A premature explosion of a charge being prepared for seismograph operations severely injured the plaintiff and killed his co-worker. The plaintiff sued the separate manufacturers of the gelatin dynamite, booster and electrical blasting cap which composed the charge. Based upon the general verdict that “we, the jury, find for the defendants, each and every one, against the plaintiff,” the District Court entered a judgment for the defendants. Plaintiff appeals asserting many errors. His objections largely boil down to an assertion that the District Judge’s charge to the jury erroneously excluded several substantive theories of recovery. The defendants counter that regardless of any so-called errors the judgment should stand as each of the defendants was entitled to a directed verdict because the evidence did not indicate with requisite certainty what caused the explosion or that any of the defendants were negligent in any regard.
Just how or why the explosion occurred is hotly disputed. Plaintiff, a driller’s helper employed by the Seismic Engineering Company, was assisting a co-worker in the drilling of holes and preparing of explosive charges for seismographic exploration in certain counties in West Texas. In such operations an explosive, usually dynamite, is discharged in the ground causing shock waves which are registered on a seismograph and recorded. From this, information about earth strata useful in oil exploration is ascertained. As is sometimes the case with a useful product, the materials were potentially extremely dangerous. Here the charge was to be composed of sticks of gelatin dynamite manufactured by Atlas but sold by Olin under its trademark, a metal cylinder booster manufactured by duPont and an electrical blasting cap manufactured by Olin.1 All three articles contained high explosives. The dynamite is in the form of hard paperboard tubes approximately two inches in diameter and 24 inches long. Each tube — stick—-is sealed by heavy paper at each end. Each “stick” is formed to permit coupling of two or more sticks. Here five sticks were joined.
The three articles are connected together to make up the seismograph charge. First, a hole is made or enlarged in one end of the dynamite by the use of a metal punch. The round metal booster is pushed in the hole formed in the end of the dynamite and the electrical blasting cap is attached to the booster. Normally this combination is placed in a hole in the ground and detonated by an electric current.
At the time of the explosion, the plaintiff’s co-worker, the driller, was assembling the charge. As was their practice, they had picked up dynamite and caps at the Seismic Company’s powder magazine that morning. Under their usual practice the driller’s helper (plaintiff) would couple the sticks of dynamite together and the driller would punch a hole in the dynamite and insert the booster and attach the cap. When the charge was assembled, they would lower it into a drilled hole in the ground with the cap and booster end of the dynamite being lowered first. The electric wires from the cap were snubbed around the charge and then ran to the reel to be connected to the switch and power source. Plaintiff had screwed the dynamite together and the driller had gone to his truck to get a booster and cap *79and had returned. Plaintiff did not specifically see the booster and cap. The driller in a squatting position had the punch in the dynamite and was pulling it out when the plaintiff walked a short distance away — about 18 feet — to the truck to get a drink of water. As he was standing by the truck, suddenly the explosives discharged instantly killing the driller and severely injuring the plaintiff.
Plaintiff’s expert and lay witnesses advance several hypotheses as to what probably happened.
One of the theories revolved around the dynamite. Workers for Seismic Engineering Company testified that on occasion they had noticed that the dynamite covering was sticky and oily. This was said to be an indication that nitroglycerine contained in the dynamite had separated and was leaking. Such a condition would set in motion two likely causes of the explosion. One, with the nitroglycerine in such a state, the punching of the dynamite or the insertion of the booster with an unavoidable presence of an abrasive, such as sand or grit, could easily detonate the dynamite. Another was that the sliding of a worker’s hand along the leaky stick of dynamite with some abrasive on it could easily ere-, ate enough friction to cause such an explosion. Arguing against this factual basis which the jury was entitled to credit, the defendants urged that considerable doubt was east by the fact that the dynamite is packaged in a waxed paper which the hot Texas sun might tend to make look somewhat sticky. This, it was claimed, was corroborated by the testimony of the plaintiff — who had only three days’ experience — that he did not notice that the dynamite used that morning was oily.
A second theory regarding the dynamite was that by the very nature of the operation, it was necessary to punch a hole in the end of the dynamite to insert the booster, which was the universal practice. The withdrawal of the punch might carry with it a film of nitroglycerine which could easily be detonated by an abrasive when the booster was purposefully pushed into the dynamite. Hence, the fault was the manufacture and sale of dynamite with an improper receptacle or means for insertion of an expected booster. In line with this, plaintiff contended that the dynamite was improperly packaged and should have — like duPont’s nitramon explosives contained in cans — provided a receptacle for the booster not requiring a forceable insertion thus eliminating the problem of the introduction of abrasives.
Plaintiff’s theory about the booster seems to be this. It was negligence to manufacture and sell a booster which was known would have to be pushed into a hole in the end of dynamite which necessarily had to be enlarged to accommodate the booster.
The theory as to the cap was that due to heat and pressure the cap, made of a plastic material, was defective since with any slight manual pressure during attachment it could have detonated, in turn causing the booster and dynamite to explode. Plaintiff also contended that Olin negligently supplied an outmoded cap (style ’52) which deteriorated easier in the heat and which should have been recalled and replaced by the safer new cap (style ’56). Olin contends that in fact a style 56 was being used and that the evidence will permit no other conclusion. But regardless of the style of the cap, plaintiff contends that it was defective.
Typical of the cross examination of an expert in cases of this kind, each of the theories advanced was questioned in an attempt to find the weak spot or underlying fallacy. In addition, the defendants offered'several theories of their own. One was that somehow the driller detonated the assembled charge by allowing it to strike a pole used to lower the charge into the holes in the ground. This, of course, would have to be weighed in the light of plaintiff’s contrary testimony and circumstances regarding the distances of the poles from the charge. Also, in an effort to show that the booster *80and cap were not involved, some of the defendants blamed the explosion on the metal punch used to make a hole in the dynamite. This is offset by testimony of the plaintiff that punching of the hole was completed before the accident and the circumstance that the punch itself was found intact after the accident with no appearance that it had been directly involved in the explosion. Here again the evidence was conflicting and properly was for the jury to resolve.
We have not attempted to inventory and discuss each bit of evidence in this 1800-page record, Travelers Ins. Co. v. Truitt, 5 Cir., 1960, 280 F.2d 784. But we have set forth these highlights, briefly discussed, to demonstrate that as to all of the defendants except duPont, the issues were properly for the jury to resolve.
We are, however, convinced that no case was ever made out against the duPont booster. There was no evidence of any kind and really no serious contention that it was the booster which suddenly exploded setting off the more powerful explosives. In fact virtually all the evidence about the booster came from Mr. Yon Ludwig, plaintiff’s expert witness, who reiterated several times that there was no fault in the booster as such.2
The only possibility seriously advanced was that perhaps the metal booster had picked up some abrasive, such as silica sand, while being handled which caused the dynamite to explode when the booster was pushed into the dynamite. It was a very real possibility that a film of nitroglycerine could have been released during the punching of the hole and the subsequent insertion of the booster with an abrasive could have detonated the dynamite. But if such was the case, the fault was not due to any defect in the booster, but solely to the manner in which it was used. Any issue as to the practice of pushing a booster into a too small hole is aimed at the receptacle for the booster in the dynamite, not the booster itself. We can find nothing to justify implicating duPont. Once evidence was presented and none of it indicated that there was any defect in the booster as such — including that of the plaintiff’s own witnesses which exonerated duPont — there would be no basis for submitting the case to the jury even on a res ipsa theory.
There is no longer any basis for doubt that a manufacturer may be liable to third persons who have no contractual relations with him for injury resulting from defects in his product which could have been avoided by the manufacturer’s exercise of due care. International Derrick & Equipment Co. v. Croix, 5 Cir., 1957, 241 F.2d 216; Roosth & Genecov Production Co. v. White, 1953, 152 Tex. 619, 262 S.W.2d 99; Restatement, Torts § 395; Prosser, Torts, § 84. One reason for the rule is the impracticability, *81if not impossibility, of effective inspection by the ultimate user of goods which may involve, if not carefully made, an extraordinary risk to human safety. In fact in many such instances the only satisfactory inspection and examination would be at the manufacturing stage. Especially is that true in this case where only the grossest of defects would be visually apparent to the ultimate user.
As the manufacturer’s duty is founded in tort, it does not depend upon the intricacies of the law of sales or upon esoteric concepts of privity of contract. The duty extends to that class of persons reasonably expected to be exposed to or use the product in justified reliance on the manufacturer’s expertise and technical ability. Here it is not contested that the plaintiff was a member of such a class — employees expected to use the explosives. Indeed, the defendants do not really dispute this. The main point of departure is in the proof required to show negligence on the part of the manufacturer. The District Court ruled that the plaintiff was not entitled to the theory of res ipsa loquitur in proving his case. Without such instructions to the jury on the theory of res ipsa, the jury on a general charge returned a general verdict for the defendants. We disagree and reverse.
We agree with Dean Prosser’s statement. “Since the injured plaintiff almost never has any direct proof of what has occurred in the manufacturer’s plant, he usually must resort to circumstantial evidence. In the ordinary case, this means that he must rely upon the doctrine of res ipsa loquitur. * * * ” Prosser, Torts, § 84 at 505. As this Court has pointed out, the doctrine of res ipsa loquitur “ * * * is simply a facet of the general law that verdicts may rest upon circumstantial evidence.” Revlon, Inc. v. Buchanan, 5 Cir., 1959, 271 F.2d 795, at page 799. See Prosser, Torts, § 42.
For the doctrine to be applicable, the plaintiff must have been injured by an instrumentality which, at the critical point in time, was within the defendant’s exclusive control. The critical point of time is not necessarily the precise time of injury. Rather, it obviously refers to the time the probable negligence inferred from the occurrence of the event took place. Ozark v. Wichita Manor, Inc., 5 Cir., 1958, 252 F.2d 671, at page 675 (on rehearing, 5 Cir., 258 F.2d 805). If the casualty is a sort which in the ordinary course of things would not occur in the absence of negligence, an inference that it was due to the defendant’s negligence is allowable. Honea v. Coca Cola Bottling Co., 1944, 143 Tex. 272, 183 S.W.2d 968, 160 A.L.R. 1445; Ozark v. Wichita Manor, Inc., supra. Thus, the theory is grounded upon the improbability of an accident occurring in the absence of negligence of the actor who controlled the forces causing the harm. Here, it is entirely reasonable to assume and, in fact, based upon ordinary experience, highly probable that these explosive charges when used under normal, contemplated circumstances would not prematurely explode in the absence of some serious defect. In fact, such is the evidence offered by the defendant. If such an unexpected explosion occurs, something seriously has gone wrong. Defendants do not deny this but simply point elsewhere for the explanation. This is translated into terms that res ipsa may not aid the plaintiff since he is unable to identify the particular cause of the explosion which must necessarily be separated out and identified to implicate any individual defendant.
Here there can be no doubt that the components of the explosive charge were under the exclusive control of the defendants at the critical manufacturing stage. As Atlas’ brief states, “Atlas could be charged with having had complete control of the dynamite manufactured by it during the manufacturing process.” “But,” it continues, “on no basis could Atlas be charged with exclusive control, or with any control at all, of either the electric blasting cap or the booster.” The defendants contend that all must fail because of this *82since the essential factor that the instrumentality causing the harm must have been in exclusive management and control of the cast defendant is missing.
But this principle “is but another way of saying that ‘the circumstances must point an accusing finger at the defendant if he is to be held liable for negligence. This is the function of the requirement that the accident be caused by instrumentalities under the management and control of the defendant'.” Ozark v. Wichita Manor, supra, 52 F.2d 671, at page 675.
For res ipsa to apply, it is not necessary that as a matter of law the jury could reach no other conclusion except that the casualty was caused by a particular identified force. It is — as it always was — a question of fact. Precisely this point was made in Ozark v. Wichita Manor, supra, 25 F.2d 671, at pages 673-674.
“The City’s expert did not undertake categorically or inferentially to dispute the main theme of plaintiffs’ expert that in all probability * * * [compressed air pressure] gave the explosive driving force to the water or air as it first came out of the hydrant when opened by the little boy.
“While this was not expressly contradicted, the jury was not compelled to accept this opinion. Concessions and questioned hypothesis, typical of cross examination of an expert, plus [other] factual evidence * * * permitted the jury to accept or reject its teaching. But it was probative evidence available for consideration by the jury under adequate instructions."
Nor is it necessary here that in order for res ipsa to apply one particular force must be severed out, identified and held as a matter of law to be the cause of the premature explosion. The various components were manufactured to be a part of one combination. The dynamite was manufactured by Atlas for Olin and Olin sold it under its trade-mark. Olin also manufactured the electric blasting cap. Clearly, Olin could not seek refuge under this musical chairs argument since it was implicated in the event that either the dynamite or the cap was negligently defective. As much is candidly stated by Olin in its brief, “concerning the dynamite, the liability, if any, of Olin and Atlas to plaintiff is identical.” Nor do we think Atlas was in any better position.3 A doctrine which is due largely to the inequity when one party has superior knowledge, unlimited access to information, and exclusive control at a critical, decisive stage cannot be overcome by such an artificial approach. The whole development of contemporary Texas law is to the contrary. Even in cases in which there is no combination as there obviously is here, the Texas courts recognize joint liability against actors completely independent and unrelated to each other in circumstances where their conduct has caused indivisible injury which cannot be accurately apportioned and identified by the plaintiff.4 Here from *83a physical standpoint the injury was caused by a combination of the three products. Where the consequences are so devastating and the risk to human life so great, manufacturers of products which are components designed to be used with other known products may not thus evade the responsibility to come in and explain. That is basically what the res ipsa doctrine requires.
The second aspect of element of control — evidence that subsequent acts have not intervened adversely to alter the instrumentality — adequately presents a factual issue for the jury. Honea v. Coca Cola Bottling Co., 1944, 143 Tex. 272, 183 S.W.2d 968, 160 A.L.R. 1445. There is sufficient evidence to warrant a jury inference that the materials were being used in the normal, contemplated manner. That it was for sale in the West Texas area and used there under contemplated conditions is established by sufficient evidence in the record. The evidence adequately permits the jury to conclude that the explosives were handled properly by the Seismic Engineering Company and in manner reasonably to have been foreseen. Nothing done by the driller or the plaintiff is conclusively shown to have adversely affected the materials. In other words adverse alteration of the instrumentality of harm was not established as a matter of law. Consequently, in such a posture, the issue
clearly presented a question for the jury to resolve.
The attempt to identify particular defects in the explosives does not prevent the case from being one for the application of res ipsa loquitur. Ozark v. Wichita Manor, Inc., 5 Cir., 1958, 258 F.2d 805; Gulf Refining Co. v. Delavan, 5 Cir., 1953, 203 F.2d 769, 771; Honea v. Coca Cola Bottling Co., supra, 183 S. W.2d 968, at page 971.
Our conclusion is that the plaintiff was deprived of an essential substantive theory of recovery. Without it, it was deprived of a basic facet of the rule of circumstantial evidence in proving, or attempting to prove, negligence on the part of the defendants.
As we pointed out in Ozark v. Wichita Manor, Inc., 5 Cir., 258 F.2d 805, at page 806 (on rehearing), this Court has twice approved a form of submission so that the jury has a usable understanding of the legal principle which permits, but does not compel, a finding of negligence. Texas &. Pacific Ry. Co. v. Buckles, 5 Cir., 232 F.2d 257, 263, see note 14, and Kansas City Southern Ry. Co. v. Justis, 5 Cir., 232 F.2d 267, 270, see note 3, 60 A.L.R.2d 628.
We think the Court also erred in the instructions given in connection with the duty to warn of defects in view of the inherently dangerous nature of the explosives.5 Specifically this instruction *84expressly required that before the jury could return a verdict against Olin or against Atlas, the jury would be required to find that such failure, in each case, to warn was the “sole and proximate cause of the injuries.” 6 The basic instruction, note 5, supra, was a substantial paraphrase of the Restatement of Torts § 388.
The defendants do not undertake to justify the correctness of an instruction requiring a finding of “sole proximate cause.” They insist, however, that while this was technically erroneous, it could not have been prejudicial in view of other portions of the charge.7
But we are unable to satisfy ourselves that this error was harmless or that the later general instruction cured the defects in the first. This instruction covered a specific theory. The Court was equally specific in its requirement that to warrant the return of a verdict against either one or both of these defendants, it was essential that as to either defendant the failure to warn was the “sole proximate” cause. As a matter of substantive Texas law on tort liability, *85there is a vast distinction between a or the proximate cause and the sole proximate cause.8
Under the instructions as actually given, it is entirely possible that as to the cap, for example, the jury reached the decision that a warning was required so that the failure to give the warning by Olin was negligence. However, having found this, the jury was not able to return a verdict against Olin because of the specific instruction requiring them to make the additional finding that this ■“was the sole and proximate cause.” This was highlighted by the repetition in identical language as to the dynamite and the defendant Atlas. The general instructions later given, see note 7, supra, do not undertake to modify or alter the former specific instruction. If they were intended to, the instructions did not make this clear. It concerned a matter of such importance both in substantive theory and supporting facts that there should not have been any doubt. When a court purposefully uses on two specific occasions a term of unusual specific legal and substantive significance, such as “sole proximate cause,” we cannot hold that the jury was free to conclude that the Court did not mean what was said or that having just said it, what was said was retracted by broad and general . terms.
There is, of course, no way to demonstrate that the jury may not have been significantly affected and a verdict based upon it may not stand.
Our conclusion is that as to Atlas and Olin the case must be retried. As to duPont, the judgment is affirmed.
Reversed and Remanded In Part and Affirmed In Part.
. The defendants were:
Atlas Powder Company — dynamite E. I. duPont de Nemours & Company— booster Olin Mathieson Chemical Corporation— cap and seller of dynamite.
. Mr. Von Ludwig testified as follows:
“Q. * * * Is there anything wrong or improper in the use of a booster such as is now in the hands of the jurors?
A. No * * *.
* * * * *
“Q. And the fault you find is in the dynamite in its packaging and not in the booster, am I correct? A. I find no fault in the booster as such.
* * * # *
“Q. Is there anything in the manufacture of a booster of that type with which you find fault? A. For the purpose intended, no.
❖ * * * *
“Q. * * * But as I understand it, you have no fault that you find with the booster as such? A. Not as a manufactured product for the purpose intended, I find no fault with it. I have no evidence that' would lead me to believe that the explosion arose out of any improper loading of the booster, for example. That’s the only defect I could see, would be if you put the wrong explosive in it. I don’t believe that happened. There is nothing to indicate it happened.
“Q. Now is there anything about the packaging of * * * [the booster] which the manufacturer should have done differently that you know about? * * *
A. No. That isn’t the problem. It’s only the accepted, standard, normal procedure of its use, the necessity of punching the hole in the stick and then pushing the booster in, that I question.”
. Neither Atlas nor Olin sought indemnity or contribution from the other. If Atlas’ position were sustained and on retrial Olin were cast on the res ipsa loquitur inference of negligent dynamite, Atlas might likely have to respond to Olin for negligent manufacture since there is no suggestion that Olin did anything to alter the dynamite as delivered to it by Atlas.
. Landers v. East Texas Salt Water Disposal Co., 1952, 151 Tex. 251, 248 S.W. 2d 731, was a complete reversal of the prior Texas (and perhaps majority) rule. The Court held: “In other words, our courts seem to bave embraced the philosophy, inherent in this class of decisions, that it is better that the injured party lose all of his damages than that any of several wrongdoers should pay more of the damages than he individually and separately caused. If such has been the law, from the standpoint of justice it should not have been; if it is the law now, it will not be hereafter. The case of Sun Oil Co. v. Robicheaux is overruled. Where the tortious acts of two or more wrongdoers join to produce an indivisible injury, that is, an injury which from its nature cannot be appor*83tioned with reasonable certainty to the individual wrongdoers, all of the wrongdoers will be held jointly and severally liable for the entire damages and the injured party may proceed to judgment against any one separately or against all in one suit.” 248 S.W.2d at page 784. Burns v. Lamb, Tex.Civ.App.1958, 312 S.W.2d 730 (error ref. n. r. e.), imposes full liability for all damage on the one negligent party though damage was also done by another person who was not negligent; Leon’s Shoe Stores, Inc. v. Hornsby, Tex.Civ.App.1957, 306 S.W.2d 402, 409 (no writ history); Kirby Lumber Corp. v. Walters, Tex.Civ.App.1955, 277 S.W.2d 796 (no writ history); Western Guaranty Loan Co. v. Dean, Tex.Civ. App.1957, 309 S.W.2d 857 (error ref. n. r. e.).
. The basic instruction was as follows:
“Now, * * * under applicable facts, you are instructed that it is the law in Texas * * * that one who supplies a chattel for another to use is subject to liability to those whom the supplier should expect to use the chattel with the consent of the other or to be in the vicinity of its probable use for bodily harm and resulting damages caused by the use of the chattel in the manner for which and by a person for whose use it is supplied if the supplier (1) knows or from the facts known to him should realize that the chattel is or is likely to be dangerous for the use for which it is supplied, and (2) has no reason to believe that those for whose use the chattel is supplied will realize its dangerous condition, and (3) fails to exercise reasonable care to inform them of its dan*84gerous condition or of the facts which make said dangerous conditions likely. The duty placed upon the manufacturer or the supplier of the manufacturer’s product is to warn such persons of the existence of the dangers which could not reasonably be expected to be apparent or obvious as applied to this case.”
. The basic instruction, note 5, supra, was followed immediately by specific application to the cap as follows.
“This means that if the caps manufactured and supplied by the defendant Olin * * * were defective as claimed by plaintiff and that the storage and use of the caps in the field under weather conditions as claimed by plaintiff would cause the caps to deteriorate, become soft and more easily detonated and that the defendant Olin * * * knew or by the exercise of reasonable care should have known that said conditions claimed by the plaintiff might or would probably exist, then there was a duty and obligation on the part of the defendant Olin * * * to warn not only those to whom it supplied the caps, but also to warn those who would be handling and using said caps of such dangers as might arise by the use of said caps by persons exercising reasonable care and prudence in the use of same and the failure to so warn would amount to negligence on the part of the defendant Olin * * * and if you further find and believe from all of the evidence that such negligence by the defendant Olin Mathieson Chemical Corporation was the sole and proximate cause of the injuries, and the damages resulting therefrom, which the plaintiff received, then your verdict should be for the plaintiff was against the defendant Olin * * (emphasis supplied).
This was repeated in substantially the same words concerning the dynamite as to the defendant Atlas.
. They refer principally to the following:
“Now, under the law there can be but one recovery of damages for an injury, but the law holds all those whose negligent acts or omissions directly contributed or united in causing the injury, liable both jointly and severally to the person injured and you are instructed that if you find from a fair preponderance of the evidence that the plaintiff was injured as a proximate result of the concurrent or combined negligence of one of the defendants with that of another, or either of them, and that such injuries would not have naturally and probably resulted to plaintiff without the negligence of one of the defendants in this case, then that defendant or those defendants are liable and the plaintiff is entitled to recover.
“In other words, as applied to this case, that simply means that while you couldn’t return a verdict against each and all of these defendants for separate amounts or for the same amount that would be added together, that when you take the evidence in this case, if you find under the definitions of negligence, and so forth, that the defendants might be liable, then it is up to you to determine which defendant or which defendants, if more than one, or all of the defendants, if more than one, and whose negligence was responsible, if you find that it was responsible, for the injuries to the plaintiff, or which defendant or defendants failed to warn, and from that, if you find that to be the proximate cause, write a verdict is keeping with what you find in that respect.” (Emphasis supplied.)
. Without suggesting that the formal refinements in the submission of special issues under the Texas practice is pertinent in the submission of a case to the jury in federal court, the following Texas authorities, among others, show the fundamental substantive distinction between a proximate cause and sole proximate cause. 30-B Tex.Jur., Negligence §§ 174, 184 (1954 & 1960 Supp.); Hodges, Special Issue Submission in Texas § 22 (1959); Southland Greyhound Lines v. Cotten, 1936, 126 Tex. 596, 91 S.W.2d 326, 329-330; Panhandle & Santa Fe By. Co. v. Ray, Tex.Civ. App.1949, 221 S.W.2d 936, 941 (no writ history).