OWEN, Justice,
delivered the opinion of the Court with respect to Parts I and II,
in which PHILLIPS, Chief Justice, GONZALEZ, HECHT and ABBOTT, Justices, join, and with respect to Part III, in which PHILLIPS, Chief Justice, GONZALEZ, HECHT, ENOCH, BAKER, ABBOTT and HANKINSON, Justices, join; and issued a concurring opinion with respect to Part IV, in which HECHT and ABBOTT, Justices, join, and with respect to Part V, in which HECHT, Justice, joins.
In this crashworthiness case, the Mileses contend that the tension eliminator in the seatbelt system of a Ford Ranger pickup truck was defectively designed and failed to prevent severe injuries to Willie Searcy. Because venue was improper in the county in which this case was tried, we reverse the judgment of the court of appeals in part and remand this case to the trial court for transfer to Dallas County and a new trial. We [379]*379affirm the judgment of the court of appeals to the extent that it holds that neither a stepparent nor a sibling may recover for loss of consortium. 922 S.W.2d 572.
I
Fourteen-year-old Willie Searcy, his younger brother Jermaine Searcy, and his stepfather Kenneth Miles were passengers in a 1988 Ford Ranger pickup truck when it was involved in a collision in Dallas County. The accident occurred on IH-35 when a Mercury Cougar traveling in the opposite direction on the other side of the interstate veered across the median into oncoming traffic and struck the Ranger. Willie Searcy is a quadriplegic as a result. He breathes only with the assistance of a ventilator and requires continuous care. Jermaine Searcy and Kenneth Miles suffered less severe injuries from which they substantially recovered.
Kenneth Miles was driving the Ranger and when the collision occurred was wearing a three-point lap and shoulder belt restraint system. Jermaine Searcy was in the middle seat and wore a lap belt. Willie Searcy occupied the right passenger seat and wore a three-point restraint identical to that worn by his stepfather. Expert witnesses for the plaintiffs testified that Willie Searcy’s injuries were more severe than they otherwise would have been because Ford defectively designed a “tension eliminator” that was part of the seat belt assembly. There was evidence that Willie Searcy had leaned forward to retrieve some trash from the floor of the pickup, and the Mileses’ experts theorized that when he did so, six to eight inches of slack were introduced into the shoulder harness and remained when Willie Searcy settled back into his seat. The Mileses contended that this caused his severe injuries either because the slack prevented the belt from adequately restraining Willie Searcy or because the belt caught his head and neck and tore his head from his spinal cord.
Susan Miles, who is Willie and Jermaine Searcy’s mother, and her husband Kenneth Miles brought suit against Ford Motor Company and Doug Stanley doing business as Doug Stanley Ford, the Dallas dealership from which Kenneth Miles had purchased the Ranger. The suit was instituted in Rusk County, although the collision occurred in Dallas County, the Ranger was purchased in Dallas County, and the Mileses and the Sear-cys all resided in Dallas County. The Miles-es contend that venue was proper in Rusk County because a Ford dealership is located there, although the Mileses concede that the Rusk County dealership has no connection with the collision or to the Ranger. Ford requested the trial court to transfer venue to Dallas County, and that motion was denied.
The damages sought from Ford and Stanley included loss of consortium by Willie Searcy’s brother and stepfather because of Willie’s severe injuries. Ford and Stanley moved for partial summary judgment on these claims, contending that the common law of Texas does not recognize a right of recovery by siblings and stepparents. The trial court granted that motion prior to trial. The remaining claims were submitted to a jury which found against Ford on all theories of liability and awarded $80 million in actual and $10 million in punitive damages. However, the jury refused to find against Doug Stanley Ford on any theory. The trial court entered judgment against Ford alone in accordance with those findings. Both Ford and the Mileses appealed. The court of appeals reversed the punitive damages award, holding that there was insufficient evidence to support the jury’s findings of gross negligence and malice. The court of appeals then remanded the gross negligence and malice issues for retrial, but refused to order a retrial of the negligence issue. The trial court’s judgment was affirmed in all other respects.
Ford challenges the judgment of the court of appeals on numerous grounds, including the remand of only gross negligence for another trial. Because the entire ease must be retried, we do not reach that issue, which is considered by Justioe Gonzalez in his concurring opinion. The judgment of the court of appeals must be reversed and the matter remanded for a new trial because venue did not lie in Rusk County.
With regard to the conditional cross-application filed by the Mileses, we affirm the judgment of the court of appeals that the [380]*380stepfather and brother of Willie Searcy cannot recover for loss of consortium. Our disposition of the case moots a second issue the Mileses raised, which is whether the court of appeals applied the correct legal standard in determining the factual sufficiency of the evidence to support the jury’s findings of gross negligence and malice. With regard to the Mileses’ third point of error, this Court does not have jurisdiction to determine whether the jury’s failure to find Doug Stanley liable was so against the great weight and preponderance of the evidence that a new trial was required.
II
The first question that we must resolve is whether the trial court erred in failing to grant Ford’s motion to transfer venue to Dallas County. The collision occurred in Dallas County, and the dealership from which Kenneth Miles purchased the Ford pickup is located in Dallas County. Ford, a foreign corporation, has its regional headquarters in Dallas, and its principal place of business in Texas is in Dallas County.
The sole basis for venue in Rusk County is that Premier Ford Mercury, Inc., a Ford dealership unrelated to the purchase of the truck by Kenneth Miles or to the collision, is located in Rusk County. The Mileses assert that venue is sustainable pursuant to former section 15.037 of the Texas Civil Practice and Remedies Code, which was in effect at the time this suit was filed. Act of June 3, 1987, 70th Leg., 1st C.S., ch. 4, § 1,1987 Tex. Gen. Laws 52, 53 (formerly codified at Tex. Crv. Prac. & Rem.Code § 15.037), repealed by Act of May 8, 1995, 74th Leg., R.S., ch. 138, § 10, 1995 Tex. Gen. Laws 978, 981.1 Section 15.037 provided that a plaintiff could sue an out-of-state corporation doing business in this state “in any county in which the company may have an agency or representative.” Tex. Civ. Prac. & Rem.Code § 15.037, Act of June 3, 1987, 70th Leg., 1st C.S., eh. 4, § 1, 1987 Tex. Gen. Laws 52, 53 (repealed 1995).
If there is any probative evidence in the entire record that Ford maintained an agency or had a representative in Rusk County, even if the preponderance of the evidence is to the contrary, we must defer to the trial court’s determination that venue was proper in the county of suit. Ruiz v. Conoco, Inc., 868 S.W.2d 752, 758 (Tex.1993). If there is no such evidence, venue did not lie in Rusk County, and the case must be transferred to Dallas County where it is undisputed that venue is proper. Id.
This Court has long held that the terms “agency” and “representative” as used in the former venue statutes are not to be interpreted in light of the law of respondeat superior. Milligan v. Southern Express, Inc., 151 Tex. 315, 250 S.W.2d 194, 197 (1952). Rather, as we held in Ruiz, “venue against a corporation may be predicated upon the presence in a county of either an agency — a more or less regular and permanent business operation — or a representative with broad powers to act for the corporation.” 868 S.W.2d at 759. Possession of broad power and discretion to act for the corporation is essential for both an agency and a representative under the former venue statutes. See id.
The Mileses argue that the Premier dealership in Rusk County was an agency of Ford Motor Company because Ford owned 77% of Premier’s stock and by virtue of that stock ownership, Ford ultimately received 78.08% of any profits generated by the dealership. The ownership of some or even all of the stock of a corporation does not establish that the corporation is an “agency” or “representative” of its stockholders within the meaning of former section 15.037. We held in Atchison, Topeka and Santa Fe Ry. Co. v. Stevens, 109 Tex. 262, 206 S.W. 921, 922 (1918), that a domestic subsidiary of a foreign corporation was not an agent of the parent company for venue purposes under a statute similar to section 15.037, even though the parent owned the subsidiary. See also Milligan, 250 S.W.2d at 196-97 (citing with ap[381]*381proval Atchison, Topeka and Santa Fe Ry. Co. v. Stevens, 109 Tex. 262, 206 S.W. 921 (Tex.1918)). The Premier dealership was a separate corporate entity irrespective of Ford’s ownership of stock.
The question for venue purposes is whether Premier possessed broad power to act for Ford. Ruiz, 868 S.W.2d at 759. Even if Ford itself maintained employees in Rusk County and owned property and conducted business in that county from which it derived income, those facts, standing alone, do not establish venue. For example, in Colorado Interstate Gas Co. v. MAPCO, Inc., 570 S.W.2d 164 (Tex.Civ.App. — Amarillo 1978, no writ), CIG’s largest operation in Texas was located in Moore County. CIG had an investment of $31 million dollars in a natural gas facility in that county and employed 140 nonsupervisory employees and 14 supervisors. In determining whether venue was proper under the predecessor to former section 15.037 (former Tex.Rev.Civ. Stat. art. 1995(27)), the court of appeals examined the actual responsibilities of CIG’s employees in Moore County. Id. at 170. Although those employees were in charge of the operation and maintenance of all pipelines, compressor stations, measurement facilities, and gas processing, and although they had frequent contact with CIG’s customers, suppliers, and the public, the court of appeals concluded that these “impressive” job descriptions did not amount to evidence that any supervisor or other employee had the broad discretionary power in conducting corporate affairs required in Milligan to establish venue. Id at 170-71.
Another instructive decision isRouse v. Shell Oil Co., 577 S.W.2d 787, 790 (Tex.Civ.App. — Corpus Christi 1979, writ dism’d), in which Shell “conduct[ed] a substantial amount of business in a more or less regular and permanent form in Lavaca County through its employees.” Shell’s production foreman supervised the maintenance, production, and reconditioning of wells in that and surrounding counties. Id. at 788. Nevertheless, his duties and obligations did not “relate to commercial or business transactions having something to do with the corporate affairs of the principal,” and the court of appeals affirmed the trial court’s determination that venue was not proper in Lavaca County. Id. at 789; see also Atchison, Topeka and Santa Fe Ry. Co. v. Sanchez, 890 S.W.2d 793, 794-95 (Tex.App. — Eastland 1994, no writ) (holding that although Santa Fe railroad conducted more or less regular and permanent business operations in Nolan County, none of its employees possessed broad powers to act for it).
In contrast, in Milligan, two companies were engaged in the business of shipping freight. 250 S.W.2d at 194. One of the companies, Northeast, acted as the local agent of the other, Southern Express, in Grayson County. Northeast entered into contracts for hauling freight in the name of Southern Express, hauled freight under the name of Southern Express, charged for services rendered by Southern Express, and kept the books of Southern Express for these operations. Id. at 195. In sum, Northeast possessed broad powers to carry out virtually every aspect of the freight business for Southern Express, although Northeast’s power to do so may have been limited to Grayson County. The same cannot be said of Ford Motor Company and the Premier dealership. Premier did not possess broad powers to carry out the business of Ford Motor Company.
The court of appeals concluded that because warranty work, pre-delivery preparations, and recall corrections were performed for Ford by the Premier dealership in Rusk County, and because the sales contracts executed by Premier created warranty obligations for Ford, there was some evidence that Premier was Ford’s agent or representative. At least two decisions of courts of appeals that predate Ruiz similarly concluded that pre-delivery inspections or the power to bind a manufacturer to a warranty would support venue under former article 1995(27). See General Motors Corp. v. Ramsey, 633 S.W.2d 646, 648 (Tex.App. — Waco 1982, writ dism’d); Allis-Chalmers Mfg. Co. v. Coplin, 445 S.W.2d 627, 628-29 (Tex.Civ.App. — Tex-arkana 1969, no writ); see also Ford Motor Co. v. Lemieux Lumber Co., 418 S.W.2d 909, 910 (Tex.Civ.App. — Beaumont 1967, no writ) (holding that venue was proper as to Ford [382]*382under article 1995(27) in a county in which a Ford dealership was located with no discussion of that statute). Two other decisions that predated Ruiz held that venue as to Ford was not sustainable under former article 1995(27) in counties that had Ford dealerships, but the opinions intimated that if there had been evidence similar to the evidence in this case, the outcome would have been different. See Ford Motor Co. v. Revert, 428 S.W.2d 139, 141 (Tex.Civ.App. — Amarillo 1968, no writ) (no evidence that dealer was authorized to make warranty repairs); Pierce v. Ford Motor Co., 401 S.W.2d 355, 357 (Tex.Civ.App. — Eastland 1966, writ dism’d w.o.j.) (no evidence of a warranty from Ford).
While warranty and recall work may entail some discretion on the part of an automobile dealer, it is not the type of “broad power of attorney” contemplated in Milligan, 250 S.W.2d at 197, nor is it indicative of “broad powers to act for the corporation” as discussed in Ruiz, 868 S.W.2d at 759. If Ford’s warranty work were carried out by Ford employees at a facility that it owned, we would not say that the employees who performed that work were agents or representatives for venue purposes. Their role would be more akin to that of a “servant.” See Milligan, 250 S.W.2d at 197. Nor is the fact that Ford’s warranty obligations arose through sales by Premier an indication that Premier had broad powers to act for Ford. Express warranties from manufacturers adhere in many sales transactions, and often, the retail seller replaces or repairs a defective or damaged product and is reimbursed or paid by the manufacturer. This does not mean that the drug store, hardware store, or grocery store where the product was purchased becomes the agent or representative of the manufacturer for venue purposes. To the extent that the decisions in General Motors Corp. v. Ramsey, Allis-Chalmers Mfg. Co. v. Coplin, Ford Motor Co. v. Lemieux Lumber Co., Ford Motor Co. v. Revert, and Pierce v. Ford Motor Co. hold or imply that an automobile dealership is the agent or representative of the manufacturer if the dealership performs warranty or recall work, they are disapproved.
The Mileses point to testimony that Ford had once told a sales manager at yet another Ford dealership, the Henderson Ford-Mercury, Inc. dealership in Rusk County, that “[y]ou are Ford Motor Company in this area [Rusk County].” Even assuming that a statement made to an employee of an unrelated entity was relevant in determining the power of Premier to act for Ford, it cannot reasonably be inferred from this testimony that Premier possessed the broad discretionary authority necessary for venue under former section 15.037. In the CIG v. MAPCO case discussed above, CIG’s employees were “CIG” in Moore County, but that fact says nothing substantive about the power to act for the foreign corporation. An employee who wears a company’s uniform or lapel pin plainly represents that company to the public, but this Court’s decision in Milli-gan made clear that “the presence in the county of a mere servant” is not the equivalent of the presence of an agent or representative for venue purposes. Milligan, 250 S.W.2d at 197. The test for determining if an employee is an agent or representative is the extent of that employee’s power to act for the corporation.
There is no evidence in the record to support the trial court’s conclusion that Premier was the agent or representative of Ford within the meaning of former Tex. Crv. Prac. & Rem.Code § 15.037, and venue did not lie in Rusk County. The Legislature has declared that improper venue cannot be harmless error. Tex. Crv. Prac. & Rem.Code § 15.064(b). Accordingly, we remand this case to the trial court with instructions to transfer the case to Dallas County.
We turn to the loss of consortium issue.
Ill
The Mileses filed a conditional application for writ of error asking that we recognize a cause of action by siblings and stepparents for loss of consortium. The trial court and the court of appeals correctly declined to permit such a recovery.
This Court held in Reagan v. Vaughn, 804 S.W.2d 463, 465-66 (Tex.1990), that a child has the right to recover for loss of consor[383]*383tium when a parent suffers a serious, permanent, and disabling injury. We admonished, however, that we were limiting this extension of the common law to the parent-child relationship and specifically rejected the suggestion that this would lead to recovery by siblings:
Respondents have suggested that recognition of this cause of action will somehow have the snowball effect of leading to recognition of actions in favor of siblings, grandparents, close friends, and so on. We have little difficulty limiting recovery to the parent-child relationship.
Id.
We adhere to the boundaries we established for loss of consortium in Reagan. We first note that historically, the common law did not recognize a cause of action for the death of a loved one. It was only with the passage of wrongful death statutes in this and other states that a spouse or child of the deceased could recover for the loss they suffered. See generally W. Page Keeton et al., PROSSER AND KEETON ON THE LAW OF TORTS § 127, at 945-47 (5th Ed.1984). Our current wrongful death statute limits recovery to “the exclusive benefit of the surviving spouse, children, and parents of the deceased.” Tex. Civ. Prac. & Rem.Code § 71.004. Our decision in Reagan was not inconsistent with the wrongful death statute when we held that under the common law, children suffer a compensable loss not only when a parent dies but also when that parent suffers a debilitating, serious, and permanent injury. However, it would be anomalous to recognize a cause of action for loss of consortium for a severe injury to a loved one when there is no recovery for the death of that same family member.
The Legislature has determined that only spouses, parents, and children may sue for wrongful death. While we are not bound by these policy decisions when we consider who may recover as a result of injuries to another person under the common law, the boundaries the Legislature has drawn do inform our decision.
We are further mindful of the course other jurisdictions have charted in this area. We noted in Reagan when we recognized a child’s right to recover for the loss of consortium of a parent that we were among a minority of jurisdictions. Reagan, 804 S.W.2d at 465. We nevertheless were persuaded by the rationale found in various decisions of other courts and in articles by commentators urging recognition of a child’s right to sue for loss of consortium. However, we have found virtually no support for further extending recovery for loss of consortium to siblings. The Mileses have cited no decision of any court embracing such a proposition, and we have found only one, Levens v. Commercial Union Insurance Co., 485 So.2d 521, 522 (La.Ct.App.1986). Nor have the Mileses offered any arguments that would suggest a reasoned basis for extending the common law in this area.
Similarly, the Mileses have offered no authority for permitting a stepparent to recover for loss of consortium, and we have found no such authority. As already noted, the majority of jurisdictions do not even permit a natural or adoptive parent to recover loss of consortium for injury to a child.2 Two [384]*384former justices of this Court did indicate that they would permit a stepparent to recover mental anguish damages as a bystander if he or she contemporaneously perceived the death or severe injury of a stepchild. See Freeman v. City of Pasadena, 744 S.W.2d 923, 924 (Tex.1988) (Ray, J., concurring). The majority of the Court, however, did not discuss that issue in Freeman and held only that the plaintiff in that case was not a bystander because he learned of the accident after it occurred. See id. at 924. Even were we to agree with the concurring opinion in Freeman, an issue not before us, a bystander cause of action is not the equivalent of a recovery of damages for loss of consortium. Bystander recovery contemplates that the plaintiff is emotionally injured from the sensory and contemporaneous observance of an accident to which he or she was in close proximity. Freeman, 744 S.W.2d at 923-24. The plaintiff is compensated for the mental anguish suffered as a result of this direct injury. Mental anguish is not an element of recovery for loss of consortium, but it generally is the only element of damage in a bystander ease. Reagan, 804 S.W.2d at 466-67.
Although we recognize that there will be cases in which a stepparent or other relative has assumed the role of a parent in a child’s life, we correctly concluded in Reagan that “the two relationships likely to be most severely affected by a negligent injury to a person are the husband and wife relationship and that of parent and child.” 804 S.W.2d at 466. The distinctions we drew in Reagan are valid ones:
The distinction between the interests of children and those of other relatives is rational and easily applied.... Thus, by limiting the plaintiffs in the consortium action to the victim’s children, the courts would ensure that the losses compensated would be both real and severe.
Id. (quoting Theama v. City of Kenosha, 117 Wis.2d 508, 344 N.W.2d 513, 521 (1984)). We correctly observed that “‘[wjhile all family members enjoy a mutual interest in consortium, the parent-child relationship is undeniably unique and the wellspring from which other family relationships derive.’ ” Id. (quoting Villareal v. State, 160 Ariz. 474, 774 P.2d 213, 217 (1989)). It is also important to note that typically, stepparents have no legal obligation to provide a home or support for stepchildren. We decline to expand our common law beyond well-settled and well-reasoned boundaries.
IV
The Court’s decision regarding venue is dispositive of the judgment that will be entered in this case. However, other substantial and significant issues have been raised, and some of those questions undoubtedly will persist on remand. To permit those issues to linger, perhaps necessitating a third trial, would be patently unjust to the parties and would result in a waste of judicial resources. Only recently, all nine members of this Court joined in considering an issue “not essential to [the] disposition of this case” in order to “provide the trial court with guidance in the retrial” of the case. Edinburg Hosp. Auth. v. Trevino, 941 S.W.2d 76, 81 (Tex.1997); see also Tyler Bank & Trust Co. v. Saunders, 159 Tex. 158, 317 S.W.2d 37, 43 (1958) (holding that since there must be a retrial, the Court felt it necessary to address the admissibility of certain evidence). Accordingly, a resolution of issues that are likely to recur is in order, and I would proceed to decide them. The first concerns the charge to the jury on defective design.
The trial court instructed the jury that a manufacturer has a duty to “reasonably design, manufacture and market the vehicle so as not to subject occupants of the vehicle to [385]*385unnecessary and avoidable injuries in the event of a collision.” I agree with the court of appeals that it was error to include this instruction in the charge because it misstated the law. 922 S.W.2d at 598. I part company with the court of appeals, however, when it concludes that the error was not reversible. Id
The charge included definitions of proximate cause and producing cause followed by this paragraph:
The manufacturer of a motor vehicle has a duty to reasonably design, manufacture and market the vehicle so as not to subject occupants of the vehicle to unnecessary and avoidable injuries in the event of a collision. A motor vehicle manufacturer is not an insurer of the safety of its product and is only required to design, manufacture and market a product that is reasonably safe for its intended use. A defective product, negligence or breach of warranty may be a proximate or producing cause of an injury, whether or not such defective product, negligence or breach of warranty actually caused the particular collision in question.
Six liability issues that each included either producing or proximate cause were then submitted, and those issues encompassed theories of design defect, manufacturing defect, marketing defect, negligence, fitness for ordinary purposes (breach of warranty), and implied warranty of fitness for a particular purpose. The instruction quoted above applied globally to each of these issues. The jury found against Ford and for the Mileses in response to all questions.
At the outset, the Mileses’ contention that Ford waived its objections to the part of the instruction that imposed a “duty ... not to subject occupants ... to unnecessary and avoidable injuries” should be briefly addressed.
A
The Mileses contend that Ford’s objections did not adequately apprise the trial court of why the instruction was error, citing Castle-berry v. Branscum, 721 S.W.2d 270, 276 (Tex.1986). The Court held in Castleberry that an objection that failed to explain “why the instruction is legally incorrect or how it would confuse the jury or prejudice the defendants,” is insufficient to preserve error. 721 S.W.2d at 277. Ford has articulated three complaints about the instruction in its application for writ of error: that it expanded upon the products liability charge approved in Acord v. General Motors Corp., 669 S.W.2d 111 (Tex.1984), that it broadened the duty of a manufacturer beyond that recognized by Texas law, and that it “nudged” the jury in favor of the Mileses in a closely contested case.
In the trial court, Ford leveled the following objections to the global instruction describing Ford’s duty:
Your Honor, our objection to this sentence focuses on the use of the words unnecessary and avoidable. And the objections are these. First of all, that by using that phrase, this sentence, this instruction is a materially incorrect statement of Texas law.
There is, Your Honor, no duty in Texas of a manufacturer to undertake, without qualifications, all steps, no matter how costly or cumbersome that might avoid an injury in the event of a collision, yet this instruction, by inclusion of the terms unnecessary and avoidable injuries, is telling the jury precisely that, that Ford had a duty in this case to do everything that was physically possible, no matter how many billions of dollars it might cost. Even if it was technology that was only available for use on the space shuttle now, that still would be something that this jury was instructed that Ford had a duty to incorporate in this Ranger pickup, because it was a physically possible thing to do, if it was something that could avoid an accident.
After further objections to the same instruction, which are not material here, Ford continued: “Not only is it surplusage, but it is designed and has the effect of nudging the jury towards a Plaintiffs’ verdict in the case.”
In Acord, we approved “the Pattern Jury Charges issue and instruction for design defect cases,” and disapproved “the addition of any other instructions in such cases, however correctly they may state the law.” 669 [386]*386S.W.2d at 116. See also Fleishman v. Guadiana, 651 S.W.2d 730, 731 (Tex.1983); Turner v. General Motors Corp., 584 S.W.2d 844, 847, 855 n. 1 (Tex.1979). Although Ford did not mention Acord by name, Ford’s objection identified at least two of the substantive defects of the submission. Ford advised the trial court that the instruction required Ford to prevent “unnecessary and avoidable injuries in the event of a collision” -without regard to cost or to the state of technology and that the instruction nudged the jury in favor of a plaintiffs’ verdict. These objections preserved Ford’s complaints.
B
The basic error in the instruction given by the trial court is that it is at odds with the risk versus utility analysis that lies at the core of products liability design defect law in Texas. See Boatland of Houston, Inc. v. Bailey, 609 S.W.2d 743 (Tex.1980). The Court observed in Boatland that “when the plaintiff alleges that a product was defectively designed because it lacked a specific feature, attention may become focused on the feasibility of that feature [and] the capacity to provide the feature without greatly increasing the product’s cost or impairing usefulness.” Id. at 746. The instruction given by the trial court does not embody the risk/utility touchstone. It is, as Ford pointed out to the trial court, an incorrect statement of the law of Texas.
The instruction tells the jury that Ford had a duty to “reasonably design, manufacture and market” the Ranger pickup bought by Kenneth Miles “so as not to subject occupants of the vehicle to unnecessary and avoidable injuries in the event of a collision.” The instruction implies that a manufacturer has a duty to prevent any “unnecessary and avoidable injuries in the event of a collision.” This Court has long recognized that ‘“[a] manufacturer or distributor of products is not an insurer,’ ” Acord, 669 S.W.2d at 114 (quoting Shamrock Fuel & Oil Sales Co. v. Tunks, 416 S.W.2d 779, 785 (Tex.1967)), and that “a manufacturer is not required to design the safest possible product,” id. (citing Henderson v. Ford Motor Co., 519 S.W.2d 87 (Tex.1974)).
The Mileses argue that the inclusion of the word “reasonable” after “duty” saves the instruction. That argument should be rejected. The jury was not told what the duty to “reasonably” design meant in this context. The jury could have concluded that cost should not be a factor. Or, the jury could have concluded that the tension eliminator was not reasonably designed to prevent Willie Searcy’s injuries even though the jury may also have concluded that the tension eliminator prevented thousands of other injuries. An example included in an amicus brief in this case illustrates the problem. Under the trial court’s instruction, a jury could find that it was not reasonable to design a vehicle capable of exceeding the speed of 55 m.p.h., which was formerly the legal limit in most states. The instruction does not tell the jury to balance the increased risk of injury in collisions that occur while traveling in excess of 55 m.p.h. against the utility of designing vehicles that can exceed 55 m.p.h. In other words, neither the risk nor the utility side of the equation was adequately captured in the trial court’s instruction which simply stated that the duty was to “reasonably” design.
This Court specifically approved the following design defect issue and instruction for use in crashworthiness cases in an effort to instruct the jury regarding its duty to weigh risk against utility:
Do you find from a preponderance of the evidence that at the time the [product] in question was manufactured by [the manufacturer] the [product] was defectively designed?
By the term “defectively designed” as used in this issue is meant a product that is unreasonably dangerous as designed, taking into consideration the utility of the product and the risk involved in its use.
Turner, 584 S.W.2d at 847 n. 1.
We approved a similar, although not identical, issue and instruction in Acord, 669 S.W.2d at 113. The question and instruction submitted by the trial court in this case regarding a design defect included an instruction that roughly tracked Acord and [387]*387Turner.3 The Mileses contend that the inclusion of risk/utility in that instruction rendered the global instruction regarding “unnecessary and avoidable injuries” harmless. That is not the case. The two instructions conflict.
The Mileses alternatively contend that the second sentence of the global instruction to which Ford objects was added at Ford’s insistence and cured any harm. The second sentence says: “A motor vehicle manufacturer is not an insurer of the safety of its product and is only required to design, manufacture and market a product that is reasonably safe for its intended use.” This language does not direct the jury to engage in a risk/utility analysis, which it is required to do in order to find a design defect. The sentence says only that the product must be “reasonably” safe. For the reasons already discussed, the inclusion of only the concept of “reasonableness” is insufficient.
It should be noted that an instruction similar to the second sentence of the instruction submitted in this case was specifically disapproved in Acord, 669 S.W.2d at 113-16, although it is a correct statement of the law, because the instruction was a comment on the weight of the evidence. This Court should not approve of the submission of improper language in jury instructions. That does not mean however, that Ford waived its objections to the instruction as a whole by asking the trial court to include an erroneous sentence under the facts of this case. Ford requested the second sentence conditioned upon the inclusion by the Mileses of the first sentence in the paragraph, to which Ford had objected, and Ford requested the second sentence only after the trial court declined to delete the first sentence. Ford advised the trial court that the entire instruction, including the second sentence, was erroneous and should not be given at all.
The erroneous instruction was harmful error. It infected the entire charge. It was given in connection with the definitions of proximate and producing cause. Either proximate or producing cause was an element in each of the six liability issues. The instruction expressly defined Ford’s duty with regard to the design, manufacture, and marketing of the Ford Ranger. The instruction ended by admonishing that a “defective product, negligence or breach of warranty may be a proximate or producing cause of an injury, whether or not such defective product, negligence or breach of warranty actually caused the particular collision.”
The instruction tainted all the findings of the jury in what was a closely contested case. The verdict was not unanimous, and, as the court of appeals .noted, “[t]he evidence is barely sufficient to find ordinary negligence ... [because] Ford relied on studies by [the National Highway Traffic Safety Administration] that consistently showed the risk-utility balance of tension eliminators weighed in favor of overall safety, and that the kind of tension eliminators Ford used were not unreasonably dangerous.” 922 S.W.2d at 589.
Two of the decisions on which the Mileses rely in arguing the error in the charge was not harmful recognize that in a close case an erroneous instruction is more likely to result in the rendition of an improper verdict. In Reinhart v. Young, 906 S.W.2d 471, 472 (Tex.1995), the trial court had submitted an instruction on unavoidable accident, and the jury found in favor of the defendant. We did not decide whether it was error to give such an instruction, but concluded that any error was not reversible because “[t]his was not a close case, where a superfluous instruction would be more likely to influence the jury improperly.” Id. at 473 (emphasis added). In Reinhart, there was “ample” evidence that the defendant was not negligent and further, we held that the error in giving an [388]*388instruction on unavoidable accident was harmless when an instruction concerning the doctrine of sudden emergency was given without objection. Id. at 473-74. Our decision in Lone Star Gas Co. v. Lemond, 897 S.W.2d 755, 756 (Tex.1995), likewise concluded that the error, if any, in submitting a surplus instruction that correctly stated the law on marketing defects was not harmful because “[u]nlike Acord, this case was not closely contended.” If a correct statement of the law is reversible error in a closely contested case, an incorrect statement of the law must certainly be reversible error.
One final observation about the charge in this case concerns the discussion in the briefing of alternative designs. We have held that “if there are no safer alternatives, a product is not unreasonably dangerous as a matter of law.” Caterpillar, Inc. v. Shears, 911 S.W.2d 379, 384 (Tex.1995). Similarly, in American Tobacco Co. v. Grinnell, 951 S.W.2d 420, 433 (Tex.1997), this Court held that “if there is no safer alternative to the cigarette manufactured by American, then its cigarettes are not unreasonably dangerous as a matter of law.” The soon-to-be issued Restatement (Third) of Torts: Products Liability defines the risk-utility test as “whether a reasonable alternative design would, at reasonable cost, have reduced the foreseeable risks of harm posed by the product and, if so, whether the omission of the alternative design rendered ... the product not reasonably safe.” The Restatement (Third) of Torts: Products Liability § 2 cmt.d (unedited page proof, February 23, 1998). Some have questioned whether the instructions previously approved in Acord and Turner remain adequate in light of the continuing developments in the law of products liability. Our Court has in the past re-examined what should and should not be included in a jury charge as the law of strict liability has evolved and matured. See Turner, 584 S.W.2d at 847 (overruling Henderson v. Ford Motor Co., 519 S.W.2d 87 (Tex.1974), and overruling General Motors Corp. v. Hopkins, 548 S.W.2d 344 (Tex.1977), upon re-examination of the definition of “unreasonably dangerous”). However, that issue is not before the Court. None of the parties requested an instruction regarding safer alternative designs.
V
A second issue likely to recur is the admissibility of certain evidence. During trial, the Mileses played two videotapes of sled tests that Ford had conducted some years ago. Ford contends this evidence was highly prejudicial and inflamed the jury.
Experts for the Mileses had at least two theories of how the slack in the seat belt exacerbated Willie Searcy’s injuries. One was that during the impact of the collision, Searcy slid forward until his body completely loaded the lap belt and then his upper body continued forward until it reached the end of the slack in the shoulder belt and slammed into that belt. Experts called by the Mileses theorized that Willie’s head then continued forward with an enormous amount of acceleration while the rest of his body decelerated, and that these forces tore his skull from his spinal column. Another expert for the Miles-es was of the opinion that the lower part of Willie Searcy’s body “submarined” under his lap belt during the collision and that slack in the shoulder belt caught his neck and almost severed his head from his spinal column.
In 1966, Ford conducted a sled test on a Thunderbird vehicle. The videotape of that test graphically depicts the heads being torn off the two dummies belted into the sled. The tape shows the collision six different times from several different angles in slow motion. It was played to the jury in its entirety then replayed, stopped, and restarted at various points while the Mileses’ expert Stephen Syson testified. The seat belt assembly in the Thunderbird was substantially different from the one in the 1988 Ranger and did not include a tension eliminator. The Mileses concede that the test is not similar to the collision in which Willie Searcy was injured.
The second videotape was of a sled test conducted in 1980 on a Ford Ranger. The dummies in that test were roughly the size of six-year-old children and were realistically dressed as children. The tape showed six slow motion views of the crash in which the child-like dummies were decapitated. The [389]*389Mileses’ expert testified that this tape showed that the “seat belt ripped the dummy's head off.” The tape was then played to the jury with starts and stops while the expert continued to testify. The expert then confirmed in response to questions by the Mileses’ counsel that what happened to Willie Searcy was “exactly what’s happening to these dummies in this sequence.”
It is undisputed that the belt system in the pickup used in this 1980 sled test had no tension eliminator and was different from the belt system in this case. Further, the Miles-es’ expert witness acknowledged that no slack had been introduced into the belt system for the 1980 test. This expert testified, however, that because the belt system in this vehicle had longer belt lengths, it provided a “simulation of the effect of slack here.”
The Mileses have never contended that the sled tests depicted in these videotapes were conducted under conditions similar to the collision in this case. Accordingly, the tapes should not have been admitted as evidence of how Willie Searcy was injured. The only basis urged by the Mileses for the admission of the tapes was for the limited purpose of impeaching Ford’s witnesses. The Mileses contend that Ford made statements in other cases and to the federal government to the effect that it had not conducted tests with slack in the seatbelts. The videotapes, the Mileses argue, pointedly demonstrate that Ford did conduct tests with slack in the seatbelts. The trial court originally ruled that the sled tests would be admitted for impeachment purposes, but gave no limiting instruction to the jury at the time the tapes were introduced into evidence.
As is evident from the testimony quoted above, the tapes were not used for impeachment. The record also reflects that after the 1966 sled test video had been played to the jury, an expert witness for the Mileses agreed that this was “information that Ford had about what seat belts would do when they weren’t snug against the chest and human neck,” and he then testified: “[tjhey at least show the effect of the seat belt and how the seat belt interacts with the neck of the test dummy, particulary children.” Ford objected to the admission of this and the other tape precisely because Ford feared that these tapes would be used to show what the Mileses contend happened to Willie Searcy and not for legitimate impeachment purposes.
The videos should not have been admitted even for impeachment purposes. While the trial court has considerable discretion in the admission of evidence, it must balance the probative impact of evidence against the possibility of unfair prejudice, confusion of the issues, or misleading the jury. Tex.R. Civ. Evid. 403. The court of appeals correctly observed that relevant photographic evidence is admissible unless it is “ ‘merely calculated to arouse the sympathy, prejudice or passion or the jury’ where the photographs do not serve to illustrate disputed issues or aid the jury in understanding the case.” 922 S.W.2d at 597 (quoting Heddin v. Delhi Gas Pipeline Co., 522 S.W.2d 886, 889-90 (Tex.1975) (holding that admission of photographs of dead pets and livestock to illustrate the damage caused by a pipeline rupture was error that required reversal)). The videos were calculated to arouse the sympathy and passions of the jury. Furthermore, there was other evidence available from testimony and documents about tests that Ford had run with slack in seatbelts. The videos were not the sole source of impeachment evidence, and I would hold that it was harmful error to admit them.
# ⅜ ⅜ ⅜ ⅜
Because the motion to transfer should have been granted, we reverse the judgment of the court of appeals in part and remand this case to the trial court for transfer to Dallas County and a new trial.