COULSON, Justice.
This is an appeal from an order sustaining a plea of privilege in a products liability case. Appellant Robert Turner sued General Motors Corporation, appellee, and Raymond Kliesing d/b/a Kliesing Motor Company in strict liability in tort for personal injuries received when his car rolled over in an accident. Suit was brought in Brazoria Cqunty, where Turner purchased the automobile from Kliesing Motors. General Motors filed a plea 'of privilege to be sued in the county of its residence in Texas, Harris County. Turner filed a controverting affidavit asserting that venue for General Motors was proper in Brazo-ria County, on the basis of Subdivision 4 [499]*499of Article 1995, Vernon’s Tex.Rev.Civ. Stat.Ann. (1964). After a plea of privilege hearing, the trial judge sustained General Motors’ plea and filed findings of fact and conclusions of law.
Subdivision 4 of Article 1995 states in pertinent part: “If two or more defendants reside in different counties, suit may be brought in any county where one of the defendants resides.” The necessary venue facts under Subdivision 4 are: 1) one defendant resides in the county of suit; 2) the party making the plea of privilege is at least a proper party; 3) the plaintiff has a bona fide claim against the resident defendant. O. P. Leonard Trust v. Hare, 305 S.W.2d 833 (Tex.Civ.App. Texarkana 1957, writ dism’d). It is uncontested that the first two venue facts are present in this case. The third venue fact consists of proving by a preponderance of the evidence each element of a bona fide claim against the resident defendant. Stockyards Nat. Bank v. Maples, 95 S.W.2d 1300 (Tex.Comm’n App.1936, opinion adopted). Turner sued General Motors and Kliesing in strict liability in tort for a defectively designed roof on his automobile which enhanced his injuries, but did not cause the accident. The trial judge, in his findings of fact, found all the facts necessary for this alleged cause of action, if such a cause of action does exist; the sustaining of the plea of privilege was, in effect, a holding that no such cause of action exists in Texas.
The question here is whether a manufacturer and retailer may be held strictly liable in tort for a defectively designed automobile which enhances the injuries of plaintiff, but does not cause the accident.
At the plea of privilege hearing, Turner testified that in April of 1971 he was driving on a two-lane farm-to-market road in his 1969 four-door, Chevrolet Impala, hardtop sedan with a center post. He was following a truck, which started to pull onto the right shoulder. Turner accelerated to fifty or sixty miles per hour to pass the truck, but the truck attempted to make a left-hand turn when Turner came up to it. Turner pulled to the right to avoid a collision and left the road. When he attempted to return to the road, he overturned and the car landed on its top. Turner estimated his speed immediately before the roll-over at twenty to thirty miles per hour. Turner’s seat belt was buckled, but the right-front portion of the roof collapsed and came into contact with his head. This contact paralyzed Turner’s hands and legs.
Mr. James Barron, called as an expert witness by Turner, stated that he had worked as a design engineer for the Chevrolet Division of General Motors from 1963 to 1965, and had then worked in the same capacity for American Motors for over five years. Barron testified that General Motors designed for future production five years in advance and that he was involved in the design of the 1969 Impala. Barron informed his superiors at General Motors of the desirability of putting a roll bar in the roof of their cars, and Barron worked on a roll bar program. He testified that a roll bar would be expensive as an option, but relatively inexpensive if put on all cars at their birth on the assembly line. The roll bar program was discontinued, and Barron was told this was due “primarily [to] cost reasons and cost in conjunction with the fact that the consumer could not see what he was paying for.” His supervisor told him that “it is difficult to pass on something to the consumer and charge him money for it if he cannot see it.”
Barron testified that it would be impossible to design a crash proof car. He defined the term “crashworthiness” as the ability of a car “to withstand normal hazard conditions.” Crashworthiness was broken down into the following categories: the structural integrity of the car’s shell; the elimination of sharp or protruding objects in the interior; passenger restraint devices; and the elimination of post-crash fire. Barron estimated that, in the context [500]*500of all possible types of accidents, roll-overs occur in twenty percent of all accidents involving “principal” injuries.
Barron drew a diagram of the Impala’s roof structure and termed the roof “cosmetic,” in that it provided protection from sun and rain, but it would not provide adequate protection in an overturn regardless of the speed. He categorized the roof as definitely defective, “uncommonly dangerous,” and “unreasonably dangerous.” The design of the Impala’s roof was called perhaps the weakest way to design a roof, and Barron said that all of the roof structure in this Impala had collapsed. There was nothing in the roof of the car which would support the car in an overturn.
Barron suggested that there were many alternative ways to design a roof more safely and specifically proposed the roll bar or roll cage (the latter is, in effect, a connected double roll bar forming a rectangle with a bar at each corner attaching the frame to the body of the car). Roll bars and roll cages had been known to Barron since 1952, and he stated that General Motors put them on test cars and racing cars. In Mr. Barron’s opinion, roll bars would greatly minimize roll-over injuries. Barron admitted that no mass-produced automobile in the United States had ever come equipped with a roll bar or roll cage and conceded that the Impala’s roof was no more dangerous than the roof in any other car produced at that time. He frankly stated that he considered the roofs on all American cars defectively designed, including those currently manufactured (West Germany’s Porsche Targa was the only production car cited by Barron as being equipped with a roll bar).
Raymond Kliesing, the defendant dealer, testified that, based upon his forty-five years of sales experience, the average consumer believes that a sedan vehicle will be a reasonably safe product in a roll-over.
The trial judge sustained General Motors’ plea of privilege and filed findings of fact and conclusions of law. The trial judge found that Kliesing was an authorized General Motors dealer, that Turner’s roof collapsed and injured him in the accident, and that the auto immediately before the accident was in substantially the same condition it was in when sold. The court’s crucial finding is that the car was
defectively designed in that the roof was not a sufficient structural support to prevent the roof from collapsing and thereby injuriously encroaching into the passenger compartment in the event of an overturn of the automobile and this defective design rendered the automobile unreasonably dangerous to the user or consumer, i. e., dangerous to an extent beyond that which would be contemplated by the ordinary user or consumer with the knowledge available to him as to the characteristics of a 1969 Chevrolet four-door sedan with a center post.
The defect was found by the trial judge to be a proximate and producing cause of the injuries, and the possibility of overturn accidents was held to be clearly foreseeable by General Motors.
The trial judge’s sustaining of General Motors’ plea of privilege was a ruling, in effect, that Turner had failed to prove a bona fide claim against Kliesing. In light of the trial judge’s findings of fact, it is clear that his implicit conclusion of law was that strict tort liability in Texas does not encompass the liability of a manufacturer or retailer of a defective product when the defect enhances the injuries of á plaintiff, but plays no role in causing the accident. The question thus presented is one of first impression in Texas.
The genesis of this issue can be found in Evans v. General Motors Corporation, 359 F.2d 822 (7th Cir. 1966) (applying Indiana law), cert. den. 385 U.S. 836, 87 S.Ct. 83, 17 L.Ed.2d 70. There the plaintiff sued under general negligence, strict liability, and implied warranty principles with the argument that the manufacturer of his automobile should be liable for his enhanced injuries due to the “X” frame without side [501]*501rails on his car, despite the fact that this defect had not caused the collision. The Seventh Circuit held that the manufacturer owed no duty to design or make an accident-proof or fool-proof car. The court also said that the intended purpose of a car does not include its participation in collisions. In Larsen v. General Motors Corporation, 391 F.2d 495 (8th Cir. 1968) (applying Michigan law), the plaintiff claimed under negligence principles that his car was defectively designed in that the steering shaft protruded in front of the axle so that his injuries were enhanced when his car was struck in the left front. The Eighth Circuit said that an auto manufacturer has a duty to design and construct its product to be reasonably fit for its intended use and free of hidden defects which would render it unsafe for that use. The court found that the real issue was one of intended use and said that the intended use of an automobile necessarily entails the risk of injury-producing accidents; such injuries are foreseeable as an incident to the normal and expected use of a car. Since Evans and Larsen, more than twenty jurisdictions have addressed the issue of crashworthiness and have split evenly. See Frericks v. General Motors Corp., 317 A.2d 494 (Md.App.1974), for a collection of cases by jurisdiction. We are persuaded by the logic of Larsen.
The evolution of strict liability in Texas has proceeded steadily since the Supreme Court, in McKisson v. Sales Affiliates, Inc., 416 S.W.2d 787 (Tex.Sup.1967), extended its scope beyond food for' human consumption and adopted Section 402A of the Restatement (Second) of Torts (1965). Section 402A provides:
(1)One who sells any product in a defective condition unreasonably dangerous to the user or consumer or to his property is subject to liability for physical harm thereby caused to the ultimate user or consumer or to his property, if
(a) the seller is engaged in the business of selling such a product, and
(b) it is expected to and does reach the user or consumer without substantial change in the condition in which it is sold.
(2) The rule stated in Subsection (1) applies although
(a) the seller has exercised all possible care in the preparation and sale of his product, and
(b) the user or consumer has not bought the product from or entered into any contractual relation with the seller.
The doctrine was extended to bystanders as well as ultimate consumers. Darryl v. Ford Motor Company, 440 S.W.2d 630 (Tex.Sup.1969). Strict liability has been held to encompass products which have been defectively designed. Pizza Inn, Inc. v. Tiffany, 454 S.W.2d 420 (Tex.Civ.App.-Waco 1970, no writ). Lessees are now protected by the doctrine. Rourke v. Garza, 511 S.W.2d 331 (Tex.Civ.App.-Houston (1st Dist.), 1974.)
Before addressing the merits of the Evans-Larsen controversy, it is necessary to discuss several Texas cases which General Motors feels have implicitly rejected the doctrine of crashworthiness. In Muncy v. General Motors Corp., 357 S.W.2d 430 (Tex.Civ.App.-Dallas 1962, no writ), an automobile was alleged to be negligently designed because the key could be removed from the ignition while the car was in gear and the engine was running. The court held that the appellant “was not using the car in the manner and for the purpose for which it was intended — at least appellants have failed to prove that the car was being so used.” Id. at 435. Kahn v. Chrysler Corporation, 221 F.Supp. 677 (S.D.Tex.1963), involved an allegation of negligent design in regard to sharp, protruding tail-fins on a car. The plaintiff was a boy who had ridden his bicycle into a fin. The court, relying heavily on Muncy, held that the “duty of the automobile manufacturer extends to the ordinary use of the vehicle. [502]*502. But the manufacturer has no obligation to so design his automobile that it will be safe for a child to ride his bicycle into it while the car is parked.” Id. at 679. Because of the Supreme Court’s decision in Otis Elevator Company v. Wood, 436 S.W.2d 324 (Tex.Sup.1968), which will be discussed infra, we believe that both these decisions have taken an overly narrow and restrictive view of the standard of “intended use.” General Motors also cites Kerby v. Abilene Christian College, 503 S.W.2d 526 (Tex.Sup.1973), which involved a finding of contributory negligence on the part of the plaintiff in failing to close the sliding door of his van; this act had substantially increased the plaintiff’s injuries but had played no part in causing the accident. The court held, 503 S.W.2d at 528:
We draw a sharp distinction between negligence contributing to the accident and negligence contributing to the damages sustained. Contributory negligence must have the causal connection with the accident that but for the conduct the accident would not have happened. Negligence that merely increases or adds to the extent of the loss or injury occasioned by another’s negligence is not such contributory negligence as will defeat recovery.
This holding explicitly restricts the basic scope of contributory negligence, does not deal with actionable primary negligence, and we think, simply expresses the Court’s rejection of the harsh doctrine of denying a plaintiff any recovery because of an omission playing no part in causing the accident from which his injuries flow. General Motors would construe Kerby to stand for the general proposition that pre-exist-ing negligence which causes injury is not actionable if another’s intervening negligence specifically sparks the accident. This is not the law of Texas. See Robert R. Walker, Inc. v. Burgdorf, 150 Tex. 603, 244 S.W.2d 506 (1951).
General Motors cites one case applying Texas law which does deal with the issue of defective design and crashworthiness. In Willis v. Chrysler Corporation, 264 F. Supp. 1010 (S.D.Tex.1967), plaintiff was traveling at seventy miles per hour and was struck head-on by another car moving at an undetermined but extremely high rate of speed. The plaintiff’s auto broke completely in two, directly behind the front seat, and suit was brought for breach of the manufacturer’s warranty. The court held that a manufacturer has “no duty to design an automobile that could withstand a high speed collision and maintain its structural integrity.” The court then, agreed with Evans that the. intended purpose of a car does not include its participation in collisions. While we simply disagree with the court’s acceptance of Evans as the law of Texas, we do not believe that Larsen imposes a duty to design an automobile which will withstand the type of high-speed, head-on collision described in Willis.
We find pertinent and agree with South Austin Drive-In Theatre v. Thomison, 421 S.W.2d 933 (Tex.Civ.App.-Austin 1967, writ ref’d n. r. e.). In that case, the driver of a riding lawn mower backed the vehicle into a child and knocked him down. The driver attempted to pull the boy away, but found him pinned by the machine. The boy lost a leg. The court rejected the manufacturer’s arguments of foreseeability, misuse, and industry custom. The manufacturer was held to have negligently designed the mower in failing to have a guard over the drive chain and gear sprocket, despite the fact that this design had not caused the accident. The court said, 421 S.W.2d at 949:
We believe it is a correct statement of the law to say that [the manufacturer] owed a duty to use reasonable care in the design and manufacture of its power mower to prevent injury to the user and to persons [the manufacturer] should reasonably expect to be in the vicinity of the mower’s probable use.
Preliminary to a discussion of the question of crashworthiness, it must be noted [503]*503that the issues and arguments pertinent to the question are essentially the same whether suit is brought under general negligence or under strict tort liability. Most of the decisions from other jurisdictions involving crashworthiness were brought under negligence, but they are nevertheless applicable here. The issue of a manufacturer’s exercise of due care under negligence becomes the issue of whether he has put an unreasonably dangerous product into the stream of commerce under strict liability, and the issues of foreseeability and intended use under negligence are transformed into the issue of normal use under strict liability.
General Motors concedes that it could foresee that its products would be involved in accidents of an infinite variety, including roll-overs. Its argument is that foreseeability cannot be equated with duty. General Motors characterizes the Larsen argument as being that, since car accidents can be foreseen, there is a duty to design cars to reduce the injuries from these accidents. If duty were commensurate with foreseeability, then an automobile manufacturer would be an insurer. Hoenig & Werber, Automobile “Crashworthiness”: An Untenable Doctrine, 20 Clev.St.L.Rev. 578, 589-90 (1971). However, duty has never properly been defined solely by foreseeability. Green, Foreseeability in Negligence Law, 61 Colum.L.Rev. 1401, 1417-18 (1961). This argument was expressed by the Seventh Circuit in Evans as follows, 359 F.2d at 825:
The intended purpose of an automobile does not include its participation in collisions with other objects, despite the manufacturer’s ability to foresee the possibility that such collisions may occur. As defendant argues, the defendant also knows that its automobiles may be driven into bodies of water, but it is not suggested that defendant has a duty to equip them with pontoons.
We agree that foreseeability alone cannot define an automobile manufacturer’s duty; this would create the duty to design a crash-proof car. However, Larsen seeks merely to hold the manufacturer to the duty of designing a crashworthy vehicle. While all agree that there is no duty to design a crash-proof car, one court has termed it a “non sequitur” to use this truism as a basis for thus saying there is no duty to design a crashworthy car. Badorek v. General Motors Corporation, 11 Cal.App.3d 902, 919, 90 Cal.Rptr. 305, 316 (1970). The rule of Larsen is not grounded upon foreseeability, but upon the %mrea-sonable risk of injury in the event of a collision. Under strict liability the question of liability in each case turns upon whether the product is in a defective condition which is “unreasonably dangerous.”
The controlling issue in the crash-worthiness inquiry is not foreseeability, but intended use. General Motors argues that the normal and proper use of an automobile does not include its participation in collisions. The Evans court said that collisions are not an intended purpose of the automobile even though the manufacturer may foresee the “possibility” of collisions. The Larsen court terms collisions a “probability;” the court stated that between one-fourth and two-thirds of all automobiles are involved during their life in a collision producing injury or death, according to O’Connell, Taming the Automobile, 58 Nw.U.L.Rev. 299, 348 (1963). It is irrelevant whether the occurrence of an accident involving a particular car is a possibility or a probability. What is germane is the fact that collisions are so frequent and common that they must be considered an unavoidable incidence of the normal and proper use of automobiles. Misuse of an automobile may occur, such as the intentional use of a car as a bulldozer. However, the normal use of an automobile is to transport people safely over the public roads. It cannot be argued that a manufacturer should not be liable for a defect which causes an accident, and no logical distinction can be drawn between that situation and one in which a defect causes injury in a foreseea[504]*504ble accident. We hold that an automobile manufacturer may be held strictly liable for a defective design which produces injuries, but not the accident.
The Supreme Court in Otis Elevator Company v. Wood, supra, gave an expansive reading to the concept of intended use. There the plaintiff went to the rescue of a small child by leaning over an escalator rail, which pulled the plaintiff in such a manner that she became suspended between the wall of the escalator and the second-floor balcony railing. The Court upheld the jury’s findings of negligent design and addressed the argument that the plaintiff was not putting the escalator to its intended use. The Court rejected this argument, relying heavily upon Spruill v. Boyle-Midway, Incorporated, 308 F.2d 79 (4th Cir. 1962). In Spruill a boy had died from drinking furniture polish, and the court rejected the manufacturer’s defense of intended use. The Court in Otis, supra, 436 S.W.2d at 328, quoted the following language from Spruill, 308 F.2d at 83-84:
“Intended use” is but a convenient adaptation of the basic test of “reasonable foreseeability” framed to more specifically fit the factual situations out of which arise questions of a manufacturer’s liability for negligence. Normally a seller or manufacturer is entitled to anticipate that the product he deals in will be used only for the purpose for which it is manufactured and sold; thus he is expected to reasonably foresee only injuries arising in the course of such use.
However, he must also be expected to anticipate the environment which is normal for the use of his product and where, as here, that environment is the home, he must anticipate the reasonably foreseeable risks of the use of his product in such an environment. These are risks which are inherent in the proper use for which is product is manufactured.
The Supreme Court then expressly approved of this view, thereby adopting what might be termed the “environment” definition of intended use. The public roads and highways are the natural environment of automobiles. Normal use under strict liability cannot logically be given a more restrictive meaning than intended use.
The doctrine of crashworthiness does not make insurers of automobile manufacturers. The Fourth Circuit, in adopting Larsen, noted that the question of whether a manufacturer had created an unreasonable risk of injury involves a traditional balancing of the gravity and likelihood of harm against the burden of precautions to avoid the harm. Dreisonstok v. Volkswagenwerk, A. G., 489 F.2d 1066 (4th Cir. 1974) (applying Virginia law). The court stated that the burden of taking precautions included a consideration of the particular purpose of the vehicle, the style or aesthetic appeal of the model, and the cost of the change and of the vehicle. So viewed, it is obvious that manufacturers are not required to produce the safest possible car, but only a reasonably safe one. For instance, if a change in design would add little to safety, render the vehicle ugly or inappropriate for its particular purpose, and add a small fortune to the purchase price, then a court should rule as a matter of law that the manufacturer had not created an unreasonable risk of harm.
The same type of balancing of factors takes place under strict liability. Section 402A states that a manufacturer is strictly liable for a product in a defective condition “unreasonably dangerous” to the user or consumer. Comment i to § 402A defines “unreasonably dangerous” as “dangerous to an extent beyond that which would be contemplated by the ordinary consumer who purchases it, with the ordinary knowledge common to the community as to its characteristics.” The same type of balancing process contemplated by the Dreisonstok crashworthy case was engaged in by this Court in the products liability case of Metal Window Products Co. v. [505]*505Magnusen, 485 S.W.2d 355 (Tex.Civ.App.-Houston [14th Dist.] 1972, writ ref’d n. r. e.). There the trial court, sitting without a jury, had found a manufacturer liable in negligence and strict liability when the plaintiff had walked into a sliding glass door and sustained injuries, although the glass had not shattered. Under § 402A liability was seen as predicated upon an initial finding of defect, and then a finding that the degree of danger rose to the level of “unreasonable.” The plaintiff asserted that the door was defectively designed in failing to have decals or a bar, but this Court doubted that this could be considered a defect at all, because such an addition would change the very nature of glass doors and destroy the illusion of spaciousness which had caused the doors to be so popular with the public. Assuming ar-guendo the existence of a defect, we said, 485 S.W.2d at 358:
[Gjiven the popularity and general acceptance of clear glass doors, it must be considered doubtful that the risk of collision without breakage due to the transparency involved outweighs the utility and value such doors have attained.
Adopting the test of Comment i as to what a reasonable consumer would expect, we stated, 485 S.W.2d at 358:
In light of the extensive use of glass doors and common knowledge as to the possibility and frequent occurrence of collisions with them, a reasonable user must be held to appreciate the risk inherent in them.
This Court held, in effect, that the door was neither defectively designed nor unreasonably dangerous.
When Comment . i is applied to crashworthiness, Dreisonstok’s balancing test for negligence governs characteristics of a product which lie within the ordinary knowledge of the community. Holford, The Limits of Strict Liability for Product Design and Manufacture, 52 Texas L.Rev. 81, 95-96 (1973). This approach would reply to the Evans’ pontoon argument that, while the average consumer as a matter of law does not expect his car to float, he may well expect the roof of that car to maintain its structural integrity in a rollover accident.
The moderate approach of Dreisonstok to the issue of crashworthiness should dispel General Motors’ fears of absolute liability. This approach may be contrasted with that of Cronin v. J. B. E. Olson Corporation, 8 Cal.2d 121, 501 P.2d 1153, 104 Cal.Rptr. 433 (1972). There, a clasp holding bread trays on a delivery truck failed to hold the trays in a collision, sending the driver through the windshield. The plaintiff recovered under his allegation of defective design in strict liability. The court adopted Larsen and then stated that it was not necessary for a plaintiff to establish that the product was unreasonably dangerous after he had shown a defect and causation, because such a requirement would place too onerous a burden upon plaintiffs. Such an approach to crashworthiness does indeed make insurers of manufacturers.
The courts adopting Evans often say that safety standards must be left to Congress, because courts lack the expertise for dealing with such complex matters, because sporadic and ad hoc court decisions will result in wrong and even contradictory standards, and because Congress has already proceeded to set safety standards. In 1966 (eighteen months before Larsen), Congress enacted the National Traffic and Motor Vehicle Safety Act, 15 U.S.C.A. §§ 1381-1431. Section 1392 instructs the Secretary of Transportation to establish motor vehicle safety standards.1 Section 1397(a) prohibits the manufacture of vehicles not [506]*506in compliance with these standards. However, § 1397(c) provides: “Compliance with any Federal motor vehicle safety standard issued under this subchapter does not exempt any person from any liability under common law.” It is obvious from this language that the federal standards were meant to supplement rather than obviate the law of negligence and products liability. It has been suggested that, while the federal regulations propounded by the Secretary do not preempt the common law, they may serve as strong evidence of negligence, if not negligence per se, in regard to vehicles before the effective date of each regulation. Nader & Page, Automobile Design and the Judicial Process, 55 Calif.L.Rev. 645, 669-70 (1967). The question of whether safety standards should be left to federal regulation presents a separate question from that of congressional intent. However, we are not aware that the argument of the necessity of federal regulation has been made in regard to design defects which cause accidents, and we cannot see any reason why design defects which cause injuries are any more in need of federal control. The danger that juries will arrive at conflicting conclusions is a hazard every manufacturer who distributes nationally runs. The complex, technical questions facing juries, aided by expert testimony, cannot be more difficult than the questions in such fields as medical malpractice. Finally, the argument that a single jury verdict may have profound consequences disrupting an essential industry has been characterized as contending that the desirability of immunity from liability is directly proportional to the magnitude of the risk created. Comment, Automobile Design Liability: Larsen v. General Motors and its Aftermath, 118 U. Pa.L.Rev. 299, 305 (1969).
Assuming the existence of liability for a design defect causing injuries, but not the accident, General Motors argues that the trial judge’s finding of defective design is supported by no evidence, insufficient evidence, or is contrary to the great weight and preponderance of the evidence. The expert witness below admitted that no mass-produced automobile had ever been manufactured in the United States with a roll bar or roll cage, and he agreed that the roof of Turner’s vehicle was no more unsafe than the roof on other vehicles of the same manufacturing era.
General Motors relies principally upon two cases for these evidence points. Dyson v. General Motors Corporation, 298 F.Supp. 1064 (E.D.Pa.1969), involved roof deformation in the roll-over accident of a hardtop. General Motors argued that to require roofs which would be perfectly safe in a roll-over would be to declare convertibles unreasonbly dangerous per se. The court adopted Larsen and said, 298 F.Supp. at 1073-1074:
[T]he manufacturer was not necessarily under an obligation to provide a hardtop model which would be as resistant to roll-over damage as a four-door sedan; but the defendant was required, in my view, to provide a hardtop automobile which was a reasonably safe version of such model, and which was not substantially less safe than other hardtop models.
In Dreisonstok, supra, a 1968 Volkswagen van was involved in a front-end collision. Such a vehicle does not have the engine in front, but plaintiff’s experts declared the van to be defectively designed by comparing it to a 1966 mid-sized Ford passenger car. The Fourth Circuit, amplifying upon Dyson, held that this was impermissible; vehicles of the same type must be compared in order to determine defective design. These two cases do not stand for the proposition that defective design must be shown by comparing plaintiff’s vehicle to similar vehicles. The expert below indicted the entire industry, a possibility which Dreisonstok does not foreclose. This is in accord with the law of Texas that, while conformance to industry custom is admissible on the question of negligence, the custom itself may be shown to be negligent. [507]*507Brown v. Lundell, 162 Tex. 84, 344 S.W.2d 863 (Tex.Sup.1961). We think the expert’s condemnation of the industry for its failure to install roll bars constitutes a sufficient showing that the custom itself was unreasonably dangerous. General Motors’ evidence points are overruled.
The judgment of the trial court is reversed, and judgment here rendered that General Motors’ plea of privilege is overruled.