether ultimate or subordinate. This does not mean a necessary connection,' that would exclude all presumptive evidence, but such as is reasonable, and not latent or conjectural.” Henderson, J., in Hart v. Newland, 10 N. C., 122, says: “Evidence is of two kinds; that which is true, directly proves the fact in issue, and that which proves another fact, from which the fact in issue may be inferred. The rules regarding competency only, apply to the first kind of evidence, and relevancy, to the second. * * * That the fact to be inferred often accompanies the fact proven, is' not suf*585ficient, it should most usually accompany it; and I would say, in the absence of all circumstances, that it should rarely otherwise happen.” In that case the action was for deceit in the sale of a slave; for the purpose of showing a scienter, the plaintiff was permitted to show that the slave was a runaway, and while hiding out, defendant’s wife had been seen carrying food to him. The learned justice, of whom Pearson, C. J., said, “his power of reflection exceeded that of any man who ever had a seat on this bench, unless Judge Haywood be considered his equal,” said, by way of illustration, in regard to the testimony: “But the strong objection in this case is, that there must be two inferences drawn, to-wit: The wife saw and fed the slave, ergo she knew he was diseased, that the wife knew it, ergo, the husband knew it, being informed by her; an error in either inference which might very well happen, would introduce a falsehood; which * * * is an object of more solicitude than the exclusion of the truth.” Hall, J., dissented, showing that two learned judges drew entirely different inferences from the same fact. The language of Judge Henderson is cited with approval by Rodman, J., in State v. Vinson, 63 N. C., 335, in which he approves Roscoe’s statement of the law. “When the fact itself cannot he proved, that which comes nearest the proof of the fact is the proof of the circumstances that necessarily and usually attend such fact.” “If the fact offered to be proved be equally consistent with the existence, or nonexistence, of the fact sought to be inferred from it, the evidence can furnish no presumption either way, as in such a case, the one fact does not most usually attend the other.” The principle upon which the admissibility of this class of testimony, with its limitations, rests is discussed by Ruffin, J., in State v. Brantley, 84 N. C., 766. He says: “Amongst other hazards and inconveniences, it was found that to allow evidence to be given touching every collateral. matter that could be supposed, however remotely, to throw any light upon *586tbe main fact sought to be established, had the effect to render trials complicated, and to confuse and mislead, rather than enlighten the juries, and at the same time to surprise the party on trial, who could not come prepared to disprove every possible circumstance, but only such as he might suppose to be germane and material. And therefore the main rule was adopted of restricting the inquiry to such facts as, though collateral to the matter at issue, had a visible, reasonable connection with it — not such a connection as would go to show that the two facts, the collateral one and the main one, sometimes, or indeed, often, go together, but such as would show that they most usually do so.” Thayer on Ev., 264-5. The general rule is much modified by the occasion of its application. As, when the intent, knowledge, etc., is the fact in issue, conduct of the defendant in other transactions of like character is admitted. State v. Murphy, 84 N. C., 742, and numerous illustrative cases. Applying the general rule to this record, we are of the opinion that the testimony of Eai-son was not relevant. To give it any probative value, the jury must infer that the hulls were fired by the engine. While it does not clearly appear it would seem that they were in a box car. Witness said that he saw “smoke coming out of the top of the car.” So far as we can see, the jury had no information in that regard. It can hardly be said that the fact shown — that the hulls were on fire — had a visible, reasonable connection with the fact in issue — that the engine emitted sparks. More than one conjecture could be reasonably advanced as to the origin of the fire in the car. It does not appear whether the doors were open when the fire was discovered- — the fact that smoke was seen coming from the top of the car would seem to indicate that the doors were closed. We do not find any evidence tending to explain the origin of the fire in the car — and we are unable to see how the jury could do so. As was said in Armstrong v. Railroad, 130 N. C., 64, “But none of the evidence connects the origin of the fire, *587with any sparks or cinders emitted from tbe engine.” If there is no evidence that the hulls were fired by sparks from the. engine, of course the fact that they were seen on fire had no visible connection with the condition of the engine, which is the condition from which it is sought to show the fact in issue — that plaintiffs’ property was fired by sparks from the engine. We would not be understood as saying that it was necessary to show, by an eye-witness, that the hulls were fired by sparks from the engine. Conditions may have been shown reasonably pointing to that conclusion, in which case the jury may have reasonably inferred that the same engine on the day previous emitted sparks at or near the plaintiffs’ factory. The hiatus in the process of reasoning is the absence of evidence that the hulls were firéd by the engine and without this the entire structure is without foundation. As was said by Ruffin; J., in Brantley’s case, supra: “We fully recognize the difficulty which a judge presiding on the circuit must experience when called hastily to determine between that which amounts to slight evidence and that which constitutes no evidence.” Cheek v. Lumber Co., 134 N. C., 225. It is impossible for us to see what weight the jury attached to Fai-son’s testimony. The rule which we find pursued by this court in such cases is to grant a new trial.
Eor the reasons stated we are of the opinion that the testimony of McKinnon was competent. He testified that he saw sparks coming from the engine the day after the fire. He does not locate the place, but we take notice of the fact that the distance between Warsaw and Olinton is- only twelve miles. He was in the rear car. He says that the sparks which he saw did not set fire to anything.
We have examined the defendant’s other exceptions and do not think that they can be sustained. Both witnesses— Duncan and Hodges — had testified to facts which tended to show, and, if believed, did show, that plaintiffs’ factory was not fired by defendant’s engine. They were asked whether *588or not they bad made contradictory statements, thus laying tbe basis for introducing impeaching evidence. It therefore became competent to show that they had made statements contradictory to their testimony. His Honor confined such testmony to proper limits, as impeaching the witnesses. State v. Wright, 15 N. C., 439; State v. Goff, 117 N. C., 755.
We concur with defendant’s counsel that the statements of the witnesses would not be competent as substantive evidence. We do not think that testimony in respect to which contradictory statements were admitted, was opinion evidence. They had testified to substantive facts. Eor instance, the witness Duncan testified that when he saw the house, the fire was flaming from the top of the crates — that the whole thing was on fire; that the fire was on top of the block, which was sitting in a ditch; that there was no grass around that block — nothing but sand. This, and similar testimony, was for the purpose of excluding the suggestion that the factory was fired by sparks from the engine. He was asked on cross-examination whether he had not told Ered Owen that the engine set the factory on fire. This he denied. It was competent to contradict him in that respect. The same is true as to the testimony of Len Hodges.
Eor the error pointed out herein there must be a
New Trial.