Full opinion text
DECISION AND ORDER ADELMAN, District Judge. On December 10, 1993, six Air National Guard maintenance crew members were killed when a KC-135R tanker aircraft exploded as they inspected its fuel system at General Billy Mitchell International Airport in Milwaukee. This action is brought by the surviving spouses of five of the Air National Guard members. Plaintiffs allege that a fuel pump installed in the airplane six days before the explosion was defectively designed and manufactured, and have sued Crane Company (“Crane”), which made the pump, and the Boeing Company (“Boeing”), which issued a set of specifications for the fuel pump and approved an earlier version. Before me are defendants’ motions for summary judgment and Crane’s motion for partial summary judgment on punitive damages. I. JURISDICTION A district court has subject matter jurisdiction over cases between citizens of different states. 28 U.S.C. § 1332(a)(1). Diversity of citizenship is present only if no defendant shares state citizenship with any plaintiff. Strawbridge v. Curtiss, 7 U.S. (3 Cranch) 267, 2 L.Ed. 435 (1806). In considering subject matter jurisdiction sua sponte, as I must, Rice v. Rice Foundation, 610 F.2d 471, 474 (7th Cir.1979), I discovered several issues which needed to be addressed, and held a telephone conference with the parties to do so. First, all plaintiffs are Wisconsin citizens in their personal capacities. (R. 60 ¶¶ 5-13.) However, their citizenship for the claims they assert as executors or administrators of decedents’ estates is determined not by their citizenship, but by that of decedents, 28 U.S.C. § 1332(c)(2), and their pleadings are silent regarding decedents’ citizenship. The parties stipulated that decedents were Wisconsin citizens. (R. 169 ¶ 1.) Second, two defendants which were Crane predecessors or are Crane subdivisions denied in their answers that they were foreign corporations and that their principal places of business were in other states. (R. 70 ¶ 18 (BFM Aerospace Corp.); R. 71 ¶ 17 (Lear Siegler, Inc.).) All parties stipulated that all Crane predecessors and subdivisions with independent existence are foreign to Wisconsin and have their principal places of business elsewhere; Crane’s counsel said that he could stipulate on their behalf. (R. 169 ¶¶ 2-3.) The third issue arises from the presence of multiple sets of Underwriters of Lloyds of London as defendants in the case. Lloyds of London is an unincorporated association whose member “names,” or underwriters, join syndicates which in turn join together to issue insurance policies. Indiana Gas Co. v. Home Ins. Co., 141 F.3d 314, 316 (7th Cir.1998), cert. denied sub nom. Certain Underwriters at Lloyd’s, London v. Indiana Gas Co., — U.S. -, 119 S.Ct. 339, 142 L.Ed.2d 280 (1998). Each name — frequently a natural person— bears unlimited personal liability, but may be sued only through the syndicate or the lead underwriter on a given policy. Id. at 316-17. Hundreds of names may be members of a given syndicate. Id. at 316. Under Seventh Circuit precedent, if a Wisconsin citizen is one of the names participating in any of the syndicates participating in any of the relevant policies, diversity is destroyed and the case must be dismissed. Id. at 319. To avoid this prospect, plaintiffs moved under Fed.R.Civ.P. 21 to dismiss the Lloyds underwriters as defendants. (R. 169 ¶ 4.) All parties (including all Lloyds underwriters who filed appearances, represented by counsel for Crane and for Boeing) stipulated that the underwriters were not necessary parties under Fed.R.Civ.P. 19(a). Id. Because I see no reason to suspect otherwise, I grant plaintiffs’ motion. I therefore find that plaintiffs and their decedents are all Wisconsin citizens, (R. 60 ¶¶ 5, 7, 9, 11, 13; R. 169 ¶ 1), and that all remaining defendants are foreign corporations with non-Wisconsin principal places of business (R. 68 ¶ 20; R. 72 ¶ 19; R. 169 ¶ 2-3). Thus, this court has diversity jurisdiction. II. FACTS KC-135R tanker aircraft provide air-to-air refueling to other military aircraft. The KC-135R’s fuel system includes several fuel tanks and pumps to move fuel from the tanks, both to supply the KC-135R’s own engines and to refuel other aircraft while flying. The fuel pumps are mounted inside the fuel tanks themselves. They are therefore continuously submerged in jet fuel, or, more dangerously, in a mixture of jet fuel and air, which is, as Crane acknowledges, an “explosive environment.” (R. 88 at 40.) One of the fuel pumps in the KC-135R which exploded was a Lear Siegler model RR12280A pump. This pump has a cast-aluminum housing and is driven by an electric motor. The motor is powered by four 18-gauge wire leads, three “hot,” and the fourth a neutral ground. These motor lead wires feed into the top of the pump, the “end bell.” The motor is cooled by jet fuel, which enters and exits the pump’s housing from the fuel tank through vent holes. One of the vent holes, referred to as the upper vent hole, is located at the top of the end bell, directly over the wire passage. The pump’s design thus calls for running “hot” electric wires through an explosive mix of jet fuel and air. To reduce the risk of fire and explosion, the wires are individually insulated. In addition, the design calls for a Teflon insulation sleeve to be placed over each wire. The KC-135R’s pump was recovered largely intact after the explosion and resulting fire. It was examined by both plaintiffs’ and defendants’ experts, and twice by the Air Force. These investigations revealed that the red motor lead wire (one of the “hot” wires) had arced inside the pump’s end bell, burning through three of its seven strands and its red insulation. Although a Teflon insulation sleeve is present on the red wire, plaintiffs’ and Crane’s experts agree that the red wire is approximately 5.5 inches long but that the Teflon sleeve runs for only 1.75 to 2 inches of that length. (R. 112 Ex. B ¶ 8, Ex. 23 at Fig. 53(b); R. 89 Ex. C ¶¶ 5, 7 & Ex.) The pump’s aluminum end bell housing showed extensive evidence of localized heat damage at the upper vent hole. The Air Force found that arcing created a pit nearly half-way through the pump’s housing and enlarged the vent hole itself. (R. 112 Ex. 2 ¶ III.4.) Energy dispersive X-ray analysis of the interior of the end bell revealed the presence of several elements — fluorine and carbon — found in the insulation on the motor lead wires. (R. 112 Ex. B ¶¶ 10-11, Ex. 22.) This suggests that the motor lead wires melted against the interior of the end bell. (R. 112 Ex. B ¶ 11.) In addition, X-ray analysis found copper on both the interior of the vent hole in the end bell and on the exteri- or of the end bell immediately outside the vent hole. (Id. ¶¶ 11, 13; R. 112 Ex. 22.) Copper melts at 1940 degrees Fahrenheit. (R. 112 Ex. 1 at J-9.) The presence of cooled copper both inside and outside the upper vent hole suggests that the intense heat of the arcing may have caused the copper in the wire to have melted into plasma and then forcibly ejected — through the vent hole, out from the pump’s end bell, and into the explosive mix of jet fuel and air inside the fuel tank. (Id.; R. 112 Ex. B ¶ 13.) X-ray analysis shows carbon on the exterior of the aluminum end bell near the upper vent hole, which plaintiffs’ expert attributes to fire soot. (R. 112 Ex. B ¶ 13.) A series of Air Force simulation experiments following the explosion found that molten metals, principally aluminum, could be ejected through a vent hole in a closed aluminum block when a copper wire inside the aluminum block was made to arc close to the vent hole, and that the energy in such ejected debris or plasma “is capable of igniting any flammable hydrocarbon.” (R. 112 Ex. 2 at 5.) III. SUMMARY JUDGMENT STANDARD Summary judgment is required “if the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to judgment as a matter of law.” Fed. R.Civ.P. 56(c). The mere existence of a factual dispute does not defeat a summary judgment motion; “the requirement is that there be no genuine issue of material fact.” Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986). For a dispute to be genuine, the evidence must be such that a “reasonable jury could return a verdict for the nonmoving party.” . Id. For the fact to be material, it must relate to a disputed matter that “might affect the outcome of the suit.” Id. Although summary judgment is a useful tool for isolating and terminating factually unsupported claims, Celotex Corp. v. Catrett, 477 U.S. 317, 323-24, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986), courts should act with caution in granting summary judgment, Anderson, 477 U.S. at 255, 106 S.Ct. 2505. When the evidence presented shows a dispute over facts that might affect the outcome of the suit under governing law, summary judgment must be denied. Id. at 248, 106 S.Ct. 2505. The moving party has the initial burden of demonstrating that he is entitled to judgment as a matter of law. Celotex, 477 U.S. at 323, 106 S.Ct. 2548. Once this burden is met, the nonmoving party must “go beyond the pleadings” and designate specific facts to support each element of the cause of action, showing a genuine issue for trial. Id. at 322-23, 106 S.Ct. 2548. Neither party may rest on mere allegations or denials in the pleadings, Anderson, 477 U.S. at 248, 106 S.Ct. 2505, or upon conclusory statements in affidavits, Palucki v. Sears, Roebuck & Co., 879 F.2d 1568, 1572 (7th Cir.1989). Both parties must produce documentary evidence to support their contentions. Whetstine v. Gates Rubber Co., 895 F.2d 388, 392 (7th Cir.1990). In evaluating a motion for summary judgment, the court must draw all inferences in a light most favorable to the nonmoving party. Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 587, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986). However, it is “not required to draw every conceivable inference from the record — only those inferences that are reasonable.” Bank Leumi Le-Israel, B.M. v. Lee, 928 F.2d 232, 236 (7th Cir.1991). IV. SUFFICIENCY OF PLAINTIFFS’ EVIDENCE Crane leaves largely unaddressed the merits of plaintiffs’ allegations of strict liability, negligence, and breach of warranty. For that reason, even if I found that plaintiffs had not presented adequate evidence to persuade a rational jury of the elements of their claims, Seventh Circuit law would prohibit me from entering summary judgment against them on that basis. Nonetheless, it is useful to summarize plaintiffs’ theory of why Crane’s design and manufacture of the RR12280A pump was defective; this is the central element to plaintiffs’ claims that Crane is liable under negligence, strict liability, and breach of warranty. If accepted, plaintiffs’ evidence would establish these defects: (1) most importantly, placing a vent hole in the motor lead wire passage, even though nothing in that passage requires cooling and the possibility of the motor lead wires’ arcing created an unnecessary risk of sparks, arcing, and molten metal being expelled through the vent hole into the combustible mix of air and jet fuel in the fuel tank (R. 112 Ex. A ¶¶ 7-8); (2) using a pump housing which was explosion proof tested on a version which did not include the upper vent hole, and which in practice was not explosion proof and failed to contain arcing and sparking within the pump (id. ¶¶ 11-12); (3) providing a Teflon insulation sleeve along only a portion of the red wire’s length, rather than along its full length, thereby increasing the risk of arcing and sparking from the areas unprotected by the Teflon insulation (id. ¶¶ 35-36); (4)allowing burrs and insufficiently smooth machining at the junction of the vent hole and the wire passage, which may have abraded the wires and caused them to arc directly in front of the vent hole (id. ¶ 23); (5) using a low class of aluminum casting for the pump housing, allowing molten aluminum to be expelled through the vent hole (id. ¶ 22); (6) using seven-strand and 600-volt type “E” wire rather than the more flexible, better insulated, and higher voltage nineteen-strand and 1000-volt type “EE” wire required in the specifications (id. ¶¶ 40^42, 45); and (7) tightly bundling, twisting, and turning the four motor lead wires, as well as not providing sufficient space for them, all of which weakened their insulation (id. ¶¶ 32-33). Crane has challenged the adequacy of plaintiffs’ evidence to support the elements of their prima facie claims in only two areas, explosion proof testing and evidence of defective machining and burrs. Neither challenge is appropriate for decision, because Crane raises these arguments only in its reply brief. First, Crane contends that any alleged deficiencies in the explosion proof testing are irrelevant, because the test was designed to see whether the pump design prevented sparks from being emitted (R. 112 Ex. 14 (MIL-P-5238A) ¶ 4.5.23.2.(a)), and would have little bearing upon whether molten metal could be emitted. Crane thus denies any causal link between the alleged deficiencies in the explosion proof testing and plaintiffs’ theory of how the explosion occurred. (R. 116 at 17.) Second, Crane contends that plaintiffs’ evidence of defective machining which may have allowed burrs at the junction of the vent hole and the wire passage is too speculative to be considered under Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), and must therefore be set aside. (R. 116 at 21-22.) However, these issues are not properly before me, because Crane raises them for the first time in its reply brief. Local Rule § 6.01(c) requires that, “A reply brief shall be limited to matters in reply.” The intent of this rule is to ensure that a responding party has a full and fair opportunity to respond to all of the moving party’s arguments and is not subjected to “sandbagging.” In addition, providing all arguments for summary judgment in the moving party’s initial brief allows the record and the parties’ arguments to be fully developed for the court’s consideration. When an issue is raised for the first time in a reply brief and the opposing party has no opportunity to respond, any part of the motion based on the new issue will be denied. Schimpf v. Gerald, Inc., 52 F.Supp.2d 976, 1000 (E.D.Wis.1999). Y. CRANE’S GOVERNMENT CONTRACTOR DEFENSE Crane has moved for summary judgment based on the government contractor defense. As we will see, Crane faces a high evidentiary burden: it must establish that the government approved each specific design feature in question in an exercise of discretion balancing technical, military, social, and safety considerations. Boyle v. United Technologies Corp., 487 U.S. 500, 511-12, 108 S.Ct. 2510, 101 L.Ed.2d 442 (1988). The Federal Tort Claims Act, 28 U.S.C. §§ 1346(b), 2671-2680, retains governmental immunity for “Any claim based upon an act or omission of an employee of the Government, exercising due care ... based upon the exercise or performance or the failure to exercise or perform a discretionary function or duty.” 28 U.S.C. § 2680(a). Boyle extended this discretionary function immunity to displace military contractors’ liability under state tort laws in certain cases: those where the government exercised discretion by ordering equipment from a contractor in a manner that prevented the contractor from “complying] with both its contractual obligations and the state-prescribed duty of care.” Boyle, 487 U.S. at 509, 108 S.Ct. 2510. Displacing the contractor’s state tort liability in such cases is appropriate, the Court ruled, because the government’s responsibility in “selecting] the appropriate design for military equipment ... often involves not only engineering analysis but judgment as to the balancing of many technical, military, and even social considerations, including specifically the trade-off between greater safety and greater combat effectiveness.” Id. at 511, 108 S.Ct. 2510. To demonstrate that such an exercise of governmental discretion and judgment truly was involved in a particular military procurement, and thereby relieve itself of tort liability under state law, a contractor must establish that: “(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.” Id. at 512, 108 S.Ct. 2510. The purpose of these requirements is to “assure that the design feature in question was considered by a Government officer, and not merely by the contractor itself.” Id. Consideration of the first Boyle requirement — -government approval of reasonably precise specifications — is largely sufficient to resolve the issues before me. The courts have identified several instances which satisfy the requirement. The most obvious occurs where the government provided the precise specifications and precise manner of construction. Boyle, 487 U.S. at 509, 108 S.Ct. 2510. The standard is also satisfied where there is a “continuous back-and-forth” between the contractor and the government, with the government “making final decisions as to specifications.” Koutsoubos v. Boeing Vertol, 755 F.2d 352, 355 (3rd Cir.1985). Under Boyle’s “approving] reasonably precise specifications” standard, as elaborated by one panel of the Fifth Circuit, “once the government has delegated authority to the private contractor to make important choices, the government does not exercise a discretionary function by merely accepting the contractor’s work.” Trevino v. General Dynamics Corp., 865 F.2d 1474, 1485 (5th Cir.1989). The government’s approval, “at a minimum, must involve judgment or policy choice to fall within the discretionary function exception.” Id. at 1484. The government’s merely approving the contractor’s final designs, without any substantive review or evaluation, is rubber-stamping and not approval under Boyle. Id. at 1480. In sum: If the government delegates its design discretion to the contractor and allows the contractor to develop the design, the government contractor defense does not apply. If the government has so delegated its discretion to the contractor, mere government acceptance of the contractor’s work does not resuscitate the defense unless there is approval based on substantive review and evaluation of the contractor’s design choices. Id. at 1486. Smith v. Xerox Corp., 866 F.2d 135, 138 (5th Cir.1989), decided by a different panel of the Fifth Circuit on the same day as Trevino, held that Boyle’s “approving] reasonably precise specifications” standard, 487 U.S. at 512, 108 S.Ct. 2510, was met where a contractor’s product incorporated government environmental specifications and the government “reviewed and approved [the contractor’s] final drawings and specifications.” I find Trevino’s analysis more persuasive because it emphasizes that the bare fact that a government office approved a given configuration in a contractor’s specifications is inadequate assurance that uniquely governmental discretion was used in balancing the necessary technical, military, social and safety trade-offs. Boyle, 487 U.S. at 511, 108 S.Ct. 2510. By contrast, Smith reports merely that the military “reviewed and approved” the contractor’s specifications, and does not specify the rigor of or judgment involved in the military’s review and approval process. Nonetheless, I do not consider Smith contrary to Trevino; Trevino ’s analysis would lead to the same result in Smith if the military’s review and approval of the contractor’s final drawings and specifications did not “merely accept[ ] the contractor’s work,” Trevino, 865 F.2d at 1485, but was substantive and reflected a policy or judgment choice, id. at 1484. To whatever extent Smith and Trevino do require different levels of governmental review for a contractor to be relieved of liability, the Seventh Circuit has followed Trevino’s more demanding requirement. See Oliver v. Oshkosh Truck Corp., 96 F.3d 992, 998-99 (7th Cir.1996) (citing Trevino twice and quoting it once on the issue; neither citing nor quoting Smith on the issue). The kind of substantive review and evaluation that is required to find government approval is illustrated in Harduvel v. General Dynamics Corp., 878 F.2d 1311 (11th Cir.1989). In that case, the plaintiff alleged that, due to design defects, wire chafing caused an F-16’s flight instruments to malfunction and so caused a fatal crash. In a manner similar to this case, the Air Force had issued a Request for Proposals for the F-16 aircraft, which included general specifications. In response, General Dynamics provided proposed detailed specifications, which were evaluated by the Air Force Systems Program Office: The Office conducted an extensive review of the aircraft, including the electrical system, examining specifications, drawings, and blueprints. One group of Air Force engineers was specifically assigned to review of the electrical system. The Air Force conducted independent review and analysis of the electrical system design, and evaluated designs in a Preliminary Design Review, Critical Design Review, and later Physical Configuration Audit. Id. at 1320 (citations to the record omitted). The Eleventh Circuit found that this level of detailed review of the electrical system alleged to be defectively designed was sufficient to satisfy Boyle’s requirement, 487 U.S. at 512, 108 S.Ct. 2510, that the government “approve[ ] reasonably precise specifications.” Harduvel, 878 F.2d at 1320. In Oliver, the Seventh Circuit likewise found final governmental decision-making for a product’s design where the Marine Corps assumed “complete responsibility” for testing a contractor’s prototype transport vehicle, and identified “literally ... hundreds” of changes for the contractor to incorporate into the design. Oliver, 96 F.3d at 996. Crucially, those changes required adding a new fuel tank — which exploded in a rear-end accident — on the exhaust system’s side of the vehicle. Meeting the government’s other size requirements led the contractor to place the new fuel tank only 1.5 inches from the exhaust pipe. The Seventh Circuit ruled that, regarding the placement of the new tank, “It is clear, therefore, the Marine Corps did not leave ‘the critical design decisions to the private contractor,’ Trevino, 865 F.2d at 1480, but rather that the Corps worked with Oshkosh every step of the way.” Oliver, 96 F.3d at 996-97 (citing Harduvel, 878 F.2d 1311, 1320; Kleemann v. McDonnell Douglas Corp., 890 F.2d 698, 700-01 (4th Cir.1989)). A similar analysis is presented in Snell v. Bell Helicopter Textron, Inc., 107 F.3d 744 (9th Cir.1997). In that case, the contractor proposed detail specifications and drawings, and, after requiring a number of changes, the military examined, reviewed, and approved final detail specifications and drawings. Despite this apparent evidence that the government had approved precise specifications, the court found that the contractor fell short of proving that the government had actually exercised discretion for the “the design feature in question,” Boyle, 487 U.S. at 512, 108 S.Ct. 2510: Each drawing was signed by a government representative to indicate approval. But that evidence does not establish as a matter of law that the government exercised its discretion with respect to the drive shaft and its components. On the contrary, testimony of the Bell official in charge of dealing with the military about the design of the helicopter, who attended all of the design meetings, was that there were no discussions with the government about the design of the critical isolation mounts. Snell, 107 F.3d at 748. I make three final observations about the government contractor defense before turning to the specifies before me. First, as Snell suggests, it is crucial to be clear about the burden and quantum of proof a defendant must provide on summary judgment when asserting the government contractor defense. At any stage of litigation, the contractor “carries the burden of proving each element of the government contractor defense.” Oliver v. Oshkosh Truck Corp., 96 F.3d 992, 997 (7th Cir.1996). When seeking summary judgment, the contractor must prove the first Boyle standard to such a degree that no reasonable juror could find otherwise, Celotex, 477 U.S. at 323, 106 S.Ct. 2548: that the government’s approval was a substantive exercise of its discretion to balance technical, military, social and safety trade-offs, Boyle, 487 U.S. at 511, 108 S.Ct. 2510, and that “the design feature in question was considered by a Government officer,” id. at 512, 108 S.Ct. 2510. Second, as indicated by the analysis above of the distinction between rubber-stamping and substantive and meaningful review, many contractors will not be eligible for this defense, even though their detail drawings were approved by the government and showed the design features which a plaintiff later alleges were defective. This may raise the question of what the contractor was expected to do to become eligible for the defense. The answer is that contractors, even government contractors, have no entitlement to blanket immunity from later allegations of negligence. When a contractor prepares designs in any other setting, it is liable for any negligence, even if its customer approves its designs. No immunity for negligence arises merely because the federal government is the customer. The government contractor defense is available only where the contractor could not both comply with its contract and satisfy its state-prescribed duty of care. Boyle, 487 U.S. at 509, 108 S.Ct. 2510. Where there is not adequate evidence that the government specifically considered and, after substantive and meaningful review, approved a given design feature proposed by the contractor, id. at 511-12, 108 S.Ct. 2510, the contractor must answer allegations of negligence just as it would be in any other case. Without such evidence, “it is impossible to say that the Government has a significant interest in that particular feature,” id. at 509, 108 S.Ct. 2510, and thus no interest is advanced by providing immunity for negligence associated with the particular feature. Third, that said, a defendant’s failure to establish the government contractor defense with regard to a particular design feature does not establish a plaintiffs corresponding affirmative case. It will often be the case that a contractor’s specifications and detail drawings show precisely what the finished product looked like, and that the government contractor defense does not apply solely because of a lack of adequate evidence that the government specifically considered and approved the design feature in question. The absence of that evidence, however, does not aid the plaintiffs ultimate burden of proving that the contractor violated a duty of care. A. Specifications and Approvals in the Procurement of the RR12280A Pump Considering Crane’s motion requires exploring the details, now more than thirty-five years old, of the design, specification, and procurement process for the RR12280A pump. Crane must establish the Boyle elements from what evidence still exists; to prevail on the government contractor defense, it must prove its case in the face of the inevitable gaps in the record after such a time. However, there is no fundamental unfairness here; whatever relevant records have been lost were at one time in its control or the control of persons with whom it was in privity of contract, and would never have been in plaintiffs’ control. 1. Overview In December, 1962, Boeing and the Air Force granted final approval to a predecessor to the RR12280A pump, the RR12280, and a very similar companion pump, the RR12220, as pumps for Boeing to install as original equipment on particular B-52 aircraft. (R. 136 Ex. 4 (SCD 10-30322) at sh. 6; Ex. 9 (SCD 10-30319) at sh. 6). Production of the RR12280 and RR12220 began shortly thereafter, but within three months, due to problems with the pumps’ stator phase insulation and motor life, Boeing issued “no further procurement” orders for both pumps. In May, 1963, Crane submitted to Boeing plans for modified versions of the two pumps, the RR12280A and RR12220A, which addressed the motor problems. In December, 1963, Boeing granted the modified versions “design proposal approval” — Boeing’s preliminary confirmation that the submissions met the basic specifications and that Crane could submit more detailed test results. Although Crane and Boeing exchanged correspondence for another two years about the test procedures needed for the RR12280A and RR12220A to win higher levels of approval from Boeing, Boeing never granted them any further approval. Independently from its submissions to Boeing for inclusion as original B-52 equipment, in May, 1964, Crane submitted the RR12280A and RR12220A to the Air Force in response to an Air Force request for proposals for replacement B-52 pumps. Crane received a contract in August, 1964 to supply the Air Force with 460 RR12280A pumps and 106 RR12220A pumps. (R. 136 Ex. 21.) In November, 1964, after Crane submitted qualification test results, the Air Force approved the pumps and requested Crane to begin production and delivery. (R. 89 Ex. 42.) Ten years later, in 1974, the Air Force begin using the RR12280A booster pump as a replacement override pump on KC-135 aircraft. 2. Relevant requirements governing military contractors Would-be military contractors in the 1960s needed to satisfy several overlapping layers of requirements. First, the military issued its own specifications. Some of these specifications were of quite general applicability. One specification, for example, provided standards for various grades of casting for all metal pieces used in aü-planes (R. 89 Ex. 18 (intended to be Ex. 37) (MIL-C-6021D)); a second provided requirements for all insulated electrical wire (R. 112 Ex. 15 (MIL-W-16878)). Other military specifications were more detailed, e.g., a 1954 general specification for fuel booster pumps, which provided pages of specifications about the performance requirements for such pumps and the testing procedures they needed to satisfy. (R. 112 Ex. 14 (MIL-P-5238A).) Second, lead contractors on particular projects issued Source Control Drawings (“SCDs”) for particular components of their overall projects. The relevant SCD here is Boeing’s SCD 10-30322, issued in November, 1960. This SCD provided general specifications for the B-52’s auxiliary booster pump and tracked which contractors’ models had received Boeing and Air Force “final approval” to be installed as original B-52 equipment. (R. 136 Ex. 4 (SCD 10-30322).) Boeing also issued a “General Requirement Supplement to Source Control Drawings for Model B-52,” which supplemented the SCD. (R. 136 Ex. 7 (SCD Supplement)). The SCD included a sketch with a largely empty box for the pump’s interior design, but with much more detail about the external cable which would power the pump. (R. 136 Ex. 4 (SCD 10-30322) at 13.) As with the military’s general specification for fuel booster pumps, Boeing’s SCD included pages of performance and testing requirements. In addition, it specifically required that the pump must satisfy the military’s general specification for fuel booster pumps. (Id. ¶ 2.1 (incorporating MIL-P-5238A by reference); ¶ 3.1.2.1 (“The pump shall conform to the applicable requirements of MIL-P-5238A unless specifically superseded by this specification.”).) Third, contractors submitting proposals in response to lead contractors’ SCDs or military requests for proposals were required to submit their proposed product’s specifications, test procedures to qualify their products, and test results. The specifications would typically include both an assembly drawing, showing how the individual components would fit together, and detail drawings, providing the information necessary to fabricate each individual component. Crane prepared assembly drawings for both the RR12280 (R. 136 Ex. 31) and the RR12280A (R. 89 Ex. 1; R. 136 Ex. 32; R. 112 Ex. 5 (drawing RR12280A)). Crane also prepared a detail drawing of the end bell housing used on both the RR12280 and RR12280A pumps (R. 89 Ex. 2; R. 112 Ex. 8; R. 136 Ex. 30 (drawing RG21604)). In addition to preparing the assembly and detail drawings, Crane submitted detailed test procedure proposals for the RR12280 to Boeing and for the RR12280A pump to both Boeing and the Air Force. Crane corresponded extensively with both Boeing and the Air Force regarding the qualification test procedures for the RR12280A, and, based on its similarity to the RR12280, won approval from both Boeing and the Air Force to conduct less extensive qualification test procedures on the RR12280A. Crane submitted extensive test results for the RR12280 pump to Boeing, receiving “final approval” (with the Air Force’s approval), and submitted test results for the RR12280A to the Air Force. Further details about Crane’s negotiations with Boeing and the Air Force and the relevant specifications are provided below as necessary. B. Government Contractor Defense as Applied to Plaintiffs’ Allegations 1. Failure to comment: The argument from silence Crane initially contends that because the Air Force approved its final detail drawings in November, 1964, after having requested four specific changes on several drawings in a June, 1964 letter (R. 89 Ex. 26), it necessarily gave substantive, meaningful review to all remaining design features. (R. 88 at 31.) Likewise, Crane contends that the Air Force’s addressing only the immediate cause of failure — faulty phase insulation causing a short circuit inside the motor — -following a two-day tear-down of two failed RR12220 pumps on February 28 and March 1, 1963, indicated Air Force approval of all other features. (R. 116 at 16, 23, 24.) I reject these arguments from silence. For summary judgment purposes, the absence of governmental comment upon a given design feature, even if other features were commented upon, simply cannot establish that, after substantive and meaningful review, a government officer considered and approved the design feature in question in an exercise of discretion balancing technical, military, social and safety trade-offs, as required by Boyle, 487 U.S. at 511-12, 108 S.Ct. 2510. It is just as possible to infer from the government’s silence that the government rubber-stamped the design feature in question, in rebanee upon the contractor’s sound discretion, and did not conduct an independent review. For summary judgment purposes, all inferences are drawn in favor of the non-moving party. It is worthy of note that three of the Air Force’s four requested design changes in the July, 1964 letter addressed the RR12280A’s rotor shaft. (R. 89 Ex. 26.) The Air Force had notified Crane only six months before, in December, 1963, that rotor shafts had sheared on the first fifteen RR12280 pumps that had been received for repair. (R. 89 Ex. 4.) At a January, 1964, meeting held at an Air Force facility to discuss this problem, Crane agreed to redesign the RR12280’s rotor shafts and to incorporate the redesigned rotor shafts into the RR12280A. (R. 89 Ex. 6 ¶ 3.d.) It is thus no surprise that when Crane submitted its RR12280A proposal four months later, in May, 1964, the Air Force looked closely at the rotor shaft design. This is not a case where the military assumed “complete responsibility” for testing Crane’s design or identified “literally ... hundreds” of changes for Crane to incorporate into its design. Oliver, 96 F.3d at 996. In Oliver, the contractor’s placing a fuel tank dangerously close to the exhaust system was directly due to the government’s requiring the contractor to change its initial design, which called for a single fifty-five gallon fuel tank on the opposite side of the vehicle from the exhaust system, to provide two seventy-five gallon fuel tanks. Id. at 996-97. By contrast, the government’s four requested changes in this case — three of which related to the rotor shaft — in no way led Crane to change its design so as to place an upper vent hole in the wire passage, to avoid explosion proof testing the pump while it was operating, to avoid placing Teflon insulation over the fub length of the motor lead wires, to use' the class of aluminum casting it did, or to route the motor lead wires as it did. The government’s silence about these design features in the face of its comments upon four other design features, three of which related to the rotor shaft, does not establish for summary judgment purposes that these design features were considered and approved by a government official after substantive and meaningful review. Crane responded to the Air Force’s request by accepting two of the requested changes and arguing against the wisdom of the others. The letter concluded, “We trust this letter and the attachments are sufficient for evaluation and acceptance of our offer.” (R. 89 Ex. 29 at 3.) Crane argues that once this statement was made, the government had discretion over every aspect of the pump’s design and testing, therefore entitling Crane to the government contractor defense. (R. 88 at 27.) This is mistaken. Design discretion originated in Crane, because Crane developed the pump. “[M]ere government acceptance of the contractor’s work does not resuscitate the defense unless there is approval based on substantive review and evaluation of the contractor’s design choices.” Trevino, 865 F.2d at 1486. The government’s failure to comment provides no evidence of approval based on such substantive review and evaluation. Similarly, the Air Force’s participating in the two-day tear-down of two non-functioning RR12220 pumps in 1962 does not even establish that Air Force representatives took conscious note of the relevant design features—placement of the upper vent hole, amount of Teflon insulation on the motor lead wires, and so on— much less that they made a substantive and meaningful review and approved them. The purpose of the inspection was to determine why the pumps had failed and how to prevent, similar future failures, not to evaluate the safety of pump features wholly unrelated to the immediate failures. Contrary to Crane’s contention (R. 116 at 16), the Air Force’s involvement in this tear-down does not establish a “continuous back and forth” between the supplier and the military regarding the location of the vent hole, the amount of Teflon insulation on the motor lead wires, or any other aspect of the pump’s design. Oliver, 96 F.3d at 998 (quoting Harduvel, 878 F.2d at 1320 (quoting Koutsoubos, 755 F.2d at 355)). The Air Force’s focusing on the immediate cause of failure, faulty phase-to-phase insulation, can in no way be considered an approval of all other features of the pump for purposes of the first Boyle element. 2. Specific Alleged Design Defects a. Location of vent hole I turn now to Crane’s government contractor defense as it applies to plaintiffs’ specific allegations. Plaintiffs first allege that the RR12280A’s upper vent hole is defectively located: the upper portion of the pump’s end bell does not need to be cooled, and placing a vent hole there, next to the motor lead wires, imposes a risk that an arc from one of the wires in the wire passage “could cause sparks or molten fragments to escape through the vent hole.” (R. 112 ¶ 8.) Crane contends that the government “approved reasonably precise specifications [for] the design feature in question,” Boyle, 487 U.S. at 512, 108 S.Ct. 2510, namely, the location of the upper vent hole. Crane bases its argument on four pieces of evidence. Three of them are comprised of the Air Force’s exposure to the vent hole on various drawings before it purchased the pump, and at the February 28 and March 1, 1962 tear-down of two failed RR12280 pumps, all without having commented upon the vent hole’s placement. (R. 116 at 14-16.) This is the argument from silence considered and rejected above. Crane’s best argument and evidence concern some notes from a November, 1962 engineering conference. Both the argument and the evidence are offered only with Crane’s reply brief. (R. 116 at 13-14 & Ex. F.) As we have already seen, a moving party may not raise a new argument in a reply. In addition, a moving party may not submit new evidence with a reply to support a new argument. Local Rule § 6.01(c); Boustead v. Barancik, 151 F.R.D. 102, 107 (E.D.Wis.1993) (Gordon, J.). So long as the non-moving party did not place the new matter in issue in its response, the new matter and the evidence supporting it may both be disregarded. Hartley v. Wisconsin Bell, Inc., 930 F.Supp. 349 (E.D.Wis.1996) (Gordon, J.); Baugh v. City of Milwaukee, 823 F.Supp. 1452, 1456 (E.D.Wis.1993) (Evans, C.J.), cited with approval in Beck v. University of Wisconsin Bd. of Regents, 75 F.3d 1130, 1134 n. * (7th Cir.1996). Crane did not refer to the November, 1962 engineering conference in its original motion, and plaintiffs did not raise it in them response. It is a wholly new issue, and the notes taken at that conference accordingly do not respond to matters placed in issue by plaintiffs’ response. Even on the merits, Crane’s new evidence would not establish that the Air Force approved the vent hole placement in an exercise of discretion balancing technical, military, social, and safety considerations. Boyle, 487 U.S. at 511, 108 S.Ct. 2510. The purpose of the conference was to respond to Crane’s request to qualify both the RR12280 and RR12220 pumps with a single testing for nine required tests. To win this permission, Crane needed to establish that the RR12280 and RR12220 pumps were identical in all relevant aspects. Notes taken by a Crane employee read, “Explosion Test: End bell openings identical.” (R. 116 Ex. F.) The notes for all nine tests total one and half pages; none is more than eight or ten words. Air Force representatives attended the conference. This new evidence, even if considered, would be insufficient to establish as a matter of law that “the design feature in question was considered by a Government officer, and not merely by the contractor itself.” Boyle, 487 U.S. at 512, 108 S.Ct. 2510. The notes suggest only that the Air Force was told what the legend at the bottom of the end bell housing drawing already indicated, that the two pumps used the same end bell housing design, and what. Crane’s later Qualification Test Report said, that the pumps had identical housings. Crane simply needed to establish that the pumps’ housings and openings were identical. The most obvious inference from the notes, “End bell openings identical,” is that Crane pointed out — or perhaps simply asserted — that the two pumps used the same drawing, RG21604, for their end bell' housings. The very brevity of the notes on the nine tests suggests that the conference participants did not review every design feature, or even every opening in the pumps’ housing. Even if the end bell housing drawing was reviewed, the meeting was not called to assess the safety of Crane’s design. There is no evidence of “discussions with the government about the design of the critical [vent hole placement].” Snell, 107 F.3d at 748. According to its own account, Crane added the upper vent hole during a period that it was not required to consult with the government. (R. 116 at 18.) It has failed to establish that at any later time “the design feature in question was considered by a Government officer, and not merely by the contractor itself.” Boyle, 487 U.S. at 512, 108 S.Ct. 2510. b. Explosion proof testing Both the military’s specifications and Boeing’s SCD set out particular explosion proof tests to qualify the B-52 booster pump. Crane conducted explosion proof testing on the RR12280 pump on June 5, 1961. Plaintiffs contend that various defects in the testing shows that Crane does not pass the second Boyle test with regard to the location of the upper vent hole: even assuming that the government “approved reasonably precise specifications” for the vent hole’s location, Crane did not “conform[ ] to those specifications,” Boyle, 487 U.S. at 512, 108 S.Ct. 2510, because it did not conform to the required testing specifications. This analysis is slightly confused. The specifications to which Crane allegedly did not conform here are testing specifications, rather than vent hole placement specifications. When a plaintiff alleges defects in how a product was tested, and the testing was governed by government specifications, the defendant may raise the Boyle defense for the testing on the same terms as it could for the product itself. In re Aircraft Crash Litig. Frederick, Md., 752 F.Supp. 1326, 1337 (S.D.Ohio 1990). Thus, a military contractor may avoid liability for defective testing if it can show (1) that the United States provided reasonably precise specifications for the testing; (2) that the testing conformed to those specifications; and (3) that the supplier warned the United States about any dangers in the use of the testing that were known to the supplier but not to the United States. Boyle, 487 U.S. at 512, 108 S.Ct. 2510. I may now examine plaintiffs’ individual arguments about Crane’s testing, each of which focuses on showing that Crane did not conform to the government’s testing specifications. First, and crucially, plaintiffs contend that the version of the pump which Crane tested did not have a vent hole at the site where they allege that the KC-135R’s pump emitted molten metal: the upper vent hole was added later. Crane’s detail drawing of the RR12280 pump’s end bell housing, RG21604, shows a 1/32” diameter hole at the top of the end bell, labeled “B3”; the drawing’s revision history indicates that Revision B3, “added 1/32 dia. hole,” was added to the drawing on June 29, 1961 — more than three weeks after the June 5 explosion proof testing on the RR12280. In addition, Boeing’s SCD required a spark plug to be placed within 1/4” of every vent, drain, and other outlet “normally open to the tank.” (R. 136 Ex. 4 (SCD 10-30322) ¶ 4.2.25.2.) The test report’s schematic drawing of the June 5, 1961 test setup describes spark plugs 2, 3, and 4 as being, respectively, 1/4” from each of three specific vent holes — the left top motor vent hole, the center top motor vent hole, and the right top motor vent hole — -and makes no reference either to a recently-added upper vent hole or to a spark plug at that location. (R. 136 Ex. 33 (TR-996) at 27A.) There is thus strong circumstantial evidence that the version of the pump that was explosion proof tested did not have an upper vent hole. Crane replies that during the relevant period, it was free to change the pump’s design as it wished, and thus might have added the vent hole not only before the detail drawing was updated to reflect the new vent hole, on June 29, 1961, but also before the June 5, 1961 explosion proof testing. (R. 116 at 18.) As evidence that it actually did add the vent hole before the explosion proof testing, Crane offers a competing interpretation of the schematic drawing. (Id. at 19.) The schematic drawing indicates that spark plug 5 was placed 1/4” from the “motor lead exit hole.” (R. 136 Ex. 33 (TR-996) at 27A.) The lab records refer to this spark plug as being placed at the “lead wire boss.” (R. 136 Ex. 33 (TR-996) at Exp. Test no. 813-1.) I understand both phrases to refer to the feed-through terminal at the top of the pump, through which the motor lead wires exited the pump into the supplying cable assembly. Crane contends that because the upper vent hole is the only hole “normally open to the tank” in the vicinity, the SCD’s requirements prove that Crane must have placed spark plug 5 by the upper vent hole, and not by the “motor lead exit hole” or “lead wire boss.” This argument assumes that which it seeks to prove. Specifically, it assumes that there was an upper vent hole when the schematic was drawn. From that assumption, it concludes that the schematic’s clear language must, by reference to another document (which goes uncited), actually refer to the assumed upper vent hole and not refer to what it says, the motor lead exit hole. Unless an upper vent hole is assumed from the start, this argument fails. In addition, this interpretation of the schematic ignores the context. As already noted, the schematic describes spark plugs 2, 3, and 4 as being a certain distance from each of three identified vent holes. If the schematic proves, as Crane contends, that spark plug 5 were the same distance from a recently-added upper vent hole, its next line would naturally describe the spark plug as “1/4” from upper vent hole,” and not what it says, “1/4” from motor lead exit hole.” Crane’s reading of the schematic is forced and unnatural. I find that its speculations do not satisfy its burden to “demonstrate the absence of a genuine issue of material fact.” Celotex, 477 U.S. at 323, 106 S.Ct. 2548 (emphasis added). Plaintiffs’ second objection to how the explosion proof testing was conducted concerns whether the pump was running during the test. No fewer than three portions of the relevant military specification required this. The explosion proof testing was to be done on an operating pump, in part, to ascertain that the pump functioned properly (R. 112 Ex. 14 (MIL-P-5238A) ¶ 4.5.23.2.(a)); the test apparatus was to include a power source to run the pump (id. ¶ 4.5.23.1.(b)(2)); and, for each air-fuel mixture, the pump was to be operated at its rated load for fifteen minutes during testing, except during charging (id. ¶ 4.5.23.2.(e)(2)). Crane’s schematic drawing for the test does not show any cables or power source to run the pump during the test, although, by contrast, it does show wiring to run the spark plugs for the test. (R. 136 Ex. 33 (TR-996) at 27A.) Moreover, the lab records are silent as to whether the pump was run during the tests. (Id. at Exp. Test no. 813-1.) The military specifications indicate that after the pump was tested while running for each air-fuel mixture, the pump was to be turned off so it could be refilled with an explosive mixture, which was then to be exploded by a spark plug. The lab reports record only one set of notes on the actual running of the test: between changes in air-fuel mixture, heaters were turned on to bring the test chamber’s heat up to the specified level. (R. 136 Ex. 33 (TR-996) at Exp. Test no. 813-1.) It is therefore not an unreasonable inference that if the pump had been running while a given air-fuel mixture was tested, and was then (1) turned off and (2) re-filled with an explosive mixture (3) which was exploded before (4) the test chamber’s heaters were turned on to prepare for testing the next air-fuel mixture, the lab report would describe all of this, and not just (4) turning on the heaters. I infer in plaintiffs’ favor as the non-moving parties that the pump was not operating during the testing. Crane appears to concede that the pump may not have been operated during explosion proof testing, but contends that Boeing’s SCD implicitly relieved it of the requirement. (R. 116 at 8.) Boeing’s SCD specified that it governed where it “specifically superceded” the military specifications. (R. 136 Ex. 4 (SCD 10-30322) ¶ 3.1.2.1.) The SCD required that the pump’s temperature not exceed 390F, and required that it have a temperature limiting device that would disconnect the pump’s power if its temperature exceeded the maximum allowed. (Id. ¶¶ 3.1.2.17-18.) The SCD’s specifications for the explosion proof testing required that: The explosion proof test shall be conducted in accordance with MIL-P-5238, paragraph 4.5.23.1, except as follows: ... The test chamber ambient shall be 160 ± 5°F. The motor case temperature shall be raised to 400 ± 10°F. The unit shall be soaked under these temperatures for a minimum of one hour prior to the explosion proof tests. (Id. ¶¶ 4.2.25, 4.2.25.1.) The military specifications required that the test chamber in which the pump was placed for the test was to be no higher than 135°F. (R. 112 Ex. 14 (MIL-P-5238A) ¶ 4.5.23.2.(d).) Crane argues that, due to the SCD’s requirement that a temperature limiting device render the pump inoperable at temperatures over 390°F, the SCD superceded the military specification’s requirement that, the pump be operated during the explosion proof tests. (R. 116 at 8.) This argument fails. The SCD twice specified that the military specification governed the SCD, unless, in one version, the SCD “specifically superceded” the military requirements (R. 136 Ex. 4 (SCD 10-30322) ¶ 3.1.2.1), and, in the other version, “except as follows” (id. ¶ 4.2.25). The SCD does not specifically supersede the military specification’s requirement that the pump be operated during explosion proof testing anywhere, including ¶ 4.2.25. In addition, the SCD explicitly details several specific divergences from the military requirements for other aspects of the explosion proof testing, which indicates that Boeing was quite able to specifically supersede the military requirements when it wished to. Moreover, it is noteworthy that these other divergences — for example, requiring that the test chamber’s ambient temperature be 165°F rather than under 135°F, and that the test be conducted with a spark plug near every vent, drain, and other opening, rather than with only one spark plug for the entire pump — all made the testing more stringent than the military’s requirements, not less. Especially because one purpose of the government’s specifications for explosion proof testingvwas to ensure that the pump functioned properly while operating (R. 112 Ex. 14 (MIL-P-5238A) ¶ 4.5.23.2.(a)), it seems highly unlikely that, if the SCD required the testing to be done on a non-operating pump, it would imply that through silence, rather than by specifically superseding the military requirement. Further, the Boyle defense requires the contractor to prove that it conformed to the government’s specifications. Boyle, 487 U.S. at 512, 108 S.Ct. 2510. Here, those specifications undisputedly required the pump to be operated during explosion proof testing. Even if, for the sake of argument, Boeing’s SCD specifically superseded those government specifications, Crane has not shown that Boeing had any governmental authority to reduce the rigor of the government’s specifications, or that Boeing’s doing so would constitute an exercise of uniquely governmental discretion in balancing technical, military, social and safety trade-offs. Boyle, 487 U.S. at 511, 108 S.Ct. 2510. Crane has thus not shown, for purposes of summary judgment, that it conducted this aspect of its explosion proof testing in conformity with the government’s specifications. The government contractor defense does not protect it from any attendant liability. Third, plaintiffs contend that, because it was not witnessed by a government observer, the explosion proof testing on the RR12280 was defective as applied to Crane’s later contract to supply RR12280A pumps. In a June, 1964 letter, the Air Force required that all qualification testing for the RR12280A pump be witnessed by a government inspector. (R. 89 Ex. 26 ¶ 2.) Crane submitted “qualification by similarity” data, including lab records, from the RR12280’s June 5, 1961 explosion proof testing to demonstrate that the RR12280A pump met the Air Force’s specifications. (R. 89 Ex. 34 (TR-1307) ¶ 6.4.) All three pages of the lab records have no initials on the line for “observer.” (R. 136 Ex. 33 (TR-996) at Exp. Test no. 813-1). Indeed, Crane identifies forty pages of lab records for other tests which government observers did initial. (R. 88 at 15 & n. 9, 29 & n. 18.) The presence of those initials strengthens the inference that government observers did not witness the explosion proof testing; and all inferences must be drawn in favor of plaintiffs as the non-moving parties. Crane replies first that the Air Force requirement that a government inspector witness all qualification tests was provided in only “a single [and] solitary letter” (R. 116 at 19), and therefore was not “reasonably precise” within the meaning of Boyle. The letter reads: “In addition to the above requirements, qualification tests are mandatory before the pumps will be acceptable to the Air Force. The tests are to be performed at your plant or an approved testing laboratory witnessed by a Government inspector.” (R. 89 Ex. 26.) I am unable to determine in what relevant way this specification is imprecise, or how repetition would make it more precise. Crane’s second reply is that the front pages of two different qualification test reports (one each for the RR12280 and RR12280A) were signed as having been “coordinated by” government officials— and that each report included the lab reports of the apparently unwitnessed explosion proof testing. (R. 89 Exs. 34 (TR 1307), 37 (intended to be Ex. 18) (TR 996).) Those signatures plainly do not establish that a government inspector witnessed the explosion proof testing. Crane has thus failed to demonstrate, for summary judgment purposes, that the government contractor defense renders it immune from its liability, if any, stemming from the defects alleged in its explosion proof testing. c. Teflon insulation The red motor wire lead which melted has Teflon insulation along two inches of its length. Plaintiffs assert that Crane’s failure to provide Teflon insulation along the entire length was negligent; Crane argues that it is protected by the government contractor defense. The parties agree that the detail drawing of the pump’s stator (R. 89 Ex. 39) shows a two-inch Teflon insulation sleeve on each motor wire lead, and that, by contrast, the assembly drawing shows Teflon insulation over the entire length. (R. Ill ¶ 74; R. 116 at 22-23.) Plaintiffs point to this disparity as evidence that there was no “reasonably precise” specification under Boyle, 487 U.S. at 512, 108 S.Ct. 2510. (R. 110 at 23.) Crane replies that according to engineering convention, detail drawings prevail if they conflict with an assembly drawing, and so the detail drawing showing two inches’ worth of Teflon insulation constitutes a “reasonably precise” specification. (R. 116 at 23.) I need not resolve this dispute. Even if the drawings, taken as a whole, do constitute a “reasonably precise” specification for Teflon insulation to cover only two inches of each motor lead wire, there is still no evidence that a government official considered this design feature and, after substantive and meaningful review, approved it in an exercise of discretion balancing technical, military, social and safety trade-offs. Boyle, 487 U.S. at 511-12, 108 S.Ct. 2510. Crane identifies as its only evidence the two earlier arguments from silence: The detail drawing of the stator was among those sent to the Air Force before it approved the pump as a whole; and the Air Force would necessarily have seen that there were only two inches of Teflon insulation on each motor lead wire in the February 28 and March 1, 1963 tear-down inspection. (R. 116 at 23.) These arguments have been considered and rejected. d. Aluminum casting quality Plaintiffs allege that the pump’s housing was made from a defective class of aluminum casting. Crane argues that even if it had used a defective class — -which it does not concede — the government approved Crane’s using that class and so Crane is protected by the government contractor defense. The best aeronautical castings, Class I, are those “the single failure of which during any operating condition would cause loss of the aircraft or one of its major components, loss of cont