WISDOM, Circuit Judge:
This diversity suit arises out of an electrical explosion at the Kelly-Springfield Tire Company plant in Tyler, Texas. On November 6, 1970, Earl Weakley had called together Thomas McKay, Charles Baker, and Martin Kennedy for a demonstration of an “isolator switch”. At the time, they believed there was no current passing through the switch. Unhappily they were wrong. When Weak-ley moved the switch, he caused a 2800 volt current to arc, burning him over nearly half his body and burning McKay, Baker, and Kennedy less seriously. The four sued Cutler-Hammer, Inc., the manufacturer of the electrical equipment, alleging that it had designed unreasonably dangerous equipment and was liable in negligence and strict liability in tort. They also sued Fischbach and Moore, Inc., the electrical contractor that installed the equipment, charging it with negligence in failing to install required safety door switches. The defendants impleaded Goodyear Tire and Rubber Company, Kelly-Springfield’s parent company, for negligence in the design of the equipment.1 The jury, in a general verdict accompanied by answers to interrogatories, held against Fischbach and Moore both in the main action and on its third party claim against Goodyear. Fischbach and Moore appeals. We affirm.
I
The equipment involved is a part of an electrical system designed to receive, channel, and regulate electrical power supplied to a group of ten rubber mills. 138,000 volts, furnished by the local power company, enter the plant at the Main Power Sub-Station where they are stepped down by a transformer to 2300 volts. The 2300 volt charge is carried by either the “working” or the “emergency” cable to the “incoming line cubicle” where it is routed through the “isolator switch” to the “motor controllers” and thence to the 400 horsepower motors that drive the mills. This circuit, though charged by the power source, carries no current if no mill motor is running. A current is made when any mill motor is started and is broken when all motors are shut down.
The isolator switch links the incoming power cables with the line feeding power to the motor controllers. It may be set to receive incoming power from the working cable or from the emergency cable or, in the neutral position, to isolate the motor controllers from the incoming power, preventing any current from reaching them. It is, however, what is known in the trade jargon as a “no load break switch”. It has no interrupting rating and cannot be used to break an electrical current. When the circuit carries the 2300 volt current, that is when any mill motor is running, the isolator switch is “under load”. Moving the switch from one position to another while it is under load is likely to generate an electrical arc. Anyone using the switch must therefore first shut down the mills motors to break the current and must make certain that no motor can be reactivated while the switch is being moved.
Two sets of devices, particularly, are designed for that purpose. One is the set of disconnect buttons and switches on the individual motor controllers located next to the incoming line cubicle. So long as these are in the disconnect position, the mill motors cannot operate. The other is the set of safety door switches mounted inside the door of the incoming line cubicle. The safety door switches are an integral part of the control wiring of the motor controllers. They close the circuit, allowing the motors to function. When the door to the incoming line cubicle is opened, each switch opens the control circuit of its corresponding motor controller, depriving the motors of electrical power and interrupting the current flowing through the isolator switch. So long as the door [1263]*1263remains open, the mill motor cannot operate and there can be no current flowing through the circuit and the isolator switch. These switches thus assure that no one can have access to, much less operate, the isolator switch while it is under load. Because these switches operate to break the electrical power and shut down the mills, opening the incoming line cubicle door when the plant is in operation disrupts production. The isolator switch, therefore, is usually handled only on weekends or at other times when the mills are shut down.
The day of the accident seemed to Weakley, it appears, a good time to de-monstate the isolator switch to recently hired electricians. Kelly-Springfield personnel were then gearing up the plant to resume production after a long strike, and the mills, Weakley remarked at the time, were still shut down. Walking toward the switch Weakley and the others passed the mills and saw none in operation. Reaching the incoming line cubicle, they looked down the banks of motor controllers, checking the pilot lights to see if any mills were running. Having confirmed his belief that the mills were still “down”, Weakley opened the door to the cubicle and heard a bump, a noise like that made when a mill controller turns on or off — virtually the same sound in either event. There were other controllers in the area on a separate electrical system that might have made the sound, but Weakley nonetheless rechecked the pilot lights. He remarked that he must have “dropped a mill off the line” and, having already done so, he might as well proceed with the demonstration. For about five minutes he explained the function and use of the isolator switch. He indicated the door switches and explained that they deener-gized the mill motors when the door was opened and prevented them from being reactivated until the door was closed. Concluding his explanation, he moved the switch, and the explosion occurred.
Investigation after the accident revealed that there were only eight, not ten, safety door switches. For mills 11 and 12, jumper wires had been used in place of door switches to close the circuit. Thus, when Weakley opened the door to the incoming line cubicle, there were no switches to prevent mills 11 and 12 from being activated in the midst of the demonstration or to deenergize them if, in spite of the pilot lights, they had already gone into operation before Weakley opened the door.
This electrical system had been designed by Goodyear technicians, subject to approval by Kelly-Springfield. The original plan called for only eight mills, so that when Cutler-Hammer shipped the incoming line cubicle, it contained only four two-way switches, enough to accommodate the mills then contemplated. When, later that year, a decision was made to expand the plant by adding mills 11 and 12, Cutler-Hammer shipped no additional door switches. There is no evidence that any were ordered. There is no dispute, however, that the final design required that two additional switches be added and that it was the duty of Fischbach and Moore to procure and install them.
II
The appellant’s principal contention is that there was insufficient evidence to support the jury’s finding, necessary to the verdict, that Fischbach and Moore had failed to install door switches for mills 11 and 12 and had, in their stead, installed the jumper wires. We conclude, to the contrary, that the evidence, though circumstantial, was nonetheless sufficient under the Boeing standard to defeat the appellant’s motions for directed verdict and judgment notwithstanding verdict. Boeing Company v. Shipman, 5 Cir. 1969, 411 F.2d 365 (en banc).2
[1264]*1264Investigation after the accident revealed four two way switches of the type originally installed when the first eight mills were completed. The jumper wires found were of the same type of wire used by Fischbach and Moore to install related equipment, a type not normally used by Kelly-Springfield. Their discoloration, a general indicator of age, was the same as that of other wires used in the original installation in 1962. There was at the same time, no physical evidence that additional switches had ever been installed for mills 11 and 12. The original bracket had been designed to hold only four double switches and was still in place at the time of the accident. To mount an additional two-way switch would have required an additional bracket. No additional bracket was found, nor were there any screw holes where such a bracket might once have been mounted. There was testimony that if Fischbach and Moore had installed additional switches for mills 11 and 12, wires would have been run from the terminal block near the bottom of the cubicle to the switches near its top, and that, if the switch had been subsequently removed, these additional wires would likely have been jumpered together near the original location of the removed switch. No such wires were found after the accident.
Fischbach and Moore relied principally on the testimony of its employee M. B. Cooper, who had superintended the installation of electrical equipment associated with mills 11 and 12. He testified by deposition that a Mr. Kirksey had installed and tested door switches for mills 11 and 12 under his supervision. Although no one could directly contra-diet this testimony, no one corroborated it. Cooper conceded that Kirksey, who did not testify, had no recollection of installing or testing the switches. Cooper’s own testimony was uncertain. He could not, in his first deposition, recall exactly how the additional switches had been added. He had, he said, either piggy-backed a second Cutler-Hammer two-way switch onto one already in place or had replaced a two-way switch with a four-way switch. He thought he had been given the switch by Goodyear’s Joe McDowell. Cutler-Hammer’s designer, C. J. Fellows, testified by deposition that he was unaware of any way these switches could be piggy-backed. Confronted with this testimony, Cooper’s memory sharpened in his second deposition. He had, he testified, replaced a two-way switch with a four-way switch given him by Kelly-Springfield’s Kenneth Schneider. The jury was of course not required to believe Cooper’s testimony.
Fischbach and Moore suggested that a four-way switch might have been removed and replaced by a two-way switch. All the Kelly-Springfield electricians who testified denied having ever removed or replaced a door switch in the incoming line cubicle, and the day book kept by the maintenance department showed no entry indicating either removal of any door switch or installation of jumpers. The suggestion that jumpers might have been used by Kelly-Springfield employees to prevent disruption of production if the door to the cubicle was accidentally opened, did not account for the alleged removal of a door switch. The jumpers would have by-passed the switch, even if it had been left in place.
[1265]*1265We are unable to say that there is no “evidence of such quality and weight that reasonable men might [not] reach different conclusions” as to whether or not Fischbach and Moore had failed to install the door switches and had installed the jumper wires. We conclude that there was a sufficient “conflict in substantial evidence” under the Boeing standard to raise a jury question. 411 F.2d at 374.
Ill
Fischbach and Moore next cites as reversible error a portion of the court’s instructions to the jury on the defense of volenti non fit injuria. The instruction on the four elements of vo-lenti was unexceptionable: (1) knowledge of facts constituting a dangerous condition or activity; (2) knowledge that the condition or activity was dangerous; (3) appreciation of the nature or extent of the danger; and (4) voluntary exposure to the danger. Myers v. Day & Zimmerman, Inc., 5 Cir. 1970, 427 F.2d 248; J. & W. Corp. v. Ball, Tex.Sup.Ct. 1967, 414 S.W.2d 143. The challenged portion of the charge read as follows:
In connection with determining whether any Plaintiff’s exposure was voluntary, you are instructed that a person is not at fault in voluntarily exposing himself to a known and appreciated danger if under the same or similar circumstances an ordinarily prudent person would have incurred the risk of such conduct involved. Thus, if there was some reasonable necessity or propriety which justified the Plaintiff in exposing himself to the known risks involved, if any, or if by the exercise of care proportionate to the danger, then such Plaintiff cannot be found to have voluntarily exposed himself to the risk.
This, as the appellees point out, is a duplicate of the charge approved by this Court in Messick v. General Motors, 5 Cir. 1972, 460 F.2d 485, 493, as an accurate statement of Texas law. Our approval was an “Erie prognostication” of how the Texas Supreme Court would apply the defense of volenti to a strict liability claim. We surmised that it would follow the Second Restatement of Torts by incorporating unreasonableness as a “subelement of voluntariness”. See Rest.2d Torts § 402A, comment (n); Note, The Fifth Circuit Predicts Relaxation of the Texas Volenti Doctrine in Strict Liability Cases, 51 Tex.L.Rev. 376 (1973). We approved this charge, however, only in relation to volenti interposed as a defense to a strict liability claim. We expressly noted that in a negligence action, “the reasonableness of the undertaking is not available to relieve a plaintiff from the application of the [volenti] doctrine.” 460 F.2d at 489. Accord Halepeska v. Callihan Interests, Inc., Tex.Sup.Ct.1963, 371 S.W.2d 368; Rest.2d Torts § 496. This charge was erroneous, then, in connection with the plaintiffs’ negligence action against Fischbach and Moore. But this is not all. The Texas Supreme Court has recently, in considered dictum, disavowed our “Erie prognostication”. Henderson v. Ford Motor Co., Tex.1974, 519 S.W.2d 87. Unreasonableness is not, the Court concluded, a necessary component of the element of voluntariness, even when vo-lenti is asserted as a defense to a strict liability claim. The Messick volenti charge is thus not only inapposite as applied to the negligence action against Fischbach and Moore but also incorrect, we now know, as a matter of Texas law.
The error in this charge does not, however, warrant reversal. The appellant failed, we conclude, to carry its burden of proving an essential element of the volenti defense, so that the appellant was not entitled to a volenti instruction on the facts before us. The error in the instruction given did not therefore “affect the substantial rights” of the appellant and must be considered harmless. Fed.R.Civ.P., Rule 61.
The heart of the volenti defense is knowledge of the circumstances that create the danger and knowledge and appreciation of the danger itself. All the plaintiffs testified, however, that they were unaware of the substitution of [1266]*1266jumper wires for the door switches. Their conduct before the accident tends to support that testimony. Indeed, Weakley had explained just before the accident that the safety door switches assured that the isolator switch could not be under load. The jury was not required, of course to believe the plaintiffs, but it could not, at the same time, predicate a finding of volenti on an adverse credibility choice alone. Myers v. Day & Zimmerman, Inc., 5 Cir. 1970, 427 F.2d 248; Rice v. Gulf States Paint Co., Tex. Civ.App.1966, 406 S.W.2d 273.
Under Texas law, volenti is an affirmative defense. Fischbach and Moore thus had the burden of proving all its essential, elements. Myers v. Day & Zimmerman, Inc., 5 Cir. 1970, 427 F.2d 248; Rabb v. Coleman, Tex.Sup.Ct.1971, 469 S.W.2d 384. It has not. met this burden. Canvassing the entire record, we find no evidence that Weakley or any of his co-plaintiffs had actual knowledge that the door switches were missing. There was evidence that Weakley might have known but had forgotten that the mills were scheduled to go into production on the night of the accident; evidence that the pilot lights were unreliable; and evidence that the sound Weak-ley heard when he opened the door to the incoming line cubicle put. him on notice that a mill had been operating. This and other evidence was certainly sufficient to raise the issue of contributory negligence, but it was not the evidence of actual knowledge of the specific danger required to raise a volenti issue.
The appellant also argues that the danger was so obvious that knowledge and appreciation of it should have been imputed to the appellees and volenti held to have been established as a matter of law. This contention is pressed with an appropriate want of conviction. It lacks merit.
The appellant also advances the related contention that contributory negligence was established as a matter of law. The essence of this argument is that Weakley acted unreasonably in failing to open the disconnect switches on the mill controllers before attempting to move the isolator switch. The appellees rejoinder is that Weakley acted reasonably in the circumstances in relying on the door switches alone. The testimony in this regard was conflicting. Some supervisory personnel testified that caution required the operator to use the disconnect switches, but other electricians disagreed and testified that they often relied on the door switches alone. There was also conflicting testimony as to whether there had been any written safety instructions including an instruction directing use of the disconnect switches. There was no conflict, however, in the testimony concerning the reliability and durability of the door switches. There was an unchallenged consensus among the witnesses that those switches were long-lived and most unlikely to fail. The issue of Weakley’s contributory negligence was, we conclude, properly submitted to the jury.
IV
The appellant also launches a broad attack on the damages awarded Earl Weakley. They were, indeed, substantial. The judgment totaled $595,184. This represented $23,864 in stipulated past medical expenses, $40,000 future medical expenses, $40,000 lost earnings to the date of trial, $90,000 past pain and suffering, $100,000 future pain and suffering, and $300,000 for lost earning capacity, considering both future productivity and future inflation. The appellant contends specifically that future inflation is too speculative a basis for assessing future damages.
This Court has writ large its disapproval of calculating future damages by reference to predictions of future inflation. Johnson & Starnes v. Penrod Drilling Co., 5 Cir. 1975, 510 F.2d 234 (en banc). Numerous other federal decisions are to the same effect. See Williams v. United States, 1 Cir. 1970, 435 F.2d 804; Petition of United States Steel Corp. v. Lamp, 6 Cir. 1970, 436 F.2d 1256, cert. denied 402 U.S. 987, 91 S.Ct. 1649, 29 L.Ed.2d 153; Sleeman v. Chesapeake & Ohio Ry. Co., 6 Cir. 1969, 414 F.2d 305; [1267]*1267McWeeney v. New York, N. H. & H. R.R., 2 Cir. 1960, 282 F.2d 34. These cases all, however, involved claims asserted on federal causes of action. Here, on the other hand, we are controlled by Erie. This and other federal courts have concluded that under Erie state law governs the measure of damages, including the admissability and jury consideration of evidence of inflation. See New Amsterdam Casualty Co. v. Soileau, 5 Cir. 1948, 167 F.2d 767, cert. denied 335 U.S. 822, 69 S.Ct. 45, 93 L.Ed. 376. Accord Magill v. Westinghouse Electric Corp., 3 Cir. 1972, 464 F.2d 294; Willmore v. Hertz, 6 Cir. 1971, 437 F.2d 357; Tullos v. Corley, 6 Cir. 1964, 337 F.2d 884. See also Rule 43(a), Fed.R.Civ.P.
Texas law thus controls. Although the Texas Supreme Court has not to our knowledge considered the question, the intermediate appellate courts have uniformly permitted juries to weigh evidence of future inflation in setting awards for future damages. See Halliburton Co. v. Olivas, Tex.Civ.App. 1974, 517 S.W.2d 349; Hammond v. Stricklen, Tex.Civ.App.1973, 498 S.W.2d 356; J. A. Robinson Sons, Inc. v. Ellis, Tex.Civ.App.1967, 412 S.W.2d 728; City of Austin v. Hoffman, Tex.Civ.App.1964, 379 S.W.2d 103; Texas Consolidated Transportation Co. v. Eubanks, Tex.Civ. App.1960, 340 S.W.2d 830; Henwood v. Moore, Tex.Civ.App.1947, 203 S.W.2d 973. We have neither found nor been cited to any chink in this consistent line of authority. We must hold, therefore, that the trial court did not err in permitting the jury to consider the evidence of future inflation. We also conclude that the district court did not abuse its discretion in allowing the jury’s assessment of damages to stand.
V
The appellant’s remaining arguments 3 relate to the judgment against it on its third-party negligence action against Goodyear Tire and Rubber Company. Of the grounds urged at trial in support of this claim, the appellant preserves and we need consider only the charge that Goodyear electrical engineers were negligent in designing equipment dangerous for its intended uses. Our review of the record persuades us that Fischbach and Moore failed to shoulder its burden of proving the design unreasonably dangerous.
Fischbach and Moore rely principally on an attempted showing that there were design alternatives which, had they been adopted, would have prevented the accident or minimized its consequences. Such a showing, by itself, is insufficient to establish liability. Henderson v. Ford Motor Co., Tex.Sup. Ct.1974, 519 S.W.2d 87, 93; Johnson v. Murray Co., Tex.Civ.App.1936, 90 S.W.2d 920, 925—26; 40 Tex.Jur.2d § 89 at 593 (1962). See Muncy v. General Motors Corp., 5 Cir. 1966, 367 F.2d 493; Muncy v. General Motors Corp., Tex.Civ.App. 1963, 357 S.W.2d 430. It is one thing to show that the defendant might have designed a safer product; quite another to show that the product he did design was unreasonably dangerous. The defendant is not obliged to design the safest possible product, or one as safe as others make or a safer product than the one he has designed, so long as the design he has adopted is reasonably safe. W. Pros[1268]*1268ser, The Law of Torts § 96 at 645 (4th ed. 1971). Accord, Henderson v. Ford Motor Co., Tex.Sup.Ct.1974, 519 S.W.2d 87; Muncy v. General Motors Corp., 5 Cir. 1966, 367 F.2d 493; Muncy v. General Motors Corp., Tex.Civ.App.1963, 357 S.W.2d 430; Johnson v. Murray Co., Tex. Civ.App.1936, 90 S.W.2d 920; F. Harper and F. James, The Law of Torts § 28.5 at 1544-45 (1956). There was substantial evidence that the final design was indeed reasonably safe.
The isolator switch was designed to be used only after the circuit had been broken by some other means. It was, moreover, designed for use only by skilled electricians who would know the means they could use to break the circuit. All the plaintiffs conceded that they knew that the isolator switch was, by definition, a no load switch, dangerous to operate under load, and that they knew the steps that could be taken to assure there would be no current. Still, the Goodyear specifications required that Cutler-Hammer provide operating instructions for the use of the equipment — an obligation the jury found to have been assumed, if not fulfilled, by Kelly-Springfield. Proper instructions, if followed, would indisputably have prevented the accident.
Exposure to the isolator switch when it might be under load was minimal. The switch was rarely used; and then, usually, only on weekends or at other times when the plant was shut down. The electricians knew that to avoid disruption of production they should use the switch during production only when absolutely necessary. Such occasions, according to Kelly-Springfield’s manager of engineering, arose perhaps once a year.
Although the danger was obvious and the exposure to it minimal, the Goodyear design nonetheless incorporated various safety features to prevent accidental use of the switch while it was under load. The motor controllers were equipped with pilot lights to indicate whether the corresponding mill motors were operating. Next to these lights, and only a step away from the incoming line cubicle, were disconnect switches that could be used to de-energize the motors and thus break the current. Finally, there were the safety door switches. There was no dispute that these switches, properly installed and operating, assured that the isolator switch could not be reached, much less operated, while it was under load. No one contested the durability and dependability of these switches. They were rarely used, whereas similar switches operated in other equipment as often as 15,000 a day. Kelly-Springfield’s former general maintenance foreman, William Norris, testified that he had never replaced a door switch or known one to fail. Other Kelly-Springfield electricians testified to the same effect. Cutler-Hammer’s W. J. Fellows, who was responsible for the final design of the incoming line cubicle, had never heard of a door switch failure. There had never before been any accident in an incoming line cubicle at the Kelly-Springfield plant, although there were a number of similar cubicles.
Testimony at trial revealed a consensus among electricians and designers that the equipment in question was safe if properly installed. There was also un-contradicted testimony that the design of the isolator switch was consistent with good engineering practice, that it complied in all respects with the requirements of the National Electric Safety Code, and that similar door switches were in common use in the industry. Finally, electricians and designers were agreed that the substitution of jumper wires for the door switches was a substantial and unjustifiable departure from design requirements. They also agreed that they would not have foreseen such a departure.
Against this evidence that the equipment was reasonably safe as designed, Fischbach and Moore mustered little contrary evidence. It produced no evidence that the design deviated from accepted industrial practice. See, e. g., Otis Elevator Company v. Wood, Tex.Sup.Ct. 1968, 436 S.W.2d 324. It produced no evidence that its suggested design alter[1269]*1269natives had been adopted by other designers for use in similar equipment. See, e. g., Otis Elevator Company v. Wood, Tex.Sup.Ct.1968, 436 S.W.2d 324; South Austin Drive-In Theatre v. Thomison, Tex.Civ.App.1967, 421 S.W.2d 933. It produced no expert testimony to support its contention that the design was unreasonably dangerous. Finally, it failed to produce evidence of previous accidents or previous switch failures.
Fischbach and Moore’s sparse evidence of negligent design is, we conclude, overwhelmed by Goodyear’s evidence that its design was reasonably safe. Fischbach and Moore produced no evidence of such quality and weight that reasonable men might conclude that Goodyear had been negligent in designing unreasonably dangerous equipment. Goodyear’s motion for a directed verdict should, accordingly, have been granted. Boeing Company v. Shipman, 411 F.2d at 374.
This conclusion makes it unnecessary for us to consider the appellant’s remaining contentions with respect to its third-party claim.
The judgment of the district court is affirmed.