OPINION
Our opinion of September 14, 1995 in case number 08-94-00333-CV is withdrawn and the following substituted in its stead.
We are presented with two appeals arising out of the same incident. Originally, Appellant, Michelle J. Feldman (“Feldman”), sued Appellees, Electronic Warfare Associates (“EWA”), The Kohler Company (“Kohler”), Stant, Inc. (“Stant”), and L & M Radiator, Inc. (“L & M”), on various product liability theories, including strict liability, negligence, and breach of warranty. The district court granted summary judgment in favor of L & M on the affirmative defense of government contractor immunity under federal law. Feldman’s claims against L & M were severed and Feldman appealed by three points of error attacking the summary judgment on legal and factual grounds. Subsequently, the trial court granted summary judgment in favor of EWA, Stant, and Kohler. Feldman also appealed these judgments, challenging the application of the government contractor defense to Feldman’s defective design and failure to warn claims against EWA; and challenging the district court’s decision that Feldman presented no factual issue as to her claims against Stant and Kohler, respectively. We join these cases for purposes of our opinion to facilitate the factual summary and the legal analysis. We affirm in part and reverse and remand in part.
SUMMARY OF THE EVIDENCE
Feldman worked as a mechanic for General Electric in El Paso, Texas. She was burned while working on an XM09S vehicle used for training United States Armed Forces and operated by General Electric. The XM09S simulates a Soviet anti-aircraft missile launcher and is utilized in “war games.” 1 Feldman’s injuries occurred when the radiator cap blew off a supplemental radiator on an XM09S she was servicing. This radiator cap was never recovered. Presumably, the explosion was caused when pressure built up in the cooling system as a result of air locks that occasionally formed in the radiator hoses. Subsequent to this accident, release valves were installed on the cooling system to prevent this type of accident from recurring. Alleging products liability and negligence claims, Feldman sued Kohler, manufacturer of the XM09S’s generator engine, EWA, who redesigned the cooling system for the vehicle under government contract, L & M, who designed and built the supplemental radiator, and Stant, manufacturer of the radiator cap.
The XM09S vehicle contained two engines; a Chrysler V-8 propelled the vehicle itself, while a Kohler four-cylinder engine powered the vehicle’s weapons systems. Both engines were originally connected to one radiator for cooling. The original engine cooling configuration led to problems with overheating and *619 EWA was retained to solve the overheating problem.
The record includes the deposition testimony of Feldman; Lawrence Davis, the EWA engineer tasked with solving the overheating problem; and Toby Brooks, a mechanical engineer with General Electric. We also have before us the affidavits of Richard Braun, executive vice-president of L & M; Alan J. Wilson, manager of sales administration for Kohler; John Blommel, director of marketing services for Stant; and Gary Smith, a professional engineer and expert witness for Feldman. The record reflects that Davis first recommended that the panels channeling airflow to the radiators be replaced, as they had apparently been removed by the operators. This proposal was rejected. Next, Davis proposed that an additional cooling fan and radiator be added in front of the Kohler engine, together with extra venting. This proposal was rejected because it changed the profile of the vehicle some one and one-half to two feet. His next proposal was that an air-cooled engine be installed in place of the existing Kohler engine, with additional venting and a blower added. This proposal was rejected because of cost and because it would create a profile change. Davis’ final proposal was that a supplemental radiator be added for the Kohler engine, and that the venting be changed; this modification was accepted. Davis approached L & M with design criteria for a radiator, which L & M designed and constructed. The modified system continued to suffer overheating problems due to air locks forming in the long radiator hoses required by the modification. 2 When the air locks were absent, the cooling system worked correctly.
Davis did not work in a vacuum. The various proposals were discussed and acted upon by a group including civilian engineers and military personnel. This group made the final decision as to what modifications would be made and no modifications could be made without approval of the government. A government team known as ADATS 3 participated in the group meetings and was responsible for the approval of all modifications. ADATS decided whether proposed changes met with the other design criteria of the vehicle, and supplied ideas as to what shoxdd or should not be done.
The design criteria required by ADATS are critical to our inquiry. First, Davis had to work with the existing vehicle. Second, the modifications could not affect the profile of the vehicle even slightly. Apparently, these requirements were meant to preserve the usefulness of the vehicle as a simulator; if it did not look like a Soviet anti-aircraft missile vehicle, it could not simulate one. Third, Davis was required to retain the existing Kohler engine, as buying new engines was too costly. Finally, the Kohler engine had to be mounted on rails for easy maintenance access. This requirement necessitated extra-long radiator hoses so that the Kohler engine could slide out of the vehicle for maintenance. 4 Davis’ suggestion that access panels be installed for engine maintenance, rather than rails, was rejected. The group’s decision to go with the supplemental radiator led to derivative criteria for the supplemental radiator. The radiator could only be placed in one position in the vehicle, mounted above the Chrysler VS engine and the existing radiator. This in turn led to restrictions on the possible dimensions of the supplemental radiator. Furthermore, the supplemental radiator could only be mounted in a horizontal position. Davis showed L & M the engine, the engine data, and the space in which the radiator would be mounted. Richard Braun’s affidavit stated that all relevant design criteria for the radiator were dictated by the pre-existing features of the vehicle, including the size of the engine, the existence or lack of a cooling fan, the configuration of the engine compartment, and the physical dimensions allowed for the radiator. Braun further stated that L & M had no discretion *620 to alter any of the specifications, which he described as “very precise.”
STANDARD OF REVIEW
The standard of review on appeal of a summary judgment is whether the successful movant at the trial level carried its burden of showing that there is no genuine issue of material fact and that judgment should be granted as a matter of law.
Lear Siegler, Inc. v. Perez,
In resolving the issue of whether the mov-ant has carried this burden, all evidence favorable to the non-movant must be taken as true and all reasonable inferences, including any doubts, must be resolved in the non-movant’s favor.
Nixon,
THE GOVERNMENT CONTRACTOR DEFENSE
L .& M and EWA were granted summary judgment as to all of Feldman’s claims based on the government contractor defense, which is predicated upon the notion that state law tort actions can significantly conflict with the federal government’s unique role as the procurer of weapons systems. Essentially, when certain conditions are met, this defense allows a government contractor to enjoy the immunity from suit that the United States enjoys. 5 This case presents the first thorough application of this defense on the merits in a Texas Court of Appeals. 6
Though various federal circuit courts had fleshed out the germinal conditions for the application of this common law defense,
see e.g., Tozer v. LTV Corp.,
The Court began its analysis by finding that the procurement of military equipment was a “uniquely federal interest.”
Id.
at 504-07,
Having concluded that in some circumstances, state laws which hold government contractors liable for design defects in military equipment present a significant conflict and must be displaced, the Court proceeded to address the scope of displacement. Adopting a three-prong test designed to locate the ultimate decision-maker, the Court found that “ [¡liability for design defects in military equipment cannot be imposed, pursuant to state law, when (1) the United States approved reasonably precise specifications; (2) the equipment conformed to those specifications; and (3) the supplier warned the United States about the dangers in the use of the equipment that were known to the supplier but not to the United States.”
Boyle,
Boyle
identified three possible situations in which the operating circumferences of federal law and state law may interact: (1) federal law and state law impose coterminous duties; (2) federal law and state law impose different, but non-mutually exclusive, duties; and (3) federal law and state law impose different, mutually exclusive duties.
Boyle,
An essential element in the application of the
Boyle
test is whether the government has given meaningful approval to the design in question. Prior to
Boyle,
a central axiom in applying the defense was that the government could not approve a proposed design with a mere “rubber stamp.”
See Tozer,
The Fifth Circuit held that the mere signing of a government contractor’s design proposal does not present the requisite approval required by
Boyle;
something more than a “rubber stamp” is needed.
Id.
at 1480;
see also Maguire,
Trevino
did little to answer what sort of review or evaluation is sufficient. Limited strictly to its facts,
Trevino
stands for the proposition that the government’s merely signing off on a proposed design is not approval for
Boyle
purposes. The government must make some sort of determination that the proposed design complies with its specifications. On the one hand, this review must be more than that involved in choosing stock equipment to meet some governmental need.
Boyle,
Boyle’s
third prong requiring that the contractor warn the government of dangers known to it but not known to the government is intimately related to the government’s approval of reasonably precise specifications. The Supreme Court stated that this requirement is necessary because, without it, “the displacement of state tort law would create some incentive ... to withhold knowledge of risks.”
Id.
at 512,
However, a question exists about the scope of the warning. Feldman argues that the contractor must warn the government of dangers about which it knew
or should have known. We
disagree. In
Bynum v. FMC Corp.,
The Court did specifically address the “known or should have known” argument in
Trevino,
The district court also held that General Dynamics had failed to prove the final element of the defense, that it had warned the government about the dangers in the use of the equipment that were known to the contractor but not to the United States. The district court committed two errors in this holding. First, it held General Dynamics to a duty to warn the government of dangers about which it ‘knew or should have known.’ After Boyle a government contractor is only responsible for warning the government of dangers about which it has actual knowledge.
Of course, because of the nature of the third prong, the danger foreseeable to the contractor must be such that it was not foreseeable to the government.
See Stout,
The government contractor defense applies equally to defective design and failure to warn cases.
Id.; Garner v. Santoro,
CLAIMS AGAINST L&M
Application of Government Contractor Immunity to Sub-contractors
Feldman asserts that L & M is not entitled to summary judgment as a matter of law because L & M is not a party to a government contract. L & M is in the position of a sub-contractor to EWA as the government contractor. Neither party cites us to case law extending or refusing to extend the defense to a sub-contractor. Based on the policy laid down by the Supreme Court in Boyle, however, we conclude as a matter of law that the defense does “trickle down” to sub-contractors. Assume the government contracts with a manufacturer for a particular device consisting of Part A, Part B, and Part C linked in series. The manufacturer only builds Part B and Part C, so it contracts with a sub-contractor for Part A. The subcontractor duly notifies the manufacturer and the government that Part A linked in series with Part B and Part C will generate an explosion. The government takes delivery of the device and it explodes, injuring many. The sub-contractor is the defendant in a class action suit containing numerous class members. As a result, the sub-contractor raises the price of Part A to the manufacturer and the manufacturer, accordingly, raises its price to the government. The result is the same as that disapproved by the Supreme Court in Boyle, just once removed. Therefore, we hold as a matter of law that a subcontractor can assert the affirmative defense of government contractor immunity.
L&M Established the Government Contractor Defense
Feldman additionally asserts that L & M failed to conclusively prove the government contractor defense.
1. Reasonably Precise Specifications Present
Feldman devotes the majority of her argument to an attack on the first prong of Boyle, asserting that the government did not provide or approve reasonably precise specifications for the XM09S modification. Feldman presents no controverting witnesses as to the testimony of Toby Brooks or Lawrence Davis, relying instead on other portions of their testimony to show that the government did not provide reasonably precise specifications. These witnesses, with Richard Braun, are the sources for all of the evidence L&M presented with its motion for summary judgment. 10
It is conclusively established in the record that all decisions with regard to the XM09S modifications were made by a group containing civilians, military personnel, and Davis. This group was dominated by the govem *626 ment presence, ADATS. ADATS decided whether proposed changes met with the design criteria, and provided design input independent of Davis. The presence and power of ADATS is the most persuasive indicator of the exercise of discretion by the government. This discretion is manifested in the limits placed upon Davis. He had to work with the existing vehicles; he could not even slightly change their appearance so that they maintained their effectiveness as threat simulators; he had to retain the existing Kohler engine; the Kohler engine had to be mounted on slide-out rails for maintenance purposes. The group decided the solution was to install a supplemental radiator. The above criteria dictated that the radiator be mounted horizontally above the Chrysler V-8 and existing radiator, that long hoses run from it to the Kohler engine, and what the size and capacity of the supplemental radiator must be. The design of the radiator was totally dictated by its placement, size, the size of the engine, the operating specifications of the engine, the configuration of the vehicle, and the cooling capacity needed. The record conclusively shows that the group, including ADATS, exercised the discretion, and left Davis and L & M none. The group, including ADATS, dictated requirements such that there was only one device Davis could put in the XM09S to solve the overheating problem, and that was a radiator of a type dictated by other XM09S requirements. Therefore, L & M has conclusively established the first prong of the government contractor defense.
2. Compliance With Government Specifications Established
Feldman does not challenge L & M’s establishment of the second prong of the government contractor defense in her brief on appeal. The record reflects that the XM09S vehicles were modified according to the plan decided on by the group, including ADATS. Feldman presented no summary judgment evidence controverting the testimony on this point. Therefore, L & M has conclusively established the second prong of the government contractor defense.
3. Warning Unnecessary Where Danger Unknown To Contractor
Richard Braun, by affidavit, stated that the only danger regarding the use of the radiator of which L & M was aware was the possibility of hot liquids being expelled from the radiator if the cap were removed when the radiator was hot. He further stated that the government was warned of this danger by a warning printed directly on the attached radiator caps which stated “DO NOT OPEN HOT.” Braun specifically testified that L & M was not aware that a radiator cap could be blown off the radiator in the XM09S, or that air locks would build up in the radiator hoses, or that a radiator cap had ever blown off an L
&
M radiator, and he asserted that a radiator of the type supplied could not build up sufficient pressure to blow off the cap. Feldman presented no summary judgment evidence rebutting L & M’s assertion it had no knowledge. The affidavit of Gary Smith alleges that L & M failed to provide adequate warnings, but Smith never asserts that L & M knew or should have known of the potential danger. Because L & M conclusively proved it had no knowledge of the alleged defect, it need not prove the third prong of the
Boyle
test.
See Boyle,
Failure to Warn Claims
Finally, Feldman argues that L & M failed to prove the government contractor defense as to her failure to warn claims. L & M argues that this assertion of error was not presented in Feldman’s response to its motion for summary judgment, and is therefore waived. Examination of Feldman’s response to L & M’s motion for summary judgment indicates she did not raise this precise argument as to this distinct claim below. Non-movants must present their issues in avoidance of summary judgment in them response to the motion for summary judgment. Tex.R.App.P. 166a(e);
McConnell v. Southside Independent School Dist.,
Failure to warn claims are reviewed under the same
Boyle
test as defective design cases.
Garner,
Disposition as to L & M
The trial court’s summary judgment as to Feldman’s failure to warn claims is reversed, and the cause remanded for further proceedings. Summary judgment as to her claim of defective design is affirmed.
CLAIMS AGAINST EWA
Application of the Boyle Test
Feldman asserts that EWA’s summary judgment evidence does not establish any part of the Boyle test. We analyze each element in turn. With regard to the first element and largely based upon Boyle and Trevino, Feldman has identified three situations in which a Boyle analysis may be conducted. According to Feldman, the government contractor defense is appropriate if the government has provided an express and precise delineation of exact design specifications, or it has been involved extensively in the design process, or it has engaged in a substantive review and evaluation of design proposals. These situations present decreasing levels of government involvement in the design of procured equipment. Feldman asserts that the government’s involvement in this case fell somewhat below these situations and was a mere “rubber stamp” of EWA’s design recommendations. She argues that EWA had ample room to exercise discretion within the constraints imposed by ADATS and the physical characteristics of the XM09S. We disagree.
Feldman’s argument depends on a very narrow view of EWA’s, and more particularly, Davis’ role in the design of the modified cooling system. In substance, she would have us believe that any unreviewed exercise of discretion on the contractor’s part eliminates the government contractor defense. However,
Boyle
expressly rejected this view. In discussing a proposed test that would apply the defense only if the contractor did not participate or participated only minimally in the design process, the Court held that the “design ultimately selected may well reflect a
*628
significant policy judgment by Government officials whether or not the contractor rather than those officials developed the design.”
Boyle,
EWA was retained to solve the overheating problem in the XM09S, and its engineer, Davis, was assigned the task of troubleshooting this problem. He presented various proposals to a group made up of government and civilian members. This group made the ultimate decision as to which proposal was implemented; not Davis, and not EWA. Furthermore, ADATS, the government player in this process, held the trump card. It expressly rejected at least three proposals based on its overriding policy considerations. No reconfiguration of the cooling system that entailed the slightest profile change in the system or that exceeded certain cost measurements was accepted. It accepted the only proposal that fit within those strict parameters. This type of back and forth problem solving and the severe nature of the constraints placed on EWA clearly satisfy Boyles’ first prong and illustrate that the government approved the design proposed by Davis based on legitimate policy considerations involving cost effectiveness, performance, and ease of maintenance. The government knew enough about the proposed design to see that it met its needs and its approval was no “rubber stamp.” Furthermore, we are of the opinion that if “approval” as defined by Boyle and Trevino calls for the type of telescoping-in on decisions made exclusively by a contractor as Feldman suggests, the constraints imposed by ADATS on EWA were so severe that no meaningful exercise of discretion could be exercised on their part. 11 Therefore, we hold that the first prong of EWA’s government contractor defense has been proved as a matter of law.
The second prong of the
Boyle
test requires conformity between the equipment and the reasonably precise specifications approved by the government.
Boyle,
The third
Boyle
prong requires the contractor to warn the government of defects known to it but not to the government.
Boyle,
Q: Do you remember the configuration of the hoses that were connected to that engine and that radiator?
A: Vaguely I remember them, and I remember they were very long. And they had to be carefully stowed, and you had to be careful when you pulled the vehicle out. My recommendation was to never try to pull the vehicle out while the engine was hot and to stow the long radiator hoses in a position which would try to prevent, or would prevent if they were stowed correctly, any air locks or this sort of thing because air lock is common if you don’t keep the vehicle or the radiator and the cylinder block and everything completely full of water at all times. If you keep them completely full of water, they generally won’t air lock. It’s just like your automobile. If you let the radiator run low, you could get an air lock in your cooling system.
[[Image here]]
Q: Did you believe at the time the L & M radiator was installed that the length of the hoses posed a potential air-lock problem in this system?
A Well, potential, yes. But if they were properly stowed, and the system was properly maintained, kept full of coolant, it could have been avoided, yeah.
Q: Was it a foreseeable potential problem that you, as a mechanical engineer, could foresee?
A Yeah, you have to foresee that there was a potential problem in the absence of proper care and maintenance.
Q: Did you propose any warnings to be placed on the system where the personnel would be able to read a warning about placement of the hoses? Did you propose anything like that?
A I don’t recall having specifically proposed it, but the head of safety there, I think was an RCA safety man, and we discussed this. Whether or not one was put on there — I did warn the operators and the safety man that they should be careful about handling this thing with those long hoses on there.
[[Image here]]
Q: While you were at EWA, were there any written warnings or instructions put on the machine regarding the use of the hoses or the way they were stored?
A I don’t recall whether I wrote anything in my notes or not, but they were discussed.
Q: Based on your experience with this particular machine, do you believe the mechanics, the people that would work on the radiator and Kohler generator, should have been informed about potential air-lock dangers with the hoses?
A I assume that’s a part of training of people that are going to work with a machine, which would have been, I guess, ADATS, army, RCA, at that time, their responsibility to train their people to operate, what had to be done to operate the machine. You have to have training to drive an automobile. Anything requires some competence on the part of the people that are using it.
Q: Yes, sir, but I want to talk specifically about these long hoses. This was not a typical design; it was atypical to the extent that you had these long hoses and two radiators on top of each other, right?
A: Yeah. I’d say it was atypical. It was atypical by virtue of constraints.
Davis also testified that he had no specific contact person at ADATS. At best, he relayed his concerns of the air lock dangers to “the safety man” who by his own recollection worked for RCA, not ADATS, the Army or the government. Furthermore, the facts that air locks frequently formed in the cooling system, that overheating continued to be a problem with the XM09S, and that a pressure release valve was added to the system after the accident all suggest that the government was not adequately warned. At the very least, they present fact issues with re *630 gard to the establishment of Boyle’s third prong. Feldman’s point of error is sustained and the ease is remanded to the district court for a determination of whether the air locks posed a danger foreseeable to EWA, whether EWA adequately warned the government of that danger, and whether EWA adequately instructed the government on the appropriate care and maintenance of the system to prevent that danger.
Failure to Warn
Feldman further asserts that EWA failed to prove the government contractor defense as to her failure to warn claims. We reiterate here that failure to warn claims are reviewed under the same
Boyle
test as defective design cases.
See Gamer v. Santoro,
Disposition as to EWA
Summary judgment in favor of EWA is reversed and remanded for further proceedings.
CLAIMS AGAINST STANT
Stant sought summary judgment on Feldman’s claims based on the assertion that Feldman presented no evidence that Stant manufactured the radiator cap in question. Central to Stant’s assertion is the fact that Feldman could not produce the actual radiator cap. However, Feldman’s claims do not depend on the actual production of the allegedly defective product; they may be proved with circumstantial evidence.
See Bell Aerospace Corp. v. Anderson,
Notwithstanding Feldman’s possible ability to establish that Stant did in fact produce the radiator cap in question, she must still present a fact issue as to whether the cap contained a defect.
See Rodriguez v. Ed Hicks Imports,
DISPOSITION AS TO STANT
Therefore, we affirm Feldman’s point of error and reverse and remand to the district court for trial on the claims against Stant.
CLAIMS AGAINST KOHLER
Kohler moved for summary judgment based on the assertion that it did not manufacture the radiator or the radiator cap, it did not place the allegedly defective product into the stream of commerce, and that as a matter of law it could not be held responsible for the acts or omissions of the manufacturer of an interchangeable part. This assertion misconstrues Feldman’s claims, which encompass not only the radiator and the radiator cap, but also the generator set manufactured by Kohler. It is impossible to tell from the record whether the defects complained about stop at the radiator and the radiator cap, or if they extend to the generator system as a whole. In fact, Feldman testified that at the time of the accident, the generator was overheating. Brooks testified that the generator would overheat and the mechanics would “shut it down, wiggle the hoses, remove the [radiator] cap, try to add more water.” When asked what happened immediately before the accident, he clarified that it was the generator, and not the radiator, that was overheating. The record also reflects that a safety valve, called a petcock, was installed on the generator some three months after the accident. 12
In support of its motion, Kohler attached the affidavit of Alan Wilson, indicating that the generator in question was manufactured as a set incorporating a radiator and radiator cap manufactured by Perfex. Wilson additionally stated that he was not aware of any communication or correspondence involving Kohler or an employee, agent, or representative of Kohler which would have discussed utilizing the generator in the XM09S system or changes or modifications in the product for purposes of the system. We note, however, that an integral unit need not be unchanged between delivery and injury in order to impose strict liability upon the manufacturer. It is sufficient that the product receive only the use and alteration which might reasonably have been foreseen or anticipated.
Dion v. Ford Motor Co.,
DISPOSITION AS TO KOHLER
Accordingly, Feldman’s complaint is sustained and the ease is reversed and remanded as it relates to the claims against Kohler.
CONCLUSION
We reverse and remand the summary judgments entered in favor of EWA, Stant, *632 and Kohler. We affirm the summary judgment granted in favor of L & M as to design defects and reverse and remand with regard to Feldman’s failure to warn claims.
Notes
. During these "war games,” military personnel are taught to recognize, track and spot, both visually and with equipment, the simulated Soviet missile launcher and to respond to its maneuvers.
. L & M did not provide or manufacture the hoses, nor did it participate in any decisions concerning the hoses.
. The only reference in the record explaining this acronym is "Air Defense something or another.”
. Feldman testified that the hoses were ten feet long.
. See Federal Tort Claims Act ("FTCA”), 28 U.S.C. § 2680(a) (1988) (providing an island of immunity for discretionary governmental decisions).
. The defense was addressed in
Trapnell v. Sysco Food Services, Inc.,
.
See Feres v. United States,
. To illustrate the second situation, the Court used an example of government procurement of an air conditioning unit.
Id.
at 509,
Boyle
provides an example of the third situation, although the entity that initially developed
*622
the particular design feature in question there is far from clear. Both the Supreme Court and the Fourth Circuit Court of Appeals sparsely described the design process.
Compare id.
at 509,
. For the purpose of this analysis, the government’s approval of a design initiated by the contractor and proposed to the government is also part of the government’s involvement in the design process.
. L&M supplemented the appellate record with the motion for summary judgment of EWA, and cites to the contract between ADATS and EWA contained in that motion in its brief on appeal. Because it was not attached as sum-maiy judgment evidence to L & M’s motion for summary judgment, or cited therein, we do not consider it on appeal. Tex.R.Civ.P. 166a(c);
see McConnell v. Southside Independent School Dist.,
. At least one Court has suggested in dicta a willingness to take this view. In
Mitchell v. Lone Star Ammunition, Inc.,
. Although generally speaking, subsequent remedial measures Eire not admissible to prove negligence, an exception is provided for its admission in products liability cases based on strict liability. TEx.R.Civ.Evid). 407(a).
. Brooks testified that the specific representation was “Well, if you put a 15-pound cap on, it just allows you to run it with a little more heat.”
