Mr. Chief Justice Neil
delivered the opinion of the Court.
These two actions were brought separately in the circuit court of Shelby county for the death of the two intestates mentioned in the margin, but, resting on the. same facts, were consolidated and heard together;
[164]*164John Koch and Alexander Johnson were members of a construction crew in the service of the Union Bridge & Construction Company, which was engaged in building the great bridge, known as the Harrihan Bridge, across the Mississippi river, at Memphis. They were killed while descending the man shaft of one of the bridge piers in course of construction on the Arkansas side of the river. It is alleged in the first count of the declaration that they were descending in the usual manner, the way they had been accustomed to descend into the shaft of pier No. 5, together with six other men, members of the same shift, when they and all the other men were overcome by deadly gas and fell from the ladder upon which they were descending and perished, their dead bodies being found at the'bottom of the shaft; that the gas which caused their death accumulated and existed in said shaft as the result of gross negligence and want of proper caré and caution on the part of the defendants, their agents and representatives. The second count states as the cause of action that the gas which caused the death o'f the two intestates accumulated and -existed in said shaft as the result of gross negli gence and want of proper care and caution on the part of defendants,- their agents and representatives, -in this, to wit:
“Defendants failed to put a lock at the top of the shaft so as to be able to keep compressed air in the shaft-... Had the shaft been supplied with a lock [165]*165at the top, all deadly gases and vapors would have been kept out of the shaft by reason of air pressure. This method of construction was the proper and scientific method to have been pursued, and by reason of the failure of'defendants to place a lock at the top of the shaft and to keep air in the shaft, they were guilty of gross negligence and want of proper care and caution, which resulted in the presence of gases in the shaft, causing the death,” etc.
The third count alleges that the intestates were directed to go down the shaft referred to, and that at the time the shaft was filled with deadly gas, which fact was unknown to the intestates, and they were overcome by the gas and died in the shaft; that shortly before defendants directed the intestates to descend the shaft, they had been informed that the decking of the caisson, at the bottom of the shaft, was burning, and they had reason to know that fumes and gasses would be generated from such burning deck of a dangerous and deadly character, but that the intes-tates were in ignorance of these facts, and so went to their death.
The plaintiffs introduced their evidence but the defendants introduced none. At the close of the introduction of the evidence the defendants moved for a peremptory instruction, which was granted by the trial judge. A verdict was accordingly rendered by the jury, and on this a judgment was based dismissing the cases of the plaintiffs. An appeal was [166]*166prosecuted to the court of civil appeals, and there the judgment was affirmed. The case was then brought to this court by the writ of certiorari.
In order to properly understand the controversy, it is necessary to explain the structure referred to in the declaration, and the nature of the work in which the intestates were engaged.
Pier No. 5, at the time the accident occurred,was within a day or two of completion. It was forty-six feet one way and twenty-five feet the other, and extended down into the bottom of the river eighty-five feet, counting from the top of the pier. The method of construction was this: The first thing was the sinking of a caisson in the bottom of the river. On the top of this there was a structure of heavy timber, called decking. On this decking there was erected a wooden form into which the concrete was poured. The weight of the concrete caused the caisson — on the bottom of which was what is called a cutting edge — to sink deeper. As the pier sank the wooden form was built higher, and the concrete filled into it, and so on until it had reached a point sufficiently high above the surface of the river. But it was necessary that men should go down into the caisson in order to eject therefrom sand, mud, stone, and pieces of wood, and whatsoever else might be found therein. Access to this caisson, called the work chamber, was given to the men by placing in the middle of the pier a steel pipe three feet in diameter, called the man shaft. This was made up of' [167]*167jointed sections eight feet éach in length, and, as the pier sank, the other sections were added at the top and so on, so as to keep the top of this pipe ahoye the water. Inside of this shaft there was a ladder by means of which the men descended into the caisson, or work chamber, at the bottom of the river, and on which they ascended to the surface again.
At the time the accident occurred, the caisson, or work chamber, was then at a depth of eighty-five feet. Various sections had been added to the man shaft. About forty feet from the bottom of this shaft there was located what was called the lock. This lock had two purposes. The main one was to gradually accustom the men to breathing the compressed air which they would have to encounter in the caisson, or work chamber. When they entered the lock they were required to open a valve which let into the lock compressed air from the work chamber, which mixed with the ordinary air, and after they had thus experienced this mixture of common air and compressed air for a time they were fitted to open the door of the lock and descend into the work chamber, and could then breathe the compressed air without, discomfort. The same plan was followed when they were about to ascend to the surface. That is to say, they stopped .in the lock for a while and admitted the common air from a valve on the upper side, and thus mixing it with the compressed air they became suf-ficently accustomed to the natural air to enable them [168]*168to breathe it without discomfort, and they would then open the door and go into the part of the shaft lying just above the lock, and they would then climb the ladder and ascend to the surface again.
At this point it. is proper to state that the compressed air was used, not only for breathing, but also to aid in the work in other ways. By its use sand and water were expelled from the caisson, by means of a pipe or hose, and by the force exercised through its pressure, water was kept from flowing into the caisson, thus enabling the men to work and dispose of sand, mud, dirt, and other things in the bottom of the caisson; likewise at periodical times the pressure of the compressed air was slackened, and- this had the effect of sinking the caisson deeper and introducing more sand and mud and other matter for removal.
The compressed air was introduced into the work chamber by means of a pipe which ran down through the pier on a line parallel with the man shaft, but separated from it a considerable distance, perhaps three feet. This air shaft passed down through the wooden decking into the workroom. It was, of course, covered with concrete along with the man shaft and other parts of the structure as the pier was built up, and it was added to in sections from time, to time as the pier sank, so as to keep its upper end above, water. Still further from the man shaft, and running parallel with, the air shaft some distance from it, was the [169]*169material shaft which was also a steel tube leading from the surface down into the work chamber. This ■was fitted with a bucket which was lowered and raised by an engine at the top. This bucket was used to carry from the work chamber pieces of stone and wood, and other heavy objects that could not be expelled by compressed air. This had a lock at the top.
The men were' accustomed to work in shifts of nine, and at the expiration of each two hours were relieved by another shift of the same number.
On the morning’ of April 6, 1914, at six o’clock, the customary shift descended to the workroom. They were to be relieved at eight o’clock. Sixteen minutes before eight o ’clock Smith, the lock tender of the gang then in the work chamber, accompanied by Lewis, another member of that gang, ascended to the surface through the man Shaft. A few. minutes after Smith and Lewis had so left and ascended to the surface the compressed air in the work chamber was slacked as usual, and as a consequence the work chamber descended further into the bottom of the river, the result of which was that additional material accumulated which was to be removed by the shift which was expected to descend' at eight o ’clock. The men .who were- in the work chamber, so expecting to be relieved, were not relieved by the arrival, of the expected new shift. They remained in the chamber from an hour and twenty minutes to an hour and a half before they had any means of escape. They [170]*170could not go up through, the man shaft in the usual way because their lock tender, Smith, had carried with him the key to the door which opened into the lock, so they waited, and finally they received a note sent down through the material shaft with the bucket notifying them to ascend to the surface in the bucket, and they did so, one at a time. About twenty minutes before any of them left the work chamber they perceived that a fire had developed in the wooden decking previously described, right around the air shaft. There is nothing in the record suggesting an explanation of the development of this fire except that the pipe which delivered the compressed air had grown hot and produced the combustion right around it, setting on fire the oakum which was packed around it on its way through the wooden decking to the work chamber. Compressing -air- heats it, and this heat is increased by its delivery through a pipe. The fire was not blazing, but only smouldering, and was extinguished by a bucket of water dipped from the bottom of the caisson, or work chamber. This fire, as stated, was not discovered until about twenty minutes before the men had left the work chamber. It is not shown how long it had been in existence before it was discovered, but there is evidence to the effect that the materials in its neighborhood were highly oxygenated, and that a fire therein would soon be burning briskly and would be drawn up as if fanned by a draft of air.
[171]*171We shall now direct our attention to the shift that was to descend at eight o’clock. As previously stated, Smith, the lock tender of the. gang already in the workroom, and Lewis had started np the man shaft at sixteen minutes before eight o ’clock, and had climed up to the surface by means of the ladder, and experienced no gas .on the inside of the shaft. Within ten minutes thereafter, that is, five minutes before eight o’clock, the relieving shift started down, composed of eight men. The gang or shift already down in the workroom not promptly ascending'through the man shaft in the usual way, some apprehension was aroused in the minds of those at the mouth of the pier, and another man, Pete Waton, was started down to investigate. When he had reached a point twenty feet below the surface he was perceived to topple and fall from the ladder. On this discovery being made the shift already in.the workroom was notified to come up through the material shaft, and did so, as already stated. Some three or four hours thereafter, between twelve and one o’clock, a man was sent down in a diving suit with an air hose attached to it in the usual way, so as to give him breathing air from the surface. He discovered all nine of the men lying dead in the bottom of the shaft -next to the lock. He brought up four of them, and the remaining bodies were removed by some other persons.
It was evidently supposed by those in charge of the work that some kind of poisonous gas had developed in the man shaft, inasmuch as the first [172]*172rescuer was equipped with the suit of.a diver. But gas had never before been known to develop in one of these shafts, and the only reason for inferring that it had developed on this occasion was the sudden falling to death of the nine men.
The plaintiff attempts to account for the gas in the following manner: As previously noted, the man shaft was composed, of lengths of eight feet of various sections. Each section was connected with the one below it by being fitted onto the other by a flange. These joints, as they were added on, were screwed tight, the space between the joints being covered by rubberoid to prevent leaking. But it was impossible to secure them so tightly that they would not leak water slightly, a small drip. Such is the nature of the work. It was also impossible to make them perfectly air tight. At four o’clock, Sunday afternoon, April 5th sixteen hours before the accident, it was discovered by the witness Howerton, a member of the shift then descending the man shaft to the work chamber below for the customary two hours’ service there, that one of the joints of the shafts, about sixteen feet above the lock was leaking air into the shaft, with a hissing sound. At six o ’clock of the same afternoon, fourteen hours before the accident, as Howerton had completed his two hours’ work.in the work chamber, and was ascending the ladder of the man shaft on his way to the surface, he discovered that another one of the joints, still higher up, was [173]*173looking, not air, but water, into the shaft. Bnt this stream was so small that aftér running thirty days after the accident, it had only accumulated three feet of water in the bottom of the shaft. It was the duty of one member of each shift, engaged in the work chamber, to keep on the lookout for leaks, and repair them. He was known as the inside lock tender because it was his duty to let his shift into and out of the lock, as the members of the shift descended to the work chamber or ascended therefrom. If the inside lock tender did not have time, to repair any leaks, it was the duty of the outside lock tender to make such repairs. The outside lock tender was stationed at the top of the shaft, and it was his duty to listen for any signals given him from below, and to attend to them when given. The outside lock tender was one Pete Watson. The evidence indicates that he was at the top of the shaft in attendance upon his duty before and at the time the accident occurred. It does not appear that any report was made to him concerning the leaks. Nor does it appear that any of the officers or .managing force of the construction company were informed of the fact of the springing of the leaks referred to. There is nothing in the record to account for these leaks, further than has already been stated, except the testimony of the witness Rogers, that if rocks or logs are permitted to remain against the cutting edge of the caisson, or work chamber, this will result in springing the joints [174]*174of the man shaft when the air is slacked, and the caisson sinks deeper. It is' in evidence that, before this accident, it developed that there were two logs encountered under the cutting edge of the caisson in the bottom of the river. An effort was made by the gang working at the time in the work chamber to remove them, but it was found impossible to do so, and the men cut off the logs as far as they could reach from the inside of the work chamber, and, as stated, allowed the caisson to sink resting on these logs. It does not appear that this fact was communicated to the managing officers or that it was known to them. The hissing-leak of air into the shaft was discovered by the witness Howerton, as he descended into the work chamber at four o ’clock Sunday afternoon, as previously stated, which was before the cutting edge of the saisson was sunk upon the two logs. The discovery of the water leak was made by him as he ascended at sis o’clock on the same afternoon, after the caisson had been sunk upon the two logs. There is nothing to account for the air leak into the shaft except the fact previously stated that it is impossible to make the joints air tight. The hissing of air into the shaft through the air leak is accounted for in the evidence only by the fact presently stated that compressed' air will, to some extent, penetrate the cement and struggle upward to an opening.
It is the theory of the plaintiff that the fire which developed, in the wooden decking before referred to [175]*175developed a gas which ascended along the air shaft, and penetrated the cement along with water to the man shaft and escaped through its leaking joints into the body of the man shaft, and poisoned the air therein. It is in proof that water will penetrate cement, and that gas will mis with water and may go with it. It is also in evidence that compressed air from the work chamber will to some extent penetrate the cement and make small air spaces, and that this air will gradually work through to the top, or to a release in some direction.
As .to whether any gas from the fire could ascend •from the ignited decking along the pipe, the only evidence upon this subject is that it would ascend until it reached the bottom of the pier where the shaft was inclosed in cement, and that it would then spread out; that it _would not ascend along the air shaft, but would spread out along the under surface of the pier. It is in evidence that cement clings very closely to iron.
It is also the theory of the plaintiff that even if the gas had ascended to the sprung joints of the man shaft. it never could have entered that shaft if the lock had been placed at the top of the shaft instead of near the bottom. It is in evidence that when the lock is at the top of the shaft the compressed air which fills the work chamber also fills the shaft and its pressure — thirty-two pounds to the square inch-keeps out water and would keep out gas, and would have prevented the entry of either. (IC
[176]*176The evidence is that there are two kinds of locks known in this work of erecting piers. The most ■ common lock is that which is constructed at the top of the shaft. This is used in erecting short piers. The customary method in erecting long or deep piers, such as was No. 5, was to place the lock near the bottom of the shaft.
We have stated that the evidence shows that the lock was expected to serve two purposes. One has already been explained, that is the equalizing of the natural air and the compressed air; the other purpose was to guard the men against the danger of the flooding of the shaft. The evidence is in conflict as to which is the better method to guard against that danger. Some of the witnesses ■ say that the lock built near the top of the shaft is the safer for that purpose; that is, to protect the men in ease of a flooding with water. Other evidence is to the effect that in long or deep piers the lower lock is the safer, because the men can get into that quickly, shut the door, and prevent the rising of the water. This lower lock was the much more expensive of the two forms of locks, and was generally used, as stated, in constructing deep piers which were operated with an elevator, instead of a ladder. This shaft, however, had a ladder instead of an elevator.
The lock of either form was constructed without any view of guarding against gas; no danger from gas having ever before been experienced, and previously stated.
[177]*177The question for solution is whether the trial judge committed error in giving a peremptory instruction in favor of the defendants on the facts above set out.
It is contended for the plaintiff that the rule res ipsa loquitur applies. The defendants insist that this rule does not apply as between master and servant, but concede that all of the facts should he considered together for the purpose of discovering whether, in the circumstances so developed, there is to he found a basis.for inferring that there was any wait of reasonable care on the part of the employer, or master, in the performance of the duties -imposed upon him by law, in providing a safe place to work, having in view, at the same time, the effect of any contributory negligence on the part of the servant, and likewise the assumption of the ordinary 'risks of work by the latter; also the negligence of fellow servants.
The rule res ipsa luquitur, in its primary or distinctive sense, may be thus defined: Where the evidence shows an injury inflicted, and also the physical thing inflicting it, and that thing does not usually, or in the ordinary course, produce such a result where due care is exercised by those in charge of it, it may be inferred that those so in charge of the thing inflicting the injury, failed to exercised such due care; that is, that they were guilty of negligence. The term re ipsa lo.quitur is often used in the more extensive sense of including all pertinent facts both [178]*178antecedent and subsequent to the infliction of the injury, which facts tend in and of themselves to indicate the responsible human agency. The former seems to ns the more accurate. It seems to us also the more useful because easily applicable to all cases, its outline being clear. The subject is fully discussed in the notes to Fitzgerald v. Southern R. Co., 6 L. R. A. (N. S.), 337, and Byers v. Carnegie Steel Co., 16 L. R. A. (N. S.), 214. The application of the rule in its primary sense is illustrated in this State by the following’ cases, in which no facts appeared except those which showed the injury and the physical 'thing which inflicted it, and from which was inferred the responsible human agency and the negligence of such human agent. These cases are Railroad v. Fort 112 Tenn., 432, 454, 80 S. W., 429; Railway Co. v. Saulsbury, 115 Tenn., 402, 90 S. W., 624; Burke v. L. & N. R. R. Co., 7 Heisk. (54 Tenn.), 451, 19 Am. Rep., 618; Martin v. McCrary, 115 Tenn., 316, 89 S. St. W., 324, 1 L. R. A. (N. S.), 530, in which it appeared that neighboring property was set on fire by sparks from an engine. It was also applied in Turnpike Co. v. Yates, 108 Tenn., 428, 67 S. W., 69, where it appeared that a pole over a turnpike gate fell on one passing through and inflicted an injury; in Mitchell v. Railroad Co., 100 Tenn., 327, 45 S. W. 337, 40 L. R. A. 426, where it appeared that on the violent blowing of the whistle of an engine under a bridge passing over a public road a team of mules attached [179]*179to a wagon passing over the bridge became scared and injured the driver; in Gorsuch, Adm’r, v. Swan, 109 Tenn., 36, 40, 69 S. W., 1113, 97 Am. St. Rep. 836, where a traveler on a public road was killed by a collision with a runaway team which was without a driver; in Chattanooga Electric Railway Co. v. Mingle, 103 Tenn., 667, 56 S. W., 23, 76 Am. St. Rep. 703, where the evidence showed merely that a heavily charged guy wire was knocked down by the stroke of the trolly of a passing car and one passing by was injured; also in Memphis Street Railway Co. v. Kartright, 110 Tenn., 277, 280, 75 S. W., 719, 100 Am. St. Rep., 807, where the evidence was that a trolly wire dislodged by the trolly of a passing car fell and struck a boy with an insulator attached to it, and thus injured him; in McHarge v. Newcomer, 117 Tenn., 595. 100 S. W., 700, 9 L. R. A. (N. S.), 298, where the evidence was that an awning in course of erection fell from the front of a store upon a sidewalk upon a street below, injuring a passing pedestrian. The subject may be further illustrated by those cases in our State 4n which it was held that the inference of negligence did not arise from mere proof- of the injury and the physical agency inflicting it. Thus in Weeks v. McNulty, 101 Tenn., 495-500, 48 S. W., 809, 43 L. R. A., 185, 70 Am. St. Rep., 693, it was held that the inference of negligence did not arise merely from proof that the burning of a hotel caused the death of a guest, the [180]*180origin of the fire not being shown. So in Railway Co. v. Manchester Mills, 88 Tenn., 653, 14 S. W., 314, where it was merely proven that fire was discovered in a freight car, the origin not being shown, the railway company in that case being protected against liability caused by fire, under a contract with the shipper against fire not caused by its negligence; and in De Glopper v. Railroad & Light Co., 123 Tenn., 633, 134 S. W., 609, 33 L. R. A. (N. S.), 913, in which it only appeared that a boy in a buggy, passing a street ear while the latter'was stalled at a grade and was rapidly revolving its wheels in one place without making any progress, received a stroke in his eye from some small object which penetrated it. In this latter case there is a valuable discussion of the rule res ipsa loquitur; also in the later case of Memphis Street Railway Co. v. Cavell 135 Tenn., 462, 187 S. W., 179. There are some other cases where negligence. was inferred simply from the infliction of an injury and the proof of the physical thing inflicting it. Among these are passenger cases, but in which was also necessarily involved the element that the injury to the passenger was a breach of the contract to safely carry. Some of these cases are Railroad Company v. Porter, 117 Tenn., 13, 94 S. W., 666; Railroad v. Kuhn, 107 Tenn., 114-117, 64 S. W. 202, and Transit Co. v. Venable, 105 Tenn., 460, 58 S. W., 861. In the first two 'of these .it appeared that the injury was caused by de[181]*181railment of the train, and in the last mentioned it was caused by a collision. In Railroad v. Mitchell, 11 Heist. (58 Tenn.), 404, the rule was invoked, but held not to apply where the evidence showed only that the passenger’s leg was run over by a car as he was getting off the train.
We see no reason why the rule res ipsa loquitur may not apply between master and servant where proof of the injury and of the physical thing inflicting it also excludes the presumption in favor of the master having performed his duty in the furnishing of a safe place to work and suitable tools to work with; also the servant’s own negligence, that of his fellow servants, and the servant’s assumption of the risks of the employment; hut obviously - such a case must he rare.
We have several master and servant cases in this State in which the rule was invoked, but not applied. Gill v. Brown, 130 Tenn., 174, 169 S. W., 752; Railroad v. Hayes, 117 Tenn., 680, 689-691, 99 S. W., 362; Railroad v. Lindamood, 111 Tenn., 457, 78 S. W., 99; Railroad v. Stewart, 13 Lea (81 Tenn.), 432. In Gill v. Brown, while the rule was learnedly discussed, the case turned on the point that another fact was discovered which indicated the master as the author of the injury. That was a case wherein it appeared that a boiler had exploded killing the servant. The fact, in addition to the mere explosion of the boiler, was that in the wreckage there were found some [182]*182rusted rivets indicating that the boiler- was in a defective condition at the time the explosion occurred. In Railroad v. Hayes, it appeared that a brakeman, while in the discharge of his duties, was injured by ' a timber structure which extended from the side of the road far enough towards the track to strike the brakeman while he was on the ladder on- the side of the car. It was held that the fact that he was so injured was not sufficient because it did not appear how long the obstruction had been in place, nor by whom it was so placed, the structure being purely temporary; some boards nailed together, and leaned towards the track. In Railroad v. Lindamood the fact proven was simply that while the brakeman was winding the brake on a freight car, the brake Slipped and threw the brakeman off the car. These fact's did not sufficiently negative the presumption in favor of the master. In Railroad Co. v. Stewart, the physical cause of the injury was the bursting of an oil cup. It was held the facts did not exclude the servant’s own negligence, he at the time being engaged in endeavoring to open the cup.
Where, in addition to the facts which constitute the res in its distinctive sense, that is the infliction' of the injury, and the physical agency inflicting it, there appear other facts which in and of themselves point to .the responsible human cause, a case of circumstantial evidence is produced rather than a pure case of res ipsa loquitur. However, in considering these [183]*183additional circumstances the inference to be drawn from the res as above defined still operates in connection with the additional facts. A good illustration of such additional facts is found in the case of Gill v. Brown, supra; also in the case of Railroad v. Northington, 91 Tenn., 56, 17 S. W., 880. In that case a servant was injured by the fact that a box on a push car in front of a handcar, which the servant was engaged with others in propelling, struck a comer of a depot platform, and fell under the car, causing a sudden jerk, which threw the servant against the handle bar, fatally injuring him. These facts, of course, did not justify the conclusion which we have before referred to, but other facts added did so, viz., that two boxes were placed on push car under _the direct orders of the foreman of the crew who was vice principal, that he was present, that he stood with one foot in each of the boxes, and that one of the boxes extended so far over the side of the push car as to produce the collision with the platform. These additional facts ■ excluded all negligence on the part of the servant himself, and of his fellow servants, likewise the assumption of risk, and the various presumptions in favor of the master. It was held that taking all these facts together a case of negligence was proven against the master.
It frequently occurs, as in the case last mentioned, that counsel do not pause after merely having proven the res, but proceed to develop their whole case, [184]*184introducing such facts as they possess to fasten the charge of negligence upon the defendant, as made in the pleadings. Under such circumstances the rule res ipsa loquitur is not of such great importance, although the inference justified thereby continues to bperate in the case, along with inferences to he drawn from other facts. The rule is of the greatest importance where, as it sometimes happens, the plaintiff is unable to prove more than the mere fact of the injury and the physical cause which produced it.
At this point it is proper to consider a question which has been much mooted, and upon which the decisions of the courts have produced some apparent confusion. This question turns upon the relation between the rule res ipsa loquitur and the burden of proof. The confusion has arisen from the two senses in which the term “burden of proof” is used in the authorities. In one sense the meaning of the term involves that burden which the party who has the affirmative of an issue must always bear until the conclusion of the case, a burden which never shifts no matter what the changing aspects of the controversy may show as the case develops on the evidence. The other meaning is expressed in the duty which devolves .upon the respective parties to meet with evidence the inferences adverse to them that may develop at any point in the trial from the beginning to the end of the case. Some authorities indicate this aspect of the. matter by the term “burden of the [185]*185evidence,” and in this sense it is said that the “burden of the evidence” shifts from time to time from one party to the other. The thought is well presented in the following excerpt from "Words and Phrases, vol. 1, p. 523, Second Series:
“The term ‘burden 'of proof’ has two distinct meanings. ‘It is used to refer: First, to the duty of establishing the truth of a given proposition or issue by such a quantum of evidence as the law demands in the case, whether civil or criminal, in which the issue arises; and, second, to the duty of producing evidence at the beginning, or at any subsequent stage of the trial, in order to make or meet a prima-facie case.’ This burden of proof never shifts during the course of a trial, but remains to the end with the party asserting the affirmative of the issue. Here the burden of proof was on plaintiff to show that defendants were guilty of negligence, and it was not upon defendants to show that they were not guilty of negligence, and this burden of proof did not shift during the trial, but remained with the plaintiff to the end. In some of the text-books and decisions it is said that, by reason of the introduction of rebutting testimony by defendant in the case, the ‘burden of proof’ is shifted, but all that is meant by this is that there is a necessity of evidence to answer the prima-facie case, or it will prevail, but the burden of maintaining the affirmative of the issue -involved in the action is upon the party alleging [186]*186the fact which constitutes the issue; and this burden remains throughout the .trial”—citing Chicago Union Traction Co. v. Mee 218 Ill., 9, 15, 75 N. E., 800-802 (2 L. R. A. [N. S.], 725, 4 Ann. Cas., 7).
Again:
“ ‘Burden of proof’ and ‘burden of evidence’ are often confused. The phrase ‘burden of proof’ is in fact more philosophical than practical. It means generally that a plaintiff, however often the evidence shifts, must, upon the whole, persuade the jury that his contention is right. The risk of nonpersuasion is all the time upon him.” Foss v. McRae, 105 Me., 140, 143, 73 Atl., 827, 829; 1 Words and Phrases (Second Series), 521.
Further:
“ ‘Burden of proof,’ in the sense of the duty of producing evidence, may he passed from one party to another as the case' progresses, while in the sense of obligation to establish the proof of the claim on which plaintiff’s case rests it is upon him throughout the trial.” Id., 522, citing Colston v. Bean, 78 Vt., 283, 285, 62 Atl., 1015, 1016.
Still further: “In the sense of the burden of the evidence, the ‘burden of proof’ may change from, one side to the other as the trial proceeds; but in the sense of maintaining the issue involved in the action, it constantly remains on the party alleging the fact which constitutes the issue, and when all the evidence has been introduced the jury must [187]*187determine whether it has been maintained.” Id. 522, citing Carroll v. Boston Elevated R. Co., 200 Mass., 527, 536, 86 N. E., 793, 797.
The same thought is a little more fully expressed in the following quotation:
“Where the party having the ‘burden of proof’ establishes a prima-fade case, and no proof to the contrary is offered, he will prevail. Therefore the other party, if he would avoid the effect of such prima-fade case, must produce evidence of equal or greater weight, to balance and control it, or he will fail. Still the proof upon both sides applies to the affirmative or negative of one and the same issue or proposition of fact; and the party whose case requires the proof of that fact has all' along the burden of proof. It does not shift though the weight in either scale, may at times preponderate.” Id. 522, citing Klunk v. Hocking Valley Ry. Co., 74 Ohio St., 125, 77 N. E., 752, 755 (quoting and adopting Powers v. Russell, 30 Mass. [13 Pick.], 76).
See, also, numerous other cases cited on the point in the same volume of Words and Phrases just referred to, and also the same title in volume 1 of the first series of that work. See, also, the subject fully discussed in the learned notes to Cleveland C., C. & St. L. R. Co. v. Hadley, 16 L. R. A. (N. S.), 527, and the note to Hughes v. Atlantic City & Shore Railroad Co., L. R. A., 1916A, 930 to 940.
[188]*188The office of the rule res ipsa loquitur is, in the main, to determine whether, on the state of facts shown, the case should go to the jury. If the court is of the opinion that the res proven furnishes some basis for an inference of negligence on the part of the defendant, nothing else appearing he should pass the case to the jury and that body may or may not make the inference as its judgment may dictate. See Hughes v. Atlantic City & S. R. Co., supra. This was the underlying thought in the following-statement of the matter from Sweeney v. Erving 228 U. S., 233, 240, 33 Sup. Ct., 416, 418, 57 L. Ed., 815, 819, Ann. Cas., 1914D, 905, viz.:
“In- our opinion, res ipsa loquitur means that the facts of the occurrence warrant the inference of negligence, not that they compel such an .inference; that they furnish circumstantial evidence of negligence where direct evidence of it may he lacking, but it is evidence to be weighed, not necessarily to be accepted as sufficient; that they call for explanation or rebuttal, not necessarily that they require' it; that they make a case to be décided by the jury, not that they forestall the verdict. Res ipsa loquitur, where it applies, does not convert the defendant’s general issue into an affirmative defense. When all the evidence is in, the question for the jury is whether the preponderance is with the plaintiff.”
This excerpt is also.quoted in two of our previous cases: Gill v. Brown and Memphis Street Railway [189]*189Co. v. Cavell, supra. It is perceived that in using the language which appears in the excerpt Mr. Justice Pitney had in mind the distinction above noted between the term “burden of proof” in its technical sense and-in its ordinary sense of “burden of the evidence,” recognizing that the latter shifts from side to side in the progress of the cause; also recognizing the fact that the jury may or may not draw the inference, it being a matter for them. From the last sentence of the excerpt it is perceived that the learned justice delivering the opinion had in view the evidence offered in the case in addition to that merely supporting the res. This is clear from the expression, “when all the evidence is in,” and also from the fact that there was other evidence besides that of the res actually introduced in the case. As we have previously indicated, the plaintiff may pause with proof of the res, that is, proof of the injury and its physical cause, or may proceed further and introduce various facts surrounding the accident, for the purpose of fixing the charge of negligence on the defendant. It is obvious that these facts, when so introduced along with the res may, instead of supporting an inference that may be drawn from it standing alone, wholly negative any such inference, and render it imperative that the trial judge should give a peremptory instruction in favor of the defendant.
Now as to the present case. The plaintiff below introduced witnesses who testified to all the cir[190]*190cumstances immediately connected with the accident, likewise those antecedent and subsequent thereto. The question is whether there is any evidence from which the jury could have rightly inferred negligence, and therefore whether the trial judge com-" mitted' error in giving a peremptory instruction in favor of the defendant, and- whether the. court of civil appeals committed error in affirming the action of the trial judge.
We are of the opinion that there was no error in the action of either court.- The theory of the plaintiff is that the two intestates and the seven other men who were found at the bottom of the shaft with them were all suffocated by foul gas that had accumulated in the man shaft. There is no direct evidence that ■ gas had accumulated, but by a process of elimination it is said that the conclusion is inevitable that gas was in the shaft and was the physical cause of the death of these men. Let that be granted; still, it appears that, ten minutes before these men went down the shaft, Smith and Lewis came out of the shaft with impunity; therefore the gas must have accumulated within that ten minutes, and there is not the slightest evidence that the defendant companies had any, notice of the fact, or any information whatever upon the subject, so it was. not possible for them to guard against it. They were as ignorant of its existence as were the men who went down the ladder.
[191]*191As to the claim that the gas was produced hy the fire in the decking, and the burning of the wood and the oakum, that it ascended hy the side of the air shaft, then permeated the concrete wall between the air shaft and the man shaft, it is sufficient to say upon this subject that the facts show beyond controversy that the fire did not develop until after the men were dead.
As to the claim that the defendants were guilty of negligence because the lock was not placed at the top of the man shaft instead of near the bottom, there is no evidence of negligence to be found here. The undisputed facts show that the purpose of the lock was to protect the men in case of flooding of the shaft as well as to enable them to pass from the outer air into the work chamber filled with compressed air, and from the latter hack to the outer air. There is no evidence that the injury occurred from flooding, or that there was any flooding of the man shaft; likewise, there is no evidence that any injury occurred by reason of passing from the outer air to the compressed air. There is evidence, as set out in the statement of facts, that, if the lock had been at the top of the man shaft, instead of near the bottom, the compressed air would have filled the man shaft, and because of its great pressure it would have kept out not only all water, but gas as well. But no negligence could he inferred from this fact as it appears from the evidence that gas had never before de[192]*192veloped in one of these man shafts, and so it was a danger that could not reasonably he foreseen; hence there was no apparent need for filling the whole of the man shaft with compressed air, to guard against a danger which no one had. ever encountered in such a place before. In addition, it appears without contradiction that in deep shafts, like the one that was under construction, No. 5, it was customary to have the lock near the bottom. It is true that several witnesses testify that it was customary to have the locks at the top of the shaft. Such is the fact, but it is a fact applicable only as to shafts that are not deep. There is no controversy but that in case of deep shafts the lock was placed as in shaft No.-5.
As to the springing of the joints of the man shaft the evidence is that the' only .cause that can be assigned for such a state of affairs as to the water leak was the sinking of the caisson or work chamber on two logs, and this was the action of the fellow servants of the decedents, and no recovery could-be had therefor. However, from what lias already been said, it is apparent that, even if gas entered through these openings, it was a danger that could not have been anticipated, and there could be no liability therefor. Moreover, it does not appear that notice of this condition of two of the jpints of the man shaft was brought to the attention of the defendants, nor is there any evidence [193]*193indicating- any duty of inspection on their part as tú snch a matter, further than the inspection by the inside lock tender already mentioned for water leaks, and it appears there was no danger of the flooding of the shaft by the small leaks referred to, and there was no flooding. This was the only danger that could have been anticipated from the springing of the joints of the shaft; further, the evidence is that these joints were screwed down -tight when the shaft was constructed, and when the accident occurred the pier was within one day of completion.
The foregoing disposes of all of the assignments of error made by the plaintiffs except the last, and that is to the effect that, inasmuch as the trial judge overruled a demurrer to the first count of the declaration, he could not thereafter direct a peremptory instruction as to the matter therein presented, since he was bound by his previous decision, that the facts' stated in that count made a good cause of action, and it is added that the evidence sustains the facts so charged. s ■
It is enough to say in response to this assignment that there was no evidence at all to sustain that count.
Since the accident occurred on the Arkansas side of the river, it is insisted that the laws of Arkansas should control. No such laws were proven formally in the, trial court, and so we must presume that the Arkansas laws are the same as our own, so far as [194]*194they bear on tbe subject. Moreover, the conclusion is fully borne out by the decisions of the supreme court of Arkansas, so far as called to oat attention by the respective parties.
It results that there was no error in the judgment of the trial court, or of the court of civil appeals, and the judgment of the latter, affirming the judgment of the trial court must itself be affirmed.