SWINDALL, J.

(1) This is the third appeal in this case, a judgment in favor of the plaintiffs having been reversed in an opinion reported in 88 Okla. 99, 212 P. 314, and a judgment also in favor of the plaintiffs having been reversed in an opinion reported in 115 Okla. 76, 241 P. 794. The present appeal is by the defendant from an order overruling its motion for a new trial after a third verdict in favor of the plaintiffs, and from the overruling of the defendant's motion for a directed verdict. The case involves the doctrine of the "last clear chance," and in the opinion on the second appeal the court held the deceased to be a trespasser, and announced the doctrine that there was no duty owed by the defendant to warn the deceased against placing himself in a position of danger, but only the duty of avoiding injuring him after observing him in a place of danger, and in defining the duty the court said that there was no liability in such a case if it were impossible after seeing him to avoid the injury. We are not satisfied with the definition of the duty so announced, as it literally imposes upon the defendant too heavy a burden, that of proving it to have been impossible to avoid the injury; as a matter of fact, even had the deceased been discovered in a place of danger, there would be no liability unless the defendant could then have avoided the injury by the exercise of ordinary care, and the burden of showing a failure to exercise ordinary care would be upon the plaintiffs. This rule is announced in: Oklahoma City Railway Co. v. Barkett, 30 Okla. 28, 118 P. 350; Atchison, T. S. F. Ry. Co. v. Baker, 21 Okla. 51, 95 P. 433, 16 L. R. A. (N. S.) 825; Clark v. St. L. S. F. R. Co., 24 Okla. 764, 108 P. 361; Crystal Ice Ice Cream Co. v. Wood, 53 Okla. 592, 157 P. 904. To recover in this case it would be necessary for the plaintiffs to prove: (1) That the deceased was in a place of danger; (2) that he was seen in such place of danger by a servant of the defendant in time to have averted injury; and (3) a failure thereafter to use ordinary care to avert the injury.

And in the former appeals in this case it was held that a primary issue was whether the deceased was seen in a place of danger, that he was a trespasser, and that the defendant was under no duty to warn him against getting in a dangerous position.

(2) The plaintiff in error urges numerous assignments of error, one of which was the overruling of its objection to the introduction of evidence in support of the second amended petition, the objection having been made because it was contended that the petition was demurrable and failed to state a cause of action for the reason that the allegation that "said employees did know, or could have known by the exercise of reasonable diligence, the position of said Ginn at the time said machine was started," being in the alternative, was not a positive allegation of either alternative, and that the latter alternative did not charge a breach of any duty, with the result that the petition failed to charge the breach of any duty. Overruling the objection was error. The rule as to alternative allegations is stated in 21 Ruling Case Law, at page 451, as follows:

"A pleading must state the cause of action by direct averments, and not by averments in the alternative. In general a violation of this rule vitiates the pleading. Where the only effect of such allegations is to make the pleading uncertain, the remedy is by motion; but where the complaint alleges in the alternative two statements of fact, one of which is sufficient to constitute a cause of action and the other not, they neutralize each other, and demurrer will lie."

The latter part of the rule applies in this case and it was applied in the following cases: Stricklin v. Chicago, M. St. P. Ry. Co., 59 Mont. 367, 197 P. 839; Grout v. Central Electric Ry. Co., 125 Mo. App. 552, 102 S.W. 1026; Anderson v. Minneapolis, St. P. S. S. M. Ry. Co., 103 Minn. 224, 114 N.W. 1123; Knoor v. Reineke, 38 Idaho, 658, 224 P. 84; Baker v. Shafter (Tex. Com. App.) 231 S.W. 349; Ball v. Youngblood (Tex. Civ. App.) 252 S.W. 872. Overruling the objection to the introduction of evidence left in the case under the pleadings on which trial was had, and, so far as issues under the allegations were concerned, an alternative ground disapproved and eliminated from the case by the opinion on the second appeal.

On the contention that the verdict was not sustained by the evidence, in addition to a review of the evidence, the plaintiff in error urges that it is a legal principle that there cannot be an inference upon an inference. That is a common expression, and it also appears in the form that "there cannot be a presumption upon a presumption," and that an "inference (or presumption) must be based upon an established fact." The existence of such a principle, either of logic or law, has been vigorously denied and wholly refuted by the highest authority upon the law of evidence, in the following language:

"It was once suggested that 'an inference upon an inference' will not be permitted, i. e., that a fact desired to be used circumstantially must itself be established by testimonial evidence; and this suggestion has been repeated by a few courts, and sometimes actually enforced. There is no such rule, nor can be. If there were, hardly a single trial could be adequately prosecuted. For example, on a charge of murder, the defendant's gun is found discharged; from this we infer that he discharged it; and from this we infer that it was his bullet which struck and killed the deceased. Or, the defendant is shown to have been sharpening a knife; and from this we argue that he had a design to use it upon the deceased; and from this we argue that the fatal stab was the result of this design. In these and innumerable daily instances we build up inference upon inference, and yet, no court ever thought of forbidding it. All departments of reasoning, all scientific work, every day's life and every day's trials, proceed upon such data. The judicial utterances that sanction the fallacious and impracticable limitation, originally put forward without authority, must be taken as valid only for the particular evidentiary facts therein ruled upon." Wigmore on Evidence (2d Ed.) vol. 1, sec. 41, pp. 258-260.

Another author attempts to draw a distinction in support of the rule but he seems to wind up in pretty thorough accord if his statements are considered with regard to correct definitions of the terms used. His language is as follows:

"The proposition that one inference cannot be founded on or drawn from another inference rests upon ample authority. It must be received, however, with certain qualifications. One eminent legal author in calling it a fallacy says": — (Quoting from Wigmore as above) — "McCabe, J., of the Supreme Court of Indiana, in a case of murder sought to be established by circumstantial evidence, said: 'This process of tallying and confirming each circumstance by the others does not infringe the general rule that one inference cannot be based on another. There is an important exception to that rule, however. A fact in the nature of an inference may itself be taken as the basis of a new inference, whether intermediate or final, provided the first inference have the required basis of a proved fact. In short, it is not merely the sum of the simple probabilities created by the numerous individual circumstances pointing to and indicating the absence of burglars and all other human beings than appellant and his wife at the scene of the murder, but it is the compound ratio of them all, tallying with and confirming each other, that made them all strong enough, when considered together, to fully justify the jury in believing and finding that no burglars, and no other human beings than appellant and his wife, were present when she was shot; from which they had a right to conclude to a moral certainty that appellant committed the murder.' We think the expression used in stigmatizing the rule as a fallacy is too broad. The safer method of utilizing it is, in its present form, subject to the exception aptly alluded to by McCabe, J., in the Indiana case." Jones on Evidence, vol. 1, sec. 6e, pp. 39-41, citing Hinshaw v. State, 147 Ind. 334, 47 N.E. 158.

McCabe, J., was considering only circumstances and their interrelation or combined effect. He was not attempting to discriminate between testimonial evidence and circumstantial evidence. He evidently used the term "moral certainty" as the equivalent of "reasonable doubt" because the case under consideration was a criminal case in which the issues did not turn upon a preponderance of the evidence as in a civil case. A reading of the whole opinion indicates that what he meant by the expression "a fact in the nature of an inference may itself be taken as the basis of a new inference, whether intermediate or final, provided the first inference have the required basis of a proved fact," was that the inference must merely be one justified by the evidence and must not be merely a guess or a conjecture. As a matter of fact, Wigmore never meant that an inference might be taken as a basis for a further inference if it were a guess or conjecture and not a justified conclusion. No other conclusion could be reached from considering his most excellent work in its treatment of relevancy, admissibility, weight, proof, demonstration, reasonable doubt, preponderance of evidence, and the respective functions of court and jury in connection with burden of proof.

A very simple case occurs to us that should disclose the utter lack of reason in the alleged rule. Suppose suit brought to recover a large sum of money, and suppose the burden of proving payment not to be on the defendant, but proof of nonpayment to be a burden on the plaintiff. Suppose that it is evidenced by circumstances that a very short time prior to the alleged payment the defendant lacked money. We would argue that payment necessarily involved ability to pay; that lack of money permitted the inference of inability to pay; that lack of money at a time shortly prior to the alleged payment permitted the inference of lack of money at the time of the alleged payment; and the lack of money at the prior date might have to be proved by circumstantial evidence. Who would urge that such circumstances in evidence would not justify submitting the issue to the jury in the face of a bare, direct testimonial allegation of payment, or would urge that if they found against the claim of payment the verdict should be set aside because it had been necessary to base an inference on an inference

A guess cannot be based on a guess, nor a guess on an inference, nor an inference on a guess, but an inference may be based upon an inference, if the inference is a justifiable conclusion based upon competent evidence either direct or circumstantial.

(3) The inference of negligence must be based upon something other than mere conjecture, speculation, or probability, and it is not sufficient to introduce evidence of a state of facts which is simply consistent with, or indicates a mere possibility, or probability as to the existence of negligence, or which suggests with equal force and leaves fully as reasonable an inference of the non-existence of negligence. But the facts must be of such a nature and so related to each other that the inference contended for is the more probable or reasonable to be drawn therefrom. 45 C. J., Negligence, sec. 834, pp. 1265-1266; 20 R. C. L., Negligence, sec. 149, pp. 180-181; Midland Valley R. Co. v. Rupe, 87 Okla. 286, 210 P. 1038; Kansas City Southern Ry. Co. v. Henderson, 54 Okla. 320, 153 P. 872; St. Louis S. F. R. Co. v. Model Laundry, 42 Okla. 501, 141 P. 970.

It is undisputed that the deceased was killed by a blow from the crank on a drilling machine. There was considerable evidence on the question of whether the engineer, as he stood at the engine upon a box about a foot high, would have been obstructed in seeing a man who might be up against the sill of the rig where the deceased was found after being struck, or in seeing the crank or crank shaft. The evidence most favorable to the plaintiffs, defendants in error, was that there would be practically no obstruction to the view. However, as to seeing a man standing up against the sill, it is unquestioned from the evidence that the sill was six inches wide, was 18 inches deep, and that its top was from 46 to 48 inches high, from which it necessarily follows that part of a man five and a half feet tall, the height of the deceased, standing up against the sill, would be behind the sill, and the height of the man would only be from 18 to 20 inches above the top of the sill.

On the day when the deceased was killed, the drilling crew, consisting of an engineer or driller, a tool dresser, and a roustabout, were preparing to draw casing and had been getting ready to spool what was known as the casing line, a seven-eighths inch line, on what was known as the calf wheel of the rig. The rig was four and a half or five feet across, with a sill on the east side where deceased was struck which was six inches across and 16 inches deep. The top of the sill was from 46 to 48 inches high. The calf wheel was on a shaft slung from beneath the sill. The engineer, when he started the engine, was ten or 15 feet away from the other two members of the crew, who were inside the sills and down under the calf wheel shaft. The deceased was a pumper and his duty was to pump water to be used in the operations of the company, but his duties did not require him to be about the drilling rig. However, on that day he had sufficient water on hand and had been around the rig for a half or three-quarters of an hour before he was struck. The engineer testified that the last time he saw him before he was killed the deceased was back near the rear wheel of the rig, from four to six feet from where he was later found with his head pinned between the crank and the sill with his chin resting on the sand-reel journals, right after he was struck by the crank. The engineer also testified that he did not look for him after he went to the engine after helping get the line ready for spooling, and doubted if he could have seen deceased from the engine in the place where he had last seen him. The last time either of the men under the rig knew of the whereabouts of the deceased, according to their testimony, was when they were getting ready to spool the line, and at the time the deceased was standing out on the walk, which ran along the east side of the rig, and was three feet or more away from the sill of the machine, and he advised them to be careful not to get entangled in the line.

The undisputed evidence was to the effect that spooling the line was a very dangerous operation that required the careful and undivided attention of all members of the crew, of the two under the calf wheel shaft to avoid getting an arm or leg entangled in the line, as it was apt to be torn off if the engine were not stopped, and of the engineer because of the dangerous situation of the other two men. After getting ready to spool the line, the engineer went to the engine and asked twice if they were ready and did not start the engine until he had been a second time answered in the affirmative. The deceased was struck and killed by the crank in its first revolution.

The crank was about two feet long and was attached to the band wheel, which ran on the main shaft, and the main shaft was forward from the calf wheel shaft towards the north end of the machine. The engineer stood near the northwest corner of the rig. The crank was used in drilling operations, what is known as the pitman being attached to the crank, the upper end of the pitman being attached to the walking beam. The crank had no function to perform in spooling the casing line, and the pitman was detached from it and tied back against the rear wheel, the crank turning with the band wheel and performing no office in connection with the work being done.

There was about four inches clearance between the crank and the sill, the crank running outside of the sill and being about three inches wide, making the outside edge of the crank revolve about seven inches from the outside edge of the sill. A walk about three feet wide ran along the east side of the rig, but its inside edge was not within about 18 inches of the outside edge of the sill the inside edge of the rear wheel being a foot from the sill, the wheel having a six-inch rim, and the walk not being within the outside edge of the rim of the wheel.

The plaintiffs contend that the deceased was standing against the sill of the machine when it was started and was seen there by the engineer before the engine was started. They tried to develop the circumstances along two lines. First they tried to prove that it was dangerous to start the engine without seeing the crank, with the idea that if the engineer saw the crank he must have seen the deceased. But it is clear from the evidence that the crank had no function whatever to perform in the work which they were then doing, and that it was not at all necessary to see it. An effort also was made to prove that the engineer could not have seen the two men under the rig without at the same time seeing a man standing up against the sill opposite the crank. The question assumed that the deceased was standing against the sill, but it is not in evidence that he was standing there. As a matter of fact it cannot be told when he got even his head into position to be struck by the crank. But if it could be assumed that the deceased was standing at the sill before he was struck, there are numerous circumstances to be considered on the point of whether the engineer could have seen him. The court erred in permitting a witness to testify as to whether the engineer in looking at the two men under the rig must or might not have seen the deceased. That clearly was a matter to be inferred or not inferred from circumstances, and the witness was in no better position to draw the inference as to the fact than were the jurors from their common, every-day experiences. There was considerable testimony in the case as to whether or not certain parts of the rig would obstruct a view of the crank or crank shaft or of a man standing against the sill, and on that point the witnesses were competent to testify, as their experience was of a higher order than the information to be obtained from the photographs, but on the question of what one looking in a certain direction must have seen within range of vision was entirely different. On that point opinions were unnecessary. Certainly we would not say that he would see more than as though there were no obstructions whatsoever that would at all hinder, and if we should concede that there was nothing to obstruct, whether or not one looking at the two men must have seen a third within a certain distance from them, was a matter upon which an opinion was not at all necessary. Testimony as to the physical facts would have put the jury into as good a position to determine that question as was the witness. However, it is to be observed that the witness did not testify as desired. He said that "if he could see one he could see all three." and his testimony was in no place stronger, whereas, on the other hand, he said, "Well, it would be up to him whether he would do that or not; he could see them." This meant no more than that he might have seen them all or might not have seen them all. This matter necessarily turns upon circumstances, and it cannot be said that the circumstances are sufficient to justify a finding that the engineer did see the deceased at the sill, even if we assume for argument's sake that he was standing there before he was struck. The deceased was found two or three feet away from where one of the men under the rig was located, but that is merely a single circumstance. The men under the rig were underneath the calf wheel shaft, opposite and nearly underneath the crank. The calf wheel shaft was slung from beneath the sills and was something less than 30 inches high. The crank was outside of the sill. The deceased, on the other hand, was outside of a sill that was six inches across and was higher than the calf wheel shaft by the dimensions of the bearings, which are not given, and by 16 inches more, the depth of the sill. If there were no serious obstructions, it seems clear that the engineer could have seen all three men at the same time, but it does not follow that he did. As a matter of fact, if the men under the rig were kneeling, or standing on their knees as it was expressed, the engineer could have seen the tops of their heads, but the testimony also is to the effect that during the operation there might be times when he could not see them at all. Where the deceased was found was a little ahead, to one side, and above where the men were. In looking at the men the engineer looked from about 15 feet away where he stood on a box about a foot high, and his line of vision was downward. In addition to that it was his duty to observe them carefully, so that not only was his gaze focused narrowly upon them, but also his attention was concentrated upon them. The natural tendency of this would be to shut out vision or consciousness of other things. Further, the parts of the rig, while they would not obstruct the vision of one looking for the deceased, and directing his gaze to the point where deceased was found, would tend to be where they would be in vision range and would lessen the probability of another object throwing a sharp outline, especially if the gaze was not focused on that point. Further than that, while the men were beneath a shaft less than 30 inches from the ground, the sill would conceal the deceased to a height of about 46 or 48 inches, or at any rate, if he were only about five and a half feet tall as the evidence showed, only about 20 inches would be above the sill. And if he was leaning over with his head where he was found pinned, it cannot be disputed that part of his head or neck would be behind a stabilizing brace four inches by four inches, standing on the sill, the top of which was higher than his head had he stood erect, and behind which he was found pinned after he was struck.

We would not hold that the jury could not find that the deceased was seen, in spite of the testimony to the contrary, as that point was decided and correctly decided in St. Louis S. F. Ry. Co. v. Bryan, 113 Okla. 39, 237 P. 613, in which it was held that the jury could accept circumstantial evidence and reject positive testimony on the same point. However, taking into consideration the fact that he was not directly in line of vision, that while he was within the scope of possible vision in the sense that it would have been possible, in some positions of all the men at least, to have seen all three, the fact that there were other things around him, that he was behind a 16-inch sill the top of which was about four feet high, that the other two men were considerably below that and inside of it, and that they were engaged in work that required the closest of attention and care on their part, that the engineer's gaze would be focused narrowly and he was obliged to look down upon the two men, and to concentrate his gaze and attention upon them, to infer that the engineer must have seen the deceased, or that it was more probable, or even as probable, that he saw him as that he did not see him, would be wholly unjustified by the circumstances of the situation. The plain conclusion would be that these men were engaged in work that required their undivided attention, that it had such attention, and that they were attending to nothing else, and that would strongly militate against the faintest probability that the deecased would have been seen at the sill of the machine. It cannot be concluded from the evidence that the deceased was at the sill of the machine or seen moving towards it because of a duty to see the crank, or that such duty would indicate to the engineer when he got into a position of danger, as there was no such duty; nor can it be argued that he was seen in the place of danger or moving into the place of danger because of the fact that if the engineer saw the other two men he must have seen, or as a more reasonable conclusion would be held to have seen, the deceased in a place of danger or moving into a place of danger.

The plaintiff introduced no evidence specifically on the point that after the deceased was seen in a place of danger injury could have been avoided by the exercise of due care. Apparently, their contention was that he was seen even before the engine was started. That assumed that he was in a position of danger before the engine started. The evidence fails to prove that he was in a place of danger before the engine started, nor does it prove that he was seen moving into a place of danger just as the engine was started.

On the second appeal it was urged that the judgment be reversed with instructions to dismiss the case, but in the opinion the court called attention to the fact that there was some evidence in the record to the effect that it was unsafe to start the engine without seeing the crank and that there was evidence that there were no serious obstructions to prevent the engineer seeing a man at the crank where the deceased was found. All possible evidence was developed on both these points at the third trial, and it is clear that it was not unsafe to start the engine without seeing the crank in the work then being done, and that the crank was a useless appendage to the machine so far as that work was concerned; also there is no way of proving that the deceased was seen in a place of danger or moving into a place of danger, and the great weight of the evidence supports the statement of the engineer, against whom most of the evidence was directed, that he did not see the deceased from the time when he left him standing near the rear wheel of the rig, out of danger, until he found him pinned against the frame after he was struck by the crank and killed.

While sympathy always goes out to those suffering the misfortune of losing a relative and source of support, damages cannot be allowed against a party in no way guilty of any breach of duty in connection with the injury resulting in the loss. No breach of duty has been proven and it seems clear that none can be proven, even after three trials. The judgment of the trial court is reversed, with instructions to dismiss the case.

LESTER, C. J., and RILEY, CULLISON, ANDREWS, McNEILL, and KORNEGAY, JJ., concur. CLARK, V. C. J., not participating. HEFNER, J., disqualified and not participating.

Note. — See under (1) 31 L. R. A. (N. S.) 1032; 36 A. L. R. 39; 39 A. L. R. 487; 45 A. L. R. 983; 53 A. L. R. 1344; 60 A. L. R. 1445; 20 R. C. L. 75; R. C. L. Perm. Supp. p. 4817; R. C. L. Pocket Part, title "Negligence," § 53; (2) 10 R. C. L. 870, R. C. L. Perm. Supp. p. 2769; R. C. L. Pocket Part, title "Evidence," § 13.