United States v. General Motors Corp.

656 F.Supp. 1555 (1987)

UNITED STATES of America, Plaintiff,
v.
GENERAL MOTORS CORPORATION, Defendant.

Civ. A. No. 83-2220.

United States District Court, District of Columbia.

April 14, 1987.

Lawrence Moloney, Thomas A. Barba, Dept. Justice, Civil Div., Washington, D.C., for plaintiff.

Thomas A. Gottschalk, Kirkland & Ellis, Washington, D.C., for defendant.

DECISION AND ORDER

JACKSON, District Judge.

The United States brings this action pursuant to the National Traffic and Motor Vehicle Safety Act of 1966, Pub.L. No. 89-563, 80 Stat. 718 (codified as amended at 15 U.S.C. § 1381 et seq. (1982 and Supp. III 1985)) (the "Act"), at the instance of the National Highway Traffic Safety Administration ("NHTSA"), U.S. Department of Transportation, against defendant General Motors Corporation ("GM"), a motor vehicle manufacturer. The complaint alleges that an entire generation of GM automobiles, its 1980 X-cars, are defective in that they are predisposed to a phenomenon known as "premature rear wheel lock-up" entailing a potential for loss of vehicle control.^[1] Counts I and II allege, respectively, that GM determined (or should have determined), pre-production, that certain components of the X-cars' rear braking system were responsible for the condition, and that, post-production, it learned that deterioration of front braking components in service were exacerbating it, but in each instance it failed in its statutory duties to notify the Secretary of Transportation and the cars' owners of, and to remedy, the "defect." Counts III and IV allege that the two recalls of some X-cars which GM did conduct in 1981 and 1983, at NHTSA's urging, were each inadequate to cure the defect. Count V alleges that GM failed to submit accurate and complete information in response to NHTSA's queries in the course of its administrative investigation of the 1980 X-cars. And Count VI charges a violation of a NHTSA regulation in GM's omission of NHTSA's "hotline" telephone number in the recall letters sent X-car owners in the 1981 recall campaign. The United States prays for a judgment declaring that GM committed the several violations alleged, an injunction directing it to recall and effectively repair all of its 1980 X-cars, and an order assessing civil monetary penalties against it.

By its answer GM denies that its 1980 X-cars are, or have ever been, defective, and that it violated the Act or the regulation as alleged.^[2]

Trial commenced March 13, 1984, and continued, with intermittent recesses, until May 16, 1985, when the Court approved and filed a stipulation of the parties closing the evidentiary record.^[3] Following post-trial briefing, closing arguments were heard on February 25, 1986, and the case submitted.

Upon the facts found as hereinafter set forth in accordance with Fed.R.Civ.P. 52(a), following trial without a jury, and the conclusions of law drawn therefrom, for the reasons stated the Court will enter judgment for defendant dismissing all counts of the complaint (except Count V) with prejudice.

I.

Enacted in 1966 "to reduce traffic accidents and deaths and injuries to persons resulting from traffic accidents," 15 U.S.C. § 1381; see generally 1966 U.S. Code Cong. & Admin.News at 2709, the Act imposes a duty upon automobile manufacturers to notify both NHTSA and the owners of their vehicles when they learn the vehicles possess safety-related defects, and then to remedy those defects without charge to the owners. 15 U.S.C. §§ 1411, 1414.^[4] The term "defect" embraces "any defect in performance, construction, components, or materials in motor vehicles or motor vehicle equipment." 15 U.S.C. § 1391(11). Prima facie proof of a defect in a class of vehicles requires only a showing that a "significant" number of them have failed in consequence of the defect, a significant number being merely a "non-de minimis" quantity; it need not be "a substantial percentage of the total." United States v. General Motors Corp., 518 F.2d 420, 438 & n. 84 (D.C.Cir.1975) ("Wheels"). Evidence of a non-de minimis number of defect-induced failures establishes a rebuttable presumption of the existence of a class-wide defect in the vehicles, and the burden of proof shifts to the manufacturer to rebut the government's prima facie showing. The manufacturer may also assert affirmative defenses, e.g., that the failures resulted from unforeseeable owner abuse or neglect of vehicle maintenance, id. at 427, 438, as to which, of course, the manufacturer has the burden of proof from the outset.

Under § 1411 the government must also show that the manufacturer not only knows of the supposed defect in its vehicles, but that it made a "good faith" determination that the defect relates to motor vehicle safety as well.^[5] A defect is "related to motor vehicle safety" if it presents an "unreasonable risk of accidents." 15 U.S.C. § 1391(1). As in the matter of determining the existence of a vehicle "defect," Wheels, 518 F.2d at 435-36, so also is "commonsense" analysis to be employed in ascertaining what constitutes an unreasonable risk, United States v. General Motors Corp., 565 F.2d 754, 757 (D.C.Cir.1977) ("Carburetors"), but, as a general proposition, any defect that involves a loss of control presumptively presents an unreasonable risk of accidents as a matter of law. United States v. General Motors Corp., 561 F.2d 923 (D.C.Cir. 1977) (per curiam) ("Pitman Arms"), cert. denied, 434 U.S. 1033, 98 S.Ct. 765, 54 L.Ed.2d 780 (1978).

II.

Formal planning for what was to become GM's "1980 X-car" began in 1975.^[6] The X-car was to be GM's first high-volume front-wheel-drive automobile with a transversely mounted engine to be sold as a "coordinated car line." Because an X-car model was to be offered by each of four of GM's car divisions, its design and development was coordinated through a "project center," established in early 1976, to which engineers from both car and component divisions were assigned. The project center was administratively a part of GM's corporate engineering staff, but all engineering decisions were, ultimately, the responsibility of the chief engineers of the several car divisions: Chevrolet, Pontiac, Oldsmobile, and Buick.

Particular divisions were assigned lead responsibility for the evolution of specific vehicle systems. Thus the Buick division acquired overall lead responsibility for the X-car's braking system. Other divisions with expertise in particular brake components were given primary responsibility for those components: the Delco-Moraine division for the front brake caliper and linings, the rear brake drum, and the master cylinder; the Inland division for the rear brake linings for automatic transmission X-cars; and the Chevrolet division for the front hub and rotor assembly, all being coordinated in their efforts by Buick's brake engineers.

As is common practice in any GM car program, the brake engineers first selected the generic type of brake components and sized them based on projected vehicle mass.^[7] Engineering drawings were made, from which prototype components were produced, tested in laboratories, and then tested on similarly sized peer cars (called "component" cars). As development progressed the evolving system was installed on various pre-production versions of the proposed X-car itself (called successively, "prototype," "pilot," and "lead unit build" cars). Test results were reviewed, and designs modified to improve performance as the tests indicated.

GM's divisional brake engineers evaluated the X-car's brake system in operation principally on four of the company's established driving "schedules": the "FMVSS-105 docket" tests, the Pike's Peak descent, a West Virginia mountain road course, and a city driving schedule in and about Los Angeles.^[8] Other vehicle systems were simultaneously being tested, by the other groups of engineers primarily concerned with them, on the same schedules, and, of particular significance for this case, on one long-distance general durability test known as the "R15-23 schedule."

Having chosen the front disc/rear drum brake design for the X-car, GM brake engineers elected to use semi-metallic linings for the disc brakes, believing them to offer superior resistance to fade at the higher brake temperatures they expected to occur at the heavier front end of the vehicle.^[9] Organic linings were to be employed on the rear drum brakes upon the supposition that they would be less susceptible to environmental degradation.

The 1980 X-car was also to be equipped with two "fixed-slope" proportioning valves in its hydraulic system (one valve per rear wheel) to limit the line pressure going to the rear brakes in moderate to heavy braking. The valves compensate for dynamic force transfer by "proportioning" rear line hydraulic pressure to incremental front line pressure above a certain "break," or "knee," point which, in the X-car, was set at 350 psi. (For example, a 41% "fixed-slope" proportioner valve allows, in theory, 41% of the amount of the incremental line pressure applied to the front brakes above the "break" to reach the rear brakes as well.) In harder brake applications, therefore, more line pressure would be directed to the front brakes relative to the rear to compensate for the dynamic transfer of normal force to the front.^[10]

Chevrolet initially proposed a 9.34-inch vented rotor for the front disc brake, but the Buick engineers, reviewing the design in March, 1976, tentatively concluded that a smaller rotor might not provide sufficient heat dissipation and gave consideration to two larger rotors: a 9.75-inch rotor in a new caliper design, and a 10-inch rotor in an existing design. After evaluating both rotors the engineers decided upon the 9.75-inch rotor upon the theory that its smaller mass would enhance fuel economy without compromising performance.

Semi-metallic linings, which were thought to offer several advantages over organic materials such as asbestos, e.g., increased fade resistance, superior high-speed effectiveness, and greater durability, were gaining favor throughout the automotive industry in the late 1970's. General Motors' brake engineers considered two semi-metallic materials for the front brakes of the 1980 X-car: the DM8032 material, which GM itself had recently developed as a successor to its own first semi-metallic lining, and the BX 7161A material, a Bendix product used on certain Ford and Chrysler vehicles. Its own tests led the GM engineers to conclude that the DM8032 offered equivalent or better performance in all parameters, and they selected it.

GM engineers also weighed two alternative rear drum brake configurations for the 1980 X-car. The "leading-trailing" system presses both brake shoes against the drum when hydraulic line pressure is applied without interaction between the shoes; the "duo-servo" system, ultimately chosen for the X-car, employs the rotating action of the drum to cause the forward, or "leading," shoe to apply additional force against the "trailing" shoe, theoretically supplying more output.

A decision in late 1977 to reroute the 1980 X-car's parking brake cable to distance it from the heat generated by the catalytic converter appeared to diminish the parking brake's mechanical efficiency, and the engineers abandoned the original plan to use 4050/4050 organic rear brake linings in favor of more "aggressive" 4035/4050 linings. Then, as the X-car program progressed through 1978, the projected weight of the vehicle increased somewhat, and the engineers grew apprehensive as to whether manual transmission X-cars would pass the federal parking brake test.^[11] They therefore made a further change from the 4035/4050 to the still more aggressive Bendix 3198/3199 rear linings (also used on some Ford and Chrysler models) on the manual transmission X-cars.

GM engineers were generally satisfied with the X-car brake system they had settled upon. Pilot and lead unit build cars passed FMVSS-105 certification tests using either the 3198/3199 or the 4035/4050 rear lining combinations, and the system achieved what GM engineers considered to be acceptable ratings on the Pike's Peak schedule for effectiveness, wear, temperature behavior, and overall performance. Of several different brake configurations tested on the West Virginia mountain schedule, an initial production configuration, using 4035/4050 rear linings and a 41% proportioning valve, received highest ratings overall for brake balance and effectiveness. The engineers were also generally content with the Los Angeles brake durability test results, although on some of the runs on the L.A. schedule drivers had submitted reports of incidents of "rear wheel (or brake) lockups."^[12]

In the latter half of 1978, a Durability Test and Development ("DT & D") group of GM's corporate engineering staff ran pre-production X-cars on a new vehicle durability test, the R15-23 schedule, which was then under development by DT & D, intended to be more abusive than the usage to which any single car in consumer service would be subjected over the life of the vehicle. Two versions of the R15-23 schedule — one of 65,000, the other of 100,000 miles — were run by X-cars. The drivers, hired from the general population, were asked to report any aspect of design, performance, or durability that displeased them or seemed to be unusual about the vehicles, from which the DT & D staff prepared written test incident reports ("TIRs") to be sent to the engineers responsible for the design of the pertinent system. In the fall of 1978, Buick began to receive TIRs from DT & D describing instances of "premature" rear wheel lockups reported by drivers running one R15-23 schedule. Buick engineers inspected and rode the suspect vehicles, concluding that the incidents were, once again, single rear wheel lockups, a conclusion confirmed for them by the discovery of unilateral glazed or cracked linings and found only on the offending wheel.^[13]

Nevertheless, in mid-December, 1978, the DT & D staff gave a status report on the R15-23 durability testing of the X-car generally to senior GM management gathered for a product review in Mesa, Arizona. One item prominently on the agenda was the subject of the reports of "rear brake overheating and premature lockup," as to which GM's then-president remarked that he did not want the X-car to go into production with any problem that might be so major as to require a "retrofit" after the cars had been built and stockpiled.

GM's management then directed that a "task force" be formed, under Buick's leadership, to investigate the DT & D incidents and advise whether the brake system should be produced as designed. The task force assembled consisted of some 20 engineers and supporting staff, drawn from Buick, DT & D, Inland, Delco-Moraine, GM Research, Chevrolet, and the corporate engineering staff. It undertook investigations of quality control in lining production, the metallurgy of the components, design of the rear brake, brake balance, and the severity of the R15-23 schedule, and ordered further vehicle and laboratory tests and engineering and mathematical analyses.

On January 23, 1979, the task force unanimously recommended against a delay in the production of the 1980 X-cars as designed, and to proceed with production on schedule. The task force's recommendation was successively presented on February 1st to the X-car project center, on February 12th to a meeting of the chief engineers, and on February 15th to a general managers' meeting attended by senior corporate management, at each of which the conclusions of the task force, viz., that the single-wheel lockups were due to unilateral overheating and would not be repeated in the field, were accepted. GM thus allowed production to proceed, and the X-car was first released to the market on April 19, 1979.^[14]

Altogether three rear brake lining combinations were used on 1980 X-cars as produced. Nearly 200,000 manual transmission X-cars were built with the Bendix 3198/3199 rear linings. Just over 30,000 automatic transmission cars had the 4035/4050 rear linings. Approximately 825,000 X-cars were equipped with the 4050/4050 combination. Roughly 246,000 automatic transmission and 47,000 manual transmission cars had the 41% proportioner valves; the remainder were equipped with the nominal 27% valve.^[15]

III.

The primary function of any motor vehicle's braking system is, of course, to enable the driver to slow the vehicle at a desired rate, varying its speed and position in traffic or on the road, or bringing it to a stop. When the brakes are applied, retarding forces develop between each tire and the pavement which cause the car to slow. The magnitude of the braking force that can be generated at each tire/road interface is limited by the adhesion characteristics of the tire and the road, which are expressed in terms of the coefficient of friction (µ) of the tire/road interface.^[16] The higher the coefficient of friction, the greater the braking force potentially available to slow or stop the car.

Braking force at the tire/road interface is created by application of the brakes. By depressing the brake pedal, the driver causes hydraulic pressure, in a disc brake, to clamp the brake linings against both sides of a metal rotor mounted on the wheel; in a drum brake, the brake linings are pressed against the inside walls of a cylindrical drum, also mounted on the wheel. The resulting friction force between the linings and the rotor or drum produces brake torque which will vary with changes in the coefficient of friction at the lining/rotor or lining/drum interfaces, the hydraulic pressures, and the physical dimensions of the brake components. The greater the brake torque created within the wheel, the greater the potential brake force that can be generated at the tire/road interface, within, of course, the limits of tire-to-road adhesion.

As a matter of principle brakes are designed to generate friction forces sufficient to take optimum advantage of tire/road coefficients of friction. Since stopping distance is also a function of mass, however, more braking force is necessary to stop a fully loaded car than a lightly loaded car in the same distance. Thus, as a rule, brakes must be designed to generate the larger braking forces sufficient to achieve the highest deceleration rates attainable for a fully loaded vehicle.

On slippery road surfaces, with lower coefficients of friction, the maximum rate of deceleration is correspondingly reduced, because the braking force that can be generated at the tire/road interface is less than on a better pavement with a higher coefficient. Yet a driver can still apply the brakes as hard (and generate the same brake torque) as when the car is on a high-coefficient road surface. Regardless of road quality, however, if the driver applies the brakes sufficiently hard so that the braking forces exceed the available friction forces at the tire/road interface, each tire at which the limit of adhesion is reached will "lock up," or skid.

"Brake" (or "wheel") "lockup," therefore, does not represent a systemic mechanical malfunction, or a broken or failed part.^[17] Brake lockup can occur, and the locked wheel will skid, notwithstanding the brake system and all its components are performing precisely as intended, simply because the driver has applied the brakes with too much force relative to extant tire and road conditions. Skidding results from the interaction of the driver, the brake system, and the tire/road interface, and while on occasion it may be both alarming and dangerous, skidding in and of itself is not a failure of vehicle "performance" nor indicative of a brake "defect."^[18]

The consequences of a skid are likewise explained by the laws of physics. In addition to steering input, a car is also controlled in its speed and direction of travel by the tire/road friction forces. When a tire is simply rolling straight, without either being accelerated or decelerated, it uses relatively little of the available friction limit to maintain its speed, leaving most of the potential tire/road friction forces available for steering and stopping. Friction forces resulting from any combination of steering and braking diminish the ability to generate control forces at the tire/road interface, and, when the limit of adhesion is reached, the tire can no longer generate any side forces for path or attitude control. Thus, since steering is accomplished through the front wheels, turning the wheels to avoid a collision is totally ineffective when they are locked and skidding.

Moreover, a sliding tire has a lower tire/road coefficient of friction than that of a rolling tire on the same surface. Since the deceleration rate of a car cannot exceed the coefficient of friction at the tire/road interface, a car can stop in a shorter distance if the limit of adhesion is not reached and the tires continue to roll during braking. For any given tire/road interface, there is a "peak µ," representing the maximum deceleration rate attainable with the tire rolling, and a "slide µ," which represents the lower rate the car can achieve while skidding on the same surface. A car with one or more of its wheels "locked" and skidding has a diminished deceleration potential, and, hence, a correspondingly lengthened stopping distance, in comparison to another with all tires close to the limit of adhesion but still rolling.

An ideal brake system would, therefore, operate to approach the limit of adhesion at all four wheels simultaneously, whatever the coefficient of the surface upon which the car is traveling, making the maximum friction forces at each tire/road interface available to the driver for control purposes. Such a car, i.e., one that will develop precisely the brake torque at each wheel as necessary to achieve simultaneous incipient four-wheel lockup as necessary to achieve simultaneous incipient four-wheel lockup is said to have "ideal brake balance."^[19] With ideal brake balance, a car possesses its shortest possible stopping distance capability; the maximum braking forces at each tire can be sustained, and the maximum deceleration rate attained, before any wheel locks up.

As a practical matter, however, ideal balance can never be achieved by any brake design throughout the entire range of operating and loading conditions to which a car is subjected.^[20] The limit of adhesion at each tire, being a function of both the tire/road coefficient and the normal (vertical) force on the wheel at any given instant, is a transient or dynamic value that will vary from stop-to-stop and even during a single stop.^[21]

The tire/road coefficients that a car may encounter are also affected by tire condition,^[22] particularly tread wear, and inflation pressure, and may vary from point to point on an apparently uniform surface due to environmental factors, e.g., patches of rain or snow, or contaminants such as gravel, vegetation, oil, or debris. And the surface characteristics of roads themselves change over time as they become worn or damaged.

There are two principal factors affecting the normal force on each tire, and therefore its maximum braking force, one being the loading condition of the car. In general, a lightly loaded vehicle has a greater percentage of its weight on the front tires, less when it is filled to capacity with passengers or cargo. Thus, even in a static state, ideal brake balance for a particular car would require a different distribution of brake torque among the four wheels in a fully-laden as opposed to a lightly-loaded condition. Then, of course, the effect of differences in static weight distribution is compounded by the dynamic transfer of normal force which occurs from the rear wheels to the front wheels during braking. As a vehicle decelerates, the inertia of its mass causes an increase in the normal force on the front wheels and a corresponding decrease in the rear, the magnitude of the rear-to-front dynamic force transfer being a function of the deceleration rate.^[23]

Because ideal brake balance for a car differs with virtually every deceleration it undergoes, some wheel will be virtually certain to lock up before another, if lockup occurs at all, and it is the sequence of lockup between the front and rear wheels that has become the central focus of this case. "Front brake lockup," or "front lock," as the terms are used here, refers to the situation in which both front wheels lock before either rear wheel during a brake application, and a car's brake system is described as "front biased" if the car will experience front brake lockup first in a stop sufficiently hard to produce lockup at all. Conversely, "rear brake lockup," or "rear lock," contemplates both rear wheels' locking before either front wheel, i.e., a "rear biased car" will lock its rear wheels first if the point of wheel lock is reached during a stop.^[24]

Certain adverse control consequences attend both a sustained front or rear brake lockup. Steering control over the direction, or path, the car is traveling is reduced in either case, because side forces cannot be developed at the sliding tires, and stopping distance is extended, because the braking forces at the sliding tires have been diminished by the lower slide coefficient. These same adverse consequences, however, also occur in instances of four-wheel lockup, whether consequent to ideal brake balance or simply a sufficiently hard brake application.^[25]

In most modern cars most of the weight is borne by the front wheels which are therefore expected to contribute more braking force in routine operation than the rear wheels. The drop in tire/road coefficient from peak to sliding which results when the front wheels alone lock extends a car's minimum stopping distance more than with rear brake lockup. A sustained front brake lockup on an X-car, for example, will extend its minimum stopping distance by 25% over that attainable when its rears lock first.^[26]

On the other hand, a front lock condition is considered "stable" in an engineering sense because, even though a coinciding lateral force, or "moment," whether endogenous or exogenous, may cause the car to oscillate, or "fishtail," its natural tendency is to return to its original attitude before the force was applied. Conversely, a rear brake lockup results in an "unstable" condition; a lateral force or moment can initiate a "yaw," or rotation around the car's vertical axis, which, unless brought under control, is self-energizing and will continue until a state of equilibrium is achieved. In sum, a sustained front lockup will extend minimum stopping distance vis-a-vis a rear lock; with sustained rear brake lockup, a car is more apt to yield to yaw, or "spin-out."

Stability, however, does not connote controllability, nor does instability render the vehicle uncontrollable.^[27] A car with front lock can be directionally controlled only by the correct driver response of brake modulation, for unless the lockup condition is relieved, the car will tend to slide in a straight line following the path it was traveling when its wheels locked. Thus, a curve cannot be negotiated nor an object directly ahead avoided, while front lock persists.

In the rear lock condition, a car will stop in a straight line only if no significant lateral force is exerted on it. Once the car does begin to yaw, the capacity of the front tires to generate sufficient side forces to enable the driver to regain control over the car's attitude by steering may be overwhelmed as the moment reinforces itself. Once again, however, as directional control may be recovered following front lock, so also may the driver bring a yawing car under control with its rears locked — by properly modulating the brakes.^[28]

IV.

Front-to-rear "brake balance" refers to the distribution of braking torque or braking force between the front and rear wheels. It is the engineering parameter for ascertaining the relative propensity of cars to proceed to either front or rear brake lockup first. No matter the brake components used, their condition, or how they work in conjunction with one another, their effect on the sequence of wheel lockup depends upon the brake torque they generate and the resulting braking forces created at the tires. By measuring either brake torque or braking force at the four wheels, engineers can determine the distribution of braking between front and rear tires, quantifying the results in terms of degrees of "front bias" or "rear bias," from which the sequence and timing of wheel lockup can be theoretically predicted for any given stop on any surface coefficient.

Engineering witnesses for both sides ultimately adopted a concept of "brake efficiency" as a uniform expression of the degree of a car's brake balance. Brake efficiency is a percentage calculated by dividing the tire/road coefficient into the maximum deceleration rate (in units of gravity) a car can achieve before experiencing front or rear brake lockup. Thus, a 100% efficient braking system equates to ideal brake balance; the car can attain a .8g deceleration rate on a .8 peak µ surface before experiencing wheel lock.^[29] A car that is 75% efficient on the same surface would experience brake lockup, either front or rear, at a .6g deceleration rate. And as between two cars known to be rear biased, one at 80% efficiency and the other 70% efficient, on the same surface, the latter will develop rear brake lockup at a lower deceleration than the more "efficient" car.

Measurements of brake torque have historically been made with "torque wheels," specially instrumented wheels employing strain gauges to record brake torque electronically throughout a deceleration. Braking force at the tire/road interface can be measured by a device developed by GM in conjunction with its trial preparation for this case called a "road transducer pad" ("RTP"), an instrumented road surface that records longitudinal forces and deceleration rates at the tire/road interface when rolling wheels are run over it during a brake application. Although the government was dubious of the RTP technique, RTP and torque wheel measurements proved to correlate well, and measurements by both methods were employed, when it served their purposes, by the engineers who testified for both sides.

The competing considerations for or against front bias or rear bias as a desirable design objective have been understood and appreciated within the brake engineering community for many years. Technological limitations simply require that compromises be made in reaching an engineering consensus as to an optimum braking distribution that addresses both stopping distance and vehicle control objectives in the myriad loading and road-surface conditions most cars will encounter in years of normal service. Differing views as to how much to emphasize stopping distance or stability at the expense of the other have provoked academic controversy in the engineering literature which is still unresolved, and has prompted those nations with mature automotive industries to adopt somewhat divergent regulatory approaches. In the United States minimum stopping distance has historically prevailed over stability as the paramount objective of effective braking, and this preoccupation, which has only recently been called into question, is expressed in the prevailing federal regulatory standard governing braking, and reflected as well in the published writings of American brake engineers and the measured brake balances of cars built by the American automotive industry from the early 1970's well into this decade.

In 1976 NHTSA promulgated Federal Motor Vehicle Safety Standard 105-75, 49 C.F.R. § 571.105 ("FMVSS-105") establishing mandatory performance requirements for every new car sold in America "to insure safe braking performance under normal and emergency conditions." 49 C.F.R. § 571.105(S2). FMVSS-105 specifies that a car be able to stop within prescribed distances on a series of "effectiveness" or stopping distance tests. Nine of the ten effectiveness stops must be made at gross vehicle weight. Only one is to be made in the lightly loaded condition. Contrasting with its exacting requirements as to stopping distances, FMVSS-105 imposes no express requirements relating to vehicle stability, other than that the vehicle not deviate from the roadway during deceleration. The locking of any one wheel is permitted at any speed, any two or more wheels may lock at speeds below 10 mph, and no wheel lockup sequence is prescribed. The emphasis on stopping distance for the heavily laden vehicle, of course, implicitly encourages brake balance design tending toward rear bias at lighter loads.^[30]

In its subsequent rulemaking proposals in 1979 and 1981 NHTSA continued to stress stopping distance as the primary safety factor. In 1979, when NHTSA proposed extending FMVSS-105 to light trucks and vans to improve their braking capability, it justified the proposal by observing that studies showed "small percentage reductions in stopping distance consistently result in proportionately larger reductions in accidents or accident severity." 44 Fed.Reg. 60116 (1979). Then, having succeeded in extending FMVSS-105 to light trucks and vans, NHTSA rejected petitions for reconsideration, calculating that a 4% improvement in stopping distance capability would result in a 5 to 9% reduction in accidents in which brakes are used, 46 Fed.Reg. 61892 (1981), and noting again that improved stopping distances not only avert some accidents but also reduce the speeds at which the remainder occur, thereby reducing their severity. Id. at 61893.

The European Common Market countries have taken a different regulatory approach in their braking standard. Regulation 13 of the United Nations Economic Commission for Europe ("ECE") demands less braking efficiency of the heavily laden car, i.e., it tolerates somewhat longer stopping distances, but prohibits wheel lock altogether and prescribes express limits to the degree of rear bias a car may exhibit, disallowing it in all but mid-range decelerations. In other words, the European approach reflects more concern with stability than stopping distance.^[31]

V.

NHTSA conducts its investigations of suspected safety-related defects in motor vehicles through its Office of Defects Investigation ("ODI"). ODI is headed by a Director, under the supervision of an Associate Administrator for Enforcement, who reports, in turn, to the NHTSA Administrator. ODI's phases of investigative activity proceed from an "engineering analysis," which may be followed by a formal case "investigation" and conclude in the "Phase I Report," the agency's initial defect determination and notification to the manufacturer. If contested, the matter then goes to a public hearing, and, ultimately, results in NHTSA's final determination and an order for recall if warranted.

NHTSA initiated an engineering analysis of GM's 1980 X-body vehicles in November, 1979, after observations made at the FMVSS-105 compliance testing (conducted by NHTSA's Office of Vehicle Safety Compliance) raised its suspicions that 1980 X-cars might be predisposed to unanticipated lockups of the rear wheels while being braked. By the time the engineering analysis had been completed in June, 1981, NHTSA had on hand some 212 complaints of the misadventures of X-car drivers, 58 of them culminating in accidents, which might have been preceded by a rear wheel lockup. A search of NHTSA's computer bank revealed some 54 additional complaints about the 1980 X-cars, compared to none for any other front wheel drive cars, including GM's own 1981 X-cars. Consequently, a formal case investigation was opened on July 2, 1981.

On July 6, 1981, the ODI Director wrote to GM stating that NHTSA believed "the rear brake system of the 1980 X-body vehicles (utilizing the 41 percent valves and aggressive brake linings) contains an engineering defect which has safety-related implications...." and urged GM to commit itself to "corrective action ... within five (5) working days." GM's response of July 8, 1981, stated that the company would "initiate a recall modification relative to the involved vehicles," and, thereafter, on August 5, 1981, GM undertook to recall 47,371 manual-transmission-equipped vehicles for a proportioning valve change. Although the recall itself covered only certain early-production manual-transmission-equipped vehicles, i.e., those manufactured with a 350 × 41% proportioning valve, GM also authorized dealers to make the proportioning valve change on automatic transmission-equipped vehicles if "it becomes necessary to satisfy an owner complaint ... in the interest of maintaining customer satisfaction." The company also authorized a change in the rear brake linings for complaining owners of later production manuals, i.e., those manufactured initially with the 350-27% proportioning valve.

ODI had developed a test plan in late June, 1981, to attempt to identify what might be mechanical causes for the complaints it was receiving, in the course of which it would examine the X-cars' rear brake linings, brake drums, proportioning valves, and parking brake cable routing. Tests conducted in July, 1981, and repeated somewhat more formally in the fall of 1981 at NHTSA's Engineering Test Facility in East Liberty, Ohio, indicated to NHTSA's engineers that the components having most pronounced effect upon rear brake output were, indeed, the proportioning valve and the rear brake linings, whether or not they were "causes" of the drivers' mishaps.

On January 14, 1983, NHTSA notified GM that the agency had made an initial determination, pursuant to 15 U.S.C. § 1412, that all 1980 X-body vehicles — manuals and automatics — equipped with more aggressive rear brake linings contained a defect which related to motor vehicle safety, and scheduled a statutory proceeding under 15 U.S.C. § 1416 to review the adequacy of the 1981 recall. Public announcement of the initial determination was given nationwide publicity, including NHTSA's release to the television networks of film clips of an X-body test vehicle undergoing a dramatic skid-and-yaw, and on February 9, 1983, GM capitulated. It notified NHTSA that it would voluntarily recall all of its 1980 X-cars equipped with manual transmissions, and certain early-production automatics, too. Accordingly, with the necessity for a compulsory recall order abated, NHTSA cancelled the public proceedings scheduled to begin the following month. On March 18, 1983, GM specified the corrective action it intended to take: not only would it install new rear brake linings and park brake cables on the recalled vehicles, it would also "inspect the front brake system and provide replacements of parts necessary to provide uniformity of the entire brake system."

While awaiting the recall specifications, on March 4th NHTSA sent GM a comprehensive "special order and document production request," a subspecies of process akin to an administrative subpoena, pursuant to 15 U.S.C. § 1401(c). The special order directed GM to provide NHTSA with copies of internal GM documents not previously furnished relating to the "premature rear brake lockup" phenomenon NHTSA was investigating. GM responded in three installments, between March 25 and March 31, 1983, the final installment enclosing the collection of TIRs generated during the R15-23 schedule, which revealed to NHTSA for the first time, it says, that not only had GM's own test drivers reported "rear wheel lockup" incidents on pre-production X-body prototypes more than two years before, but also that GM management had felt compelled to create an unprecedented "task force" to deal with the very problem NHTSA was investigating.

Less than a week later NHTSA issued a formal administrative subpoena to GM requiring it to produce both records and one or more knowledgeable officials to testify about them. Thereafter, administrative depositions and document production continued until NHTSA aborted the administrative proceedings in favor of the instant action.

On August 3, 1983, when the complaint was filed, NHTSA claimed to be in possession of more than 2,000 consumer complaints it classified as incidents of probable "premature rear wheel lockup." Included among them were multiple instances of accidents, injuries, and fatalities; if the incidents were attributable to a vehicle defect, the defect was indisputably safety-related.

VI.

For reasons to become apparent, notwithstanding the extensive engineering tests conducted by both sides on X-cars, other cars, and their various components, the government chose to rely primarily, throughout trial and thereafter, upon evidence of X-car "consumers'" personal experiences to meet its burden of proof that GM's 1980 X-cars are generically afflicted with a safety-related defect. That evidence took several forms: 1) in-court testimony of 12 live-witness consumers who had lost control of their 1980 X-cars while attempting to slow or stop; 2) depositions of absent consumers in which similar incidents were described; 3) unsolicited written complaints sent to the government or directly to GM from around the country recounting still more such incidents; 4) an assessment of complaints received, analyzed by a vehicle dynamicist and a statistician in conjunction with one another, which purportedly establishes that a substantial majority of them reported experiences which were "consistent with" instability accompanying early rear brake lockup; and 5) comparisons of the absolute numbers and normalized rates of complaints received by the government regarding the 1980 X-car, as compared to the lesser numbers of complaints made about other cars.

According to the government, the 1980 X-car has been the subject of the largest number of reports of what it terms "yaw instability" of any car in NHTSA's history. As of the end of February, 1985, more than two months before the end of the trial, the total number of complaints received by NHTSA which it was prepared to classisfy officially as cases of "probable premature rear brake lockup" exceeded 4,000. When trial began the number was in excess of 3,500, of which over 300 reported injuries or fatalities, and nearly a third mentioned at least property damage accidents.^[32] The number of similar complaints about other car models, by contrast, was negligible, NHTSA says, except for two others (one of them a GM product) also currently under investigation.

The testimony of the government's 12 typical "consumers" who appeared at trial (owners and/or drivers of 1980-model GM X-cars from various parts of the United States and Canada) is the centerpiece of the government's circumstantial proof-of-defect-by-failure-of-performance-alone evidence. Pretermitting questions of admissibility, the Court assumes that their experiences are representative of those related in truncated form in most, if not all, of the affidavits/declarations, depositions, and unsworn complaints offered by the government for cumulative purposes, or, in other words, the affiants/declarants, deponents and complainants would have given similar testimony if called.

The consumers had certain characteristics in common; in others they were unique. Four were male, eight female. The males acknowledged responsibility for the maintenance of their vehicles, while the females generally entrusted maintenance to male members of their households. They were all mature, ranging between 37 and 73 years of age. All had many years' driving experience (none less than 20) with many different types of automobiles, and all had generally good driving records. Their driving experience had been acquired in geographic locations as diverse as southern California and Ontario, Canada, New York City and Waukesha, Wisconsin. And the experiences they described which the government contends evinces the "defect" in their X-cars occurred in every climatic condition, ranging from heavy snow to dry summer heat, and on every sort of road: freeways, interstates, two-lane country roads, multi-lane urban arteries, bridge approaches, and residential neighborhood streets. None of them had ever experienced the phenomenon to which they testified in any other vehicles they had driven in all their years of driving. All professed to be able to distinguish it from control difficulties associated with ice or snow. All had made some effort to ascertain and correct the cause, both through and outside GM channels (some with more persistence than others), without success. Some retained their 1980 X-cars; others had traded them in for cars made by other manufacturers, and several had traded for later-year-model X-cars.

The phenomenon to which each testified, however — variously described, but recognizable as similar nevertheless — was a "yaw," or a swerve by the rear to a marked degree, either left or right, from the axis of travel upon the application of "moderate" to "moderately hard" brake pressure, while traveling at relatively modest rates of speed. Each asserted that he or she "lost control" of the car's direction of travel altogether until coming to rest, at angles varying from 30° to 180° from the original heading. Each had had more than one such experience, some as many as six, which were separated, however, by intervals of weeks or months, sometimes years, during which the vehicle had behaved unremarkably. Since the incidents would occur intermittently and without warning, they could not be anticipated and prepared for. Several of them had resulted in collisions with stationary roadside objects or other vehicles unable to avoid them, inflicting significant property damage. None of the consumers was ever able to convince General Motors, or a GM dealer, that what had happened to them was the fault primarily of the vehicle and was amenable to mechanical correction.

The testimony of the consumer witnesses was supplemented with a complaint analysis by the government's vehicle dynamicist who reviewed 109 declarations procured by government attorneys from other X-car owners who had complained to NHTSA of similar experiences.^[33] Three-fourths of them, in his opinion, described experiences "consistent with" rear brake lockup instability, and, using the dynamacist's criteria for so classifying them, a government statistician made projections with respect to the remainder of the complaints, concluding that over 1,400 of them reported incidents which could likewise be characterized as "consistent with" rear wheel lockup instability. (Almost half of them had come from owners of automatic transmission X-cars which had never been recalled.) By cumulating the consumers' testimony with that of the deponents, three-quarters of the declarants, and the 1400-plus complainants, the government contended it had demonstrated, by a "failure of performance" of a clearly non-de minimis number of vehicles, the existence of a safety-related defect in 1980 X-cars which two recalls had failed to remedy.

Corroborative of the consumer evidence, the government asserted, was the documentation divulged by GM under compulsion which not only confirmed that the consumers' experiences were not illusory, but supplied proof that GM had, in fact, determined for itself that its 1980 X-cars were prone to "rear wheel lockup" before they went into production and had nevertheless allowed them to be sold to the public. In the aggregate, according to NHTSA, GM's internal writings constituted a virtual admission of a breach of its statutory duty.

The chief of NHTSA's Defects Evaluation Division, ODI, testified to his analysis of the thousands of documents supplied by GM by the end of the administrative inquiry. According to his reconstruction of the X-car brake design process from the documents, beginning in mid-1978 GM test drivers had begun to submit their reports of "rear brake" or "rear wheel" lockups, and the reports had been circulated throughout the corporation to the considerable consternation of senior management. At the December, 1978, meeting at Mesa, Arizona, ostensibly called to review the general "durability" test results on the X-car before the start of production, the most exalted of GM's executives, including its president, were told that the testing had revealed, among other design "deficiencies," a problem with "rear brake lockup" which remained "unresolved" despite efforts to correct it.

After the Mesa meeting, and the formation of the "task force" to consider whether to delay the start of production and to propose solutions for the "lockup problem," in late January, 1979 (less than two weeks after its initial meeting) the task force declined to recommend a delay of production start-up, even though it had not found an explanation for the problem and its testing had not been completed, while simultaneously suggesting, nevertheless, various design changes clearly indicating that it knew brake design should be moving in the direction of diminished rear braking, e.g., reducing the hydraulic pressure to the rear brakes by changing the proportioning valve slope and/or break point; using less aggressive rear brake linings; and installing finned drums to reduce rear brake temperatures.^[34]

Even after the X-car had gone into production in January, 1979, GM continued its "durability" testing, using production X-cars in addition to lead unit build models, and test drivers continued to report incidents of brake lockup, prompting a GM senior vice president of engineering to expostulate, in a contemporaneous memorandum NHTSA finds particularly incriminating:

Don't you know that you never lock the rear wheel brakes first?!!

How are such product decisions made?

What event caused the design responsible division to change their mind on this matter?

How could we miss something so obvious?!!

How can GM put out such a system?

Engineering Staff is not doing its job!

As late as May 7th the same vice president wrote:

"Every time I ask, I am told the ``X' car brakes are fixed. These tests do not indicate they are. What do we have that does?"

(On July 11, 1979, finned drums were first installed on test cars, and thereafter the internal lockup reports apparently ceased during the remainder of the durability testing.)

Soon after the first X-cars went on sale in April, 1979, General Motors, too, began to collect complaints from its employees and customers of the sort that NHTSA characterized as "consistent with" rear brake lockup. In 1980, GM began to catalogue the complaints according to the various brake component configurations then in service. Its records reveal that X-cars with 27% proportioning valves, 4050/4050 rear linings, and finned drums were generating fewer complaints from the field than any of the other configurations. Yet it made no effort to do anything about those otherwise equipped until its first recall the following year.^[35]

VII.

Once the government had completed its prima facie presentation, and the Court had denied GM's motion to dismiss pursuant to Fed.R.Civ.P. 41(b), first GM (in its case-in-chief), then the government (in rebuttal), and, finally, GM (in surrebuttal) presented extensive evidence of the results of actual tests made on X-cars, their commercial competitors, and the components of both. The trial record is replete with engineering data.^[36]

The Court finds the vehicle test results to be the most objective, least ambiguous or equivocal, and hence the most convincing evidence adduced. It is the primary basis for this decision.

General Motors measured the brake balance of 73 "current configuration" X-cars,^[37] 24 of which had been the subjects of consumer complaints. NHTSA measured the brake balance of 30 X-cars (including later models), three of which were complaint vehicles. NHTSA also determined the wheel lock sequence of more than 140 additional current configuration X-cars, without, however, measuring their brake balances.^[38] Then, in order to compare the X-car measurements to some external standard, both parties also tested a number of competitive, or "peer," cars (referred to by GM as "state-of-the-art" cars). GM initially tested 57 competitive vehicles, nearly all of which dated from the 1979-83 model years. Then, reacting to NHTSA's intimations that the competitors might not be fairly representative, GM tested an additional 97 cars from particular model lines, and a statistically random sample of 101 other competitive cars — 104 Ford Fairmonts, 57 AMC Alliances, and 55 Chrysler Cordobas, all but the Alliances dating from the 1979-81 model years. NHTSA evaluated 41 competitive cars for wheel lock sequence, but did not measure the brake balance of any.

The results of the test programs were definitively consistent, regardless of which party conducted the tests and recorded the data, or the manner in which the test cars were obtained. The results conclusively disprove the existence of any common engineering idiosyncrasy in the braking performance of 1980 X-cars not found in their competitors, no matter how configured, and whether or not it could be termed a "defect."

First, the percentage of competitive cars found to be rear biased in the "as received" condition, lightly loaded, was not only substantial; it exceeded that of the X-cars. NHTSA's initial evaluation of wheel lock sequence of 102 X-cars and 41 peer cars indicated that a driver was more likely to encounter rear bias and/or rear brake lockup in a non-GM car than in an X-car on both low and high coefficient surfaces.^[39] In each of the three samples of competitive cars chosen by GM and NHTSA for testing, well over 50% of them were rear biased on a .5 peak µ surface and an even greater percentage were rear biased at higher declerations.^[40]

Second, the braking efficiencies of rear-biased competitive cars in customer service were found to be generally lower than the efficiencies of rear biased X-cars, including so-called "complaint cars." Whereas competitive cars ranged as low as 60% in efficiency, the most inefficient X-cars were at 70% or higher.

Third, the brake balance of all X-cars measured in terms of their braking efficiencies fell well within the brake balance "envelope" established by the extremes of the competitors from the same and later model years.^[41]

Fourth, the envelope of current-configuration X-car braking efficiencies at all decelerations was more compact, and clustered in closer proximity to the ideal brake balance curve, than those of other model lines from which a substantial number of vehicles were tested, e.g., the Alliance, Cordoba and Fairmont. Of models of which multiple samples of at least ten vehicles were tested, not one was shown to have a narrower range of brake efficiency, or to possess lesser degrees of rear bias, than the 103 current configuration X-cars.

Fifth, when the "design intent" brake balance of current configuration X-cars was tested with the brakes "rebuilt and burnished," it was revealed to be front biased, even in the lightly loaded condition.^[42]

In making these findings, the Court accepts the parties' respective measurements, whether made with torque wheels or the RTP, having been given no persuasive reason to do otherwise. Each side expressed reservations about the other's test protocols, but the correlations and consistency of the measurements allay any uncertainty over the extent to which test procedures may have affected the results. Ultimately, NHTSA's brake engineers tacitly conceded that the tests conducted by both did not demonstrate, individually or collectively, a greater degree of rear bias on X-cars than their competitors.

Count I of the complaint alleged that, "for reasons relating to several distinct components" thereof, the rear braking system of the 1980 X-cars caused "premature rear wheel lock-up." Yet the government was never able to identify any "components" of the rear braking system which, separately or in conjunction with one another, consistently caused the X-cars' rear wheels to lock before the front wheels, increased rear brake output over time, or rendered X-cars as a group rear biased either when new or with use. (The measured vehicle data, to the contrary, reflected that the X-cars' rear brake output generally decreased with use).

Count II of the complaint alleged that corrosion of the front brake components diminished their effectiveness with a consequent shift in brake balance to the rear. Yet front brake corrosion was not shown to reduce front brake output, thus leaving brake balance between front and rear unaltered. GM engineers measured the brake balances of 19 X-cars received from consumers with varying degrees of corrosion of their front brake components; the most severely corroded were nevertheless front biased, and no correlation was found between front brake output and the degree of corrosion. Even corroded caliper pins could not resist the clamping force of a brake application, and the only effect of rotor corrosion appeared to be an insignificant increase in front brake output.^[43]

Having found among the test vehicle population no 1980 X-cars possessing greater degrees of rear bias than competitive cars, or, for that matter, exhibiting any particular propensity to rear brake lockup, NHTSA postulated the existence of "worst case" vehicles somewhere in the undiscovered X-car universe by combining the extremes of adverse brake torque measurements made upon different X-cars tested. Such projections, however, are not only purely hypothetical, and do not even remotely approach by the measurements actually made on more than 100 X-cars, they were all but disavowed by NHTSA engineers who acknowledged that "worst-case" projections are essentially speculation rather than a valid engineering analysis.^[44]

Finally, instrumented "complaint" X-cars, other X-cars, and competitive vehicles were driven through a series of handling and control tests by GM engineers on wet and dry surfaces at all deceleration ranges. The tests were filmed, and they demonstrate repeatedly and dramatically that any car — indeed, all cars — will yaw in conducive circumstances with the front wheels locked and the rears still rolling. They will also frequently maintain a constant direction of travel notwithstanding a complete rear wheel lock alone.

In short, it appears that it is the unique character of each application of each vehicle's brakes, each deceleration being a never-to-be-replicated confluence of factors such as the immediate surface coefficient of friction, rate of deceleration, tire and brake lining condition, vehicle loading, direction of travel, driver reaction, and the like (of which brake balance at the moment is but one), which will ultimately determine whether, and to what extent, a braked vehicle will yaw. There is simply no engineering evidence of any peculiar property of X-cars generally that renders them in any way exceptional insofar as having a predisposition to yaw.^[45]

VIII.

It is, of course, only "safety related defects" in its vehicles which raise for the manufacturer a duty to notify and repair under the Act. If direct evidence of a "defect" in 1980 X-cars' braking systems is lacking, its presence might nevertheless be inferred circumstantially from accident statistics showing X-cars to be disproportionately involved in accidents of the sort likely to begin with skid-related yaws. It was GM, however, not the government, that presented a "risk analysis" comparing the relative rates of accident involvement of 1980 X-cars with three groups of competitive cars, drawing upon accident data from two of NHTSA's own sources and ten state compilations.^[46] The data bases encompassed either three- or four-year periods.

GM's risk analysis disclosed that, in each data base surveyed, 1980 X-cars consistently exhibited a relevant accident rate no worse than, and in most instances better than, the rate for not only peer car groups but also all 1980 models.^[47] Data collected for "all" accidents, "skidding" accidents, and accidents on "wet and snowy roads," demonstrate 1980 X-cars to have a proportionately lower rate of each, and the only such situation-specific data (from the state of Michigan) show the X-cars less likely to be involved in "skidding accidents concluding in side impacts" which NHTSA submits are more likely to occur with a yawing vehicle.

The government did not challenge either the data or GM's analysis of it, but only discounted its significance on the ground that incidents of rear brake lockup are relatively rare and would likely be masked by the vastly greater number of accidents for which "driver error" is responsible. Whether or not the data are sufficiently sensitive to enable a disproportionate risk of accidents due to vehicle- (as opposed to driver-) induced skidding to be detected were it present, to the extent that statistical incidence can ever prove specific events, the risk analysis supports a finding that the 1980 X-car does not have a generic brake defect that leads to lockup or skidding. The risk analysis, in other words, is consistent with the engineering test data in tending to prove the absence, not the presence, of a "safety-related defect" in the X-car.

IX.

The gravamen of this action is that General Motors failed to comply with the notice-and-remedy provisions of § 1411 of the Act, which require a manufacturer to remedy vehicles it knows to possess a "defect" that "relates to motor vehicle safety." Counts I and II charged GM with allowing its 1980 X-cars to enter the market and remain on the road with a braking system, front and rear, it knew, "for reasons relating to several distinct components of that system," was predisposed to "premature rear wheel lock-up" with consequent loss of vehicle control. Counts III and IV alleged violations of § 1414, which requires the manufacturer to repair or replace defective vehicles, in that GM's two recalls of 1980 X-cars neither encompassed enough cars to reach all those with the defect nor remedied the defect in those they did reach. And Count VI charges GM with failing to comply with NHTSA's "hotline" regulation requiring obligatory defect recall notices to bear NHTSA's toll-free telephone number. 49 C.F.R. § 577.5(g)(1)(vii) (1981).

Relief under every count of the complaint presently before the Court is, therefore, contingent upon proof of a violation of § 1411 of the Act.^[48] Unless its 1980 X-cars possessed a safety-related "defect" within the meaning of the Act, GM had no legal obligation to repair or replace any vehicle, to conduct any notification and recall campaign, or to include any specific information in any notices it sent. And the government bears the burdens of proof and persuasion on the elements of each count. See Center for Auto Safety, Inc. v. Lewis, 685 F.2d 656, 663 (D.C.Cir.1982) ("Transmissions"); United States v. General Motors Corp., 518 F.2d 420, 438 (D.C.Cir.1975) ("Wheels").

The analytic framework to be employed by a trial court in "performance defect" cases under the Act was first set forth in Wheels, which involved a large number of broken wheels on pickup trucks.^[49] The court of appeals identified three elements of a case for compulsory recall: (1) functional "failures" in the vehicle; (2) a "significant" number of such failures; and (3) a causal relationship between the failures and the integrity and/or operation of vehicle components, rather than driver fault (at least that which is unforeseeable). In Wheels, however, it was never necessary for the court to define a functional "failure," because it was undisputed that a broken wheel was one. Id. at 426-28.

This case is apparently the first enforcement action in which a manufacturer has denied that its cars experienced functional failures. GM asserts here that neither skidding nor rear brake lockup per se constitutes a functional failure, since all cars may be expected to lock wheels under some circumstances, and the X-car has not been shown to have any peculiar propensity to lock up, rear or front, more frequently than cars generally. Moreover, it also appears that every previous case litigated to a final decision under the Act involved a clearly identified broken or separated part causally connected to an uncontroverted and significant deviation from intended and expected vehicle performance.^[50] The defendant-manufacturers relied upon various affirmative defenses rather than general denials that their vehicles were defective at all.

The government's position here is essentially as follows: (1) a failure of the vehicle simply to perform as expected is a "defect;" (2) consumer experiences alone are sufficient to prove performance failure; and (3) the government is not required to come forward with an "engineering explanation" for that failure of performance. For that formulation it cites the Wheels and Pitman Arms cases, noting that the majority in Pitman Arms declined to join Judge Leventhal in an opinion which assumed an obligation on both sides to offer technical proof for their positions. Moreover, according to the government, the comparative performance of peer cars is irrelevant (again citing Wheels). That a manufacturer has built to the "state of the art" is no defense if there are a significant number of failures to perform as expected. The only defense, according to the government, is "gross vehicle abuse" by the owner.^[51]

In Wheels, however, the court of appeals declared that it was the agency's obligation to demonstrate that failures had occurred, not merely that consumers had complained. Wheels, 518 F.2d at 427. It noted that whether "a defect exists in a particular case thus turns on the nature of the component involved," id., and required that the agency offer "competent evidence showing a significant number of failures." Id. at 442.

The Court concludes that anecdotal accounts of skidding events are not sufficiently reliable, i.e., are not competent evidence from which to infer the existence of any specific brake problem. The driver of a car simply cannot gauge its brake balance as it decelerates, even in the grossest sense. The knowledge necessary to discern, let alone calculate or quantify, brake balance or brake efficiency is lacking, even if the witness' recollection is meticulously accurate, factual not fanciful, and truthfully related. Drivers can describe only what happened to them, which is an altogether insufficient basis upon which to make a judgment as to the technical adequacy of the braking system of their cars, especially when, as here, the testimony is of skidding incidents which occurred sporadically months or years apart.^[52]

Those courts which, in other contexts, have allowed proof of vehicle defects to be made by circumstantial evidence of a loss of vehicle control have done so only with respect to a single vehicle involved in one misadventure, and have also required that other conditions analogous to those which precede the invocation of the doctrine of res ipsa loquitur be present: (1) the evidence must demonstrate an unusual event unlikely to occur with a fully functional car; (2) the occurrence must be inconsistent with causes other than vehicle malfunction, and, thus, admit of only one reasonable inference; and (3) the inference must not be contradicted by direct evidence to the contrary. See, e.g., Brothers v. General Motors Corp., 202 Mont. 477, 658 P.2d 1108 (1983); Stewart v. Ford Motor Co., 553 F.2d 130 (D.C.Cir.1977). The drivers' several descriptions of their incidents of skidding and yaw in this case satisfy, at best, only the first of these conditions, and yet the government would have the Court draw the inference not only that each and every one of them was exclusively attributable to the same systemic brake defect, but also that an entire generation of automobiles must necessarily be similarly afflicted.

The evidence here, on the other hand, conclusively establishes that skid-and-yaw can and does result from, in addition to brake imbalance, differential road friction, road camber or slope, curved paths of travel, worn, underinflated, or mismatched tires, driver steering inputs, combined braking and cornering, and lane change maneuvers, for none of which the car's braking system can be held responsible. That some of NHTSA's consumers' motoring experiences were "consistent with" rear brake lockup is no more diagnostic of a "defect" in their vehicles than are certain general physical symptoms experienced by humans diagnostic of any specific illness with which they may be consistent.

X.

In addition to proving the existence of a vehicle defect under the Act, the government must also prove that any resulting performance failure relates to motor vehicle safety, that is, it presents an unreasonable risk of accidents or injuries. 15 U.S.C. §§ 1391(1), 1411; Wheels, 518 F.2d at 426, 435. The government once again tenders the per curiam decision in Pitman Arms, 561 F.2d 923, as establishing a per se rule: any vehicle-endogenous reason for a diminution of the driver's control capabilities poses an "unreasonable" risk as a matter of law. Since rear brake lockup, when it occurs, results in at least a partial loss of control, the government contends that it satisfies the statutory requirement of safety-relatedness for a mandatory recall and repair.

Re-examination of the Wheels decision, and a review of subsequent decisions in analogous contexts under other federal safety legislation, however, persuade the Court that Wheels and Pitman Arms should not be read today as establishing a rigid rule turning entirely upon a diminution of control in the abstract. The unreasonableness of any risk to safety must be assessed relatively in at least three dimensions: (1) the severity of the harm it threatens; (2) the frequency with which that harm occurs in the threatened population relative to its incidence in the general population; and (3) the economic, social, and safety consequences of reducing the risk to a so-called "reasonable" level. See Industrial Union Department, AFL-CIO v. American Petroleum Institute, 448 U.S. 607, 100 S.Ct. 2844, 65 L.Ed.2d 1010 (1980) (plurality opinion) ("Benzene"); American Textile Manufacturers Institute, Inc. v. Donovan, 452 U.S. 490, 101 S.Ct. 2478, 69 L.Ed.2d 185 (1981) ("Cotton Dust"); Center for Auto Safety v. Peck, 751 F.2d 1336 (D.C.Cir.1985) ("Bumpers").

In Wheels, the district court had held that a large number of performance "failures" constituted irrebuttable proof of the presence of a statutory "defect" regardless of the cause of those failures. 518 F.2d at 436. The court of appeals reversed, on the ground that a manufacturer is not required to remedy even a large number of failures if their cause is a factor like age, wear, or unanticipated abuse, i.e., causes not inhering in imperfections in the vehicle as manufactured. 518 F.2d at 436. And since courts are to consider costs in addressing the question of "unreasonable risk," they are likewise to consider those same costs in addressing the question of "defect." 518 F.2d at 435. In other words, a manufacturer is not expected to build a vehicle that will never fail, no matter the cost. 518 F.2d at 435-36. Manufacturers are not obliged "to use tires that do not wear out, lights that never burn out, and brakes that do not need adjusting or relining." 518 F.2d at 436.

This implicit recognition in Wheels of a relative, rather than absolute, risk as the statutory measure of a manufacturer's duty to repair was expressly articulated in the court of appeals' recent opinion in the Bumpers case, in which the D.C. Circuit upheld NHTSA's decision to relax an impact-resistance standard for vehicle bumpers. Acknowledging that the action entailed some increase in risk to the public, the court nevertheless interpreted the Act as requiring NHTSA to regulate only as to "significant risks." 751 F.2d at 1344 n. 5, 1345, 1348. An "insignificant risk" was per se reasonable, regardless of the costs associated with the remedy:

"The principle that an ``unreasonable risk' provision requires even insignificant risks to be eliminated if that can be done at (presumably) insignificant cost would turn many areas of regulation into unending pursuit of the trivial."

Id. at 1344 n. 5. The court declared that the Act was not directed "toward any conceivable safety hazard, no matter how insignificant; rather, the Act is directed at ``unreasonable' risks." Id. at 1345 (citation omitted). In interpreting the "unreasonable risk" formulation of the Act to refer to those of "significance," the court of appeals was following the lead of the Supreme Court in Benzene and Cotton Dust in incorporating the concept of "significant risk" into an interpretation of other federal safety legislation.

Only when the risk appears "significant," based both on severity and relative frequency factors, does it become necessary to proceed to a "``common-sense' balancing of safety benefits and economic cost." Wheels, 518 F.2d at 435 (footnote omitted). In other words, a significant risk that can be remedied at a proportionate cost, and without a corresponding sacrifice of public safety in other respects, is generally to be regarded as an "unreasonable risk" which the Act mandates that the manufacturer must rectify. See Bumpers, 751 F.2d at 1344 n. 5; Wheels, 518 F.2d at 435-36. Conversely, if the only "remedies" are ineffective, prohibitively expensive, or affirmatively detrimental to public safety, even a significant risk may nevertheless be "reasonable" as a matter of law. Id.

Any skid, of course, involves some loss of control and is, thus, potentially hazardous. Assuming GM could render every X-car so front biased that its rear wheels would never lock, however, the relative severity of the risks associated with front brake lockup, rear brake lockup, and four-wheel lockup are both debatable and unresolvable on this record. The severity of the risks associated with rear brake lockup are ameliorated by the potential for shorter stopping distances (or at least lower speeds at impact) than with fronts locked, and NHTSA's prior positions in rulemaking suggest that, but for the necessity of contending otherwise in this case, it agrees that shorter stopping distances and slower speeds in collision do, in fact, reduce the risk of accidents and injury. Thus, the risk of loss of control with rear brake lockup may or may not be more severe than the consequences of a front brake lockup. The government, however, has failed to prove to this Court's satisfaction that it is.

With respect to the frequency with which the risk (of accidents and injuries) is encountered in X-cars with their brakes as presently configured, GM's risk analysis evidence demonstrates that the likelihood of involvement in a skidding accident is no higher, and as a rule is lower, for the 1980 X-car population than for the automobile population at large. If there truly is a relationship between brake bias and the frequency with which skid-related accidents are threatened, the results of the risk analysis are explained by the engineering measurements of brake efficiencies showing that the 1980 X-cars are, as a group, already less rear biased, and thus less susceptible to rear brake lockup than are most of their competitors. The evidence thus precludes a finding of "significance" in the frequency sense of the term "unreasonable risk."

Because the government has not established that the X-car braking system represents an unreasonable risk of accidents and injuries of "significance" in either the severity or frequency parameters, it is perhaps unnecessary to comment upon the safety benefits/detriments of the proposed "remedy" to determine whether it might also entail "unreasonable" risk. Benzene, 448 U.S. at 634-42, 100 S.Ct. 2859-64; Bumpers, 751 F.2d at 1344 n. 5. But the somewhat amorphous remedy NHTSA submits would be appropriate — it resists being committed to specific recommendations — has substantial safety implications of its own, unlike the remedies appropriate in prior cases under the Act, viz., to replace or repair an offending part with absolutely no corresponding negative impact upon public safety. NHTSA suggests generally that the appropriate remedy here is to assure that the X-car population is still more front biased, leaving to GM the manner in which it is to be accomplished.^[53] However, the engineering tests have shown that current configuration X-cars, at least those tested by both sides, already exhibit front bias or nearly ideal brake balance in the lightly loaded condition and are fully front biased in the heavily loaded condition. Shifting the brake balance toward more front bias would actually move the cars farther from the ideal in all loading conditions, the result being, as previously noted, to render them prone to earlier front lock, skidding incidents with longer stopping distances, and the control losses associated with front skids.

For all of the foregoing reasons, therefore, the Court concludes the government has failed to meet its burden of showing that current configuration 1980 X-cars now present, or have ever presented, an "unreasonable risk" of accidents due to a "defect" that causes "premature rear brake lockup," and Counts I and II will be dismissed with prejudice.

Counts III and IV allege that GM failed to comply with § 1414 of the Act in connection with its 1981 and 1983 recall campaigns. Count III charges that in July (i.e., August), 1981, GM "knowingly conducted an inadequate recall campaign" of some 47,000 manual transmission 1980 X-cars originally equipped with 41% proportioner valves which were to be replaced with 27% valves. Count IV makes similar allegations with respect to the February (i.e., March), 1983, recall of approximately 240,000 1980 X-cars to replace their original brake linings with the less "aggressive" 4050/4050 variety.

Implicit in the Court's conclusion that current-configuration 1980 X-cars do not possess a safety-related defect is the corollary that whatever deficiencies some earlier incarnations of them may have exhibited have been adequately remedied, and GM is under no present duty to take further action. Section 1414 of the Act provides, in pertinent part:

"(a)(1) If notification is required under section 1411 of this title or by an order under section 1412(b) of this title ... then the manufacturer ... shall cause such defect or failure to comply in such motor vehicle ... to be remedied without charge." (Emphasis added)

Recall is, thus, expressly contingent upon the existence of a safety-related defect giving rise to a duty to notify and repair.

The court of appeals previously stated in a case that followed upon completed administrative proceedings before NHTSA, "the plain meaning of this language in § 1414(a)(1) is that a determination and order under section 1412(b) are prerequisites to the remedy obligations under section 1414(a)(1).... Absent a section 1412(b) determination and order section 1414 does not apply...." Transmissions, 685 F.2d at 662. Since administrative proceedings in this case were aborted by the government in favor of an immediate resort to federal court, by analogy only knowledge on the part of GM that its 1980 X-cars were defective, and that the defect was safety-related, would have raised a duty under § 1411 to recall and repair. GM has never conceded, however, that its vehicles are defective. It acceded to NHTSA's insistent, if informal, demands that it take some action, and, over protest, voluntarily elected to conduct both the 1981 and 1983 recalls for business reasons: to avoid costly and prolonged litigation in 1981, and, following NHTSA's "initial determination" in 1983, to placate consumers aroused by the attendant adverse nationwide publicity.

Had the extensive engineering testing succeeded in isolating an idiosyncrasy in the X-car's braking system to explain the extraordinary number of consumer complaints about it, or had the accident statistics demonstrated an abnormally elevated incidence of X-car involvement in the sorts of accidents likely to occur as a result of the systemic malfunction NHTSA suspected, then the internal GM documents might supply convincing corroboration of GM's knowledge of the "defect" from the outset. As it is, without proof that there is, or ever was, a "defect," they prove only that brake engineers have yet to devise the infallible braking system, and that GM's engineers, as well as their counterparts elsewhere in the industry, continue in quest of it, and also continue to argue, sometimes heatedly, about how its imperfect substitute should work in the meantime. Counts III and IV will likewise be dismissed.

Count VI alleges that GM failed to advise owners of those 1980 X-cars recalled in July/August, 1981, of NHTSA's toll-free "Auto Safety Hotline" telephone number by which owners dissatisfied with a manufacturer's efforts to remedy a defect may notify NHTSA to that effect.^[54] The regulation, 49 C.F.R. § 577.5(g)(1)(vii) (1981), adopted by NHTSA in January, 1981, 46 Fed.Reg. 6971 (1981), required that it be done in appropriate cases, and it is undisputed that GM did not do so. The Court is unpersuaded by the reasons GM gives for not doing so, viz., that the regulation was invalidly adopted without notice and comment, and that NHTSA had never before enforced it anyway, but, as with the duty to recall and remedy itself, the duty to give notification thereof in the form prescribed by NHTSA is subject to the same precondition: that the manufacturer has or should have determined that the vehicle "contains a defect which relates to motor vehicle safety." 49 C.F.R. § 577.5(a). Having previously concluded that the proof fails to establish that the 1980 X-cars were defective, or that GM had or should have determined that they were, the Court further concludes that GM was legally a volunteer in making both recalls, and that neither notification was, therefore, required to conform to any particular form. Count VI, too, will be dismissed.

ORDERED, that judgment be entered for defendant General Motors on Counts I, II, III, IV, and VI of the complaint, and the same are dismissed with prejudice.^[55]

NOTES

[1] GM introduced the 1980 X-car for public sale in April, 1979, selling approximately 1.1 million X-cars under trade names of the Chevrolet Citation, Pontiac Phoenix, Oldsmobile Omega, and Buick Skylark during the 1980 model year ending in September, 1980.

[2] The action was filed August 3, 1983. GM moved to dismiss the complaint. On October 31, 1983, the government filed a motion for a preliminary injunction. On December 1st the Court denied GM's motion to dismiss, 574 F.Supp. 1047 (D.D.C.1983), and set hearing on the motion for a preliminary injunction for February 1, 1984. When the case was advanced for an early trial in March, 1984, the government withdrew its motion for a preliminary injunction.

Count V was severed for separate trial on February 6, 1984, on defendant's motion.

[3] Because the case was expedited to trial, discovery was allowed to continue simultaneously, being accomplished primarily during the recesses. Instead of the six weeks originally allotted, the trial eventually consumed 113 court days. Testimony was received from 33 trial witnesses, 20 of them experts. The trial record comprises over 16,000 pages of transcript and nearly 3,700 exhibits.

The Court commends counsel for both parties for their exceptional service to their clients and the Court throughout the proceedings.

[4] The manufacturer's self-start remedy provisions now found in § 1414 did not appear in the original 1966 Act but were added by the Motor Vehicle and Schoolbus Safety Amendments of 1974, Pub.L. 93-492 (Oct. 27, 1974).

Under the more familiar parallel provision of the Act, 15 U.S.C. § 1412, if the Secretary (i.e., NHTSA) determines that certain vehicles contain a defect relating to safety, then the Secretary may administratively order the manufacturer to take remedial action, and the Secretary's order is judicially enforceable under 15 U.S.C. § 1415. The Act vests the Secretary of Transportation with broad powers to conduct any investigation necessary to its enforcement. 15 U.S.C. § 1401(a)(1)(B). Vehicle manufacturers are required to maintain information, and to produce it upon request, in conjunction with an investigation. 15 U.S.C. § 1401(b). They may be asked for performance or technical data, and may be required to furnish written answers under oath. 15 U.S.C. § 1401. If an investigation develops evidence of a violation, the Secretary may refer the matter to the Attorney General who may bring an enforcement action in a United States District Court to recover civil penalties as well as to obtain appropriate injunctive relief. 15 U.S.C. §§ 1398, 1399, 1401. This Court has previously held, however, that an action brought by the United States under § 1411 may go forward in the absence of a final administrative order under § 1412. United States v. General Motors Corp., 574 F.Supp. 1047, 1049 (D.D.C.1983).

[5] This Court has also previously held that a manufacturer cannot evade its statutory obligations "by the expedient of declining ... to reach its own conclusion as to the relationship between a defect in its vehicles and ... safety." United States v. General Motors Corp., 574 F.Supp. at 1050. Thus, a manufacturer incurs its duties to notify and remedy whether it actually determined, or it should have determined, that its vehicles are defective and the defect is safety-related.

[6] The term "X-car" has no significance other than as an internal designator adopted by GM for that particular car body.

[7] GM at all times contemplated that the 1980 X-cars would be equipped with front disc brakes and rear drum brakes, a combination common then and now on both GM and non-GM automobiles.

[8] FMVSS-105 (see Part IV, infra, at 1567) establishes the government's performance requirements with respect to stopping distance in the fully and lightly-loaded conditions, fade and recovery, water recovery and parking brake grade-holding capacity. 49 C.F.R. § 571.105. The Pike's Peak test assesses the thermal capacity of a brake system by subjecting the vehicle to repeated brake applications during a steep mountain descent. The West Virginia mountain test permits an evaluation of fade, wear and general effectiveness over a 1,000-mile course through mountainous terrain. The Los Angeles city traffic test measures brake durability on a 5,000-mile urban schedule run through Los Angeles in consecutive eight-hour shifts of 200 miles each.

[9] All similar brake systems convert the kinetic energy of the vehicle into thermal energy as the brake linings press against the rotating rotor or drum mounted on the wheel.

[10] GM engineers at the time preferred fixed-slope rear proportioning to a system without proportioning (as found on some contemporary competitive cars) and to systems employing "sensing" valves that vary the percentage of rear line pressure in relation to the height of the rear springs of the vehicle. The proportioning valve was then in common use in the United States.

[11] FMVSS-105 required the parking brakes of manual transmission vehicles to hold on a 30% grade, in neutral gear, in both the forward and reverse directions. Automatic transmission vehicles needed only to hold on a 20% grade, having the additional protection of the "park" position of the transmission. 49 C.F.R. § 571.105. (With both the transmission in "park" and the parking brake engaged, automatic transmission vehicles must hold on a 30% grade.)

[12] Although the documents themselves do not say so, GM insists that it assessed all such incidents as "one-wheel takeovers," i.e., a single rear brake generating a significantly disproportionate amount of brake torque relative to the other, attributed to sustained exposure to high operating temperatures, producing changes in the chemical composition of brake linings.

[13] The DT & D and Buick engineers maintain that they never regarded the single-wheel lockup incidents as signifying a potential control, or safety, problem with the X-cars. Not only had they occurred only at low speeds towards the end of a stop, but a single rear wheel lockup does not of itself result in loss of vehicle control. The rolling tire on the opposite wheel, still operating below the limit of adhesion, generates sufficient side forces in conjunction with the front tires to maintain control of both path and attitude. See Part III, infra.

[14] The task force did, however, recommend adoption of three "running production changes" in the X-car's rear brake system, all of which evince its continuing concern with its performance. One recommendation was to substitute a nominal 27% proportioning valve for the 41% valve, resulting directly in a decrease in rear braking force at higher decelerations. The other two, ostensibly to improve the rear brake temperature behavior, were a reversion to the 4050/4050 lining combination on automatic transmission X-cars, and the use of a "finned" drum (with greater capacity to dissipate heat) in lieu of the smooth drum initially installed. The lining change was implemented in March, 1979; the 27% proportioner valve and finned drum came into production during August, 1979.

[15] The actual slope of the 27% valve is, in fact, approximately 30-31%.

[16] The "coefficient of friction" is a mathematical calculation, µ= F/N, where F is the force required to slide one object over another and N is the vertical, or "normal" force, perpendicular to it.

[17] Brakes lock the wheels when the driver applies the brakes with sufficient force to cause a tire to cease rotation, and a non-rotating wheel on a moving car is said to be "locked up." The terms "brake lockup" and "wheel lockup" are colloquialisms that are used interchangeably in this case to refer to a skid.

[18] A brake lockup condition can be relieved in either of two ways: the driver may modulate the pedal to reduce the brake torque so that the locked wheel resumes rotation, or the tire may skid on to a less slippery portion of the pavement, increasing friction forces available between the tire and road sufficiently to cause the wheel to resume rotation.

[19] Brake balance is said to be "ideal" when the distribution, or proportion, of braking forces among the tires is equal to the distribution of normal forces on the tires. A car having 40% of its normal force on each front tire and 10% on each rear tire during a stop that also develops 40% of its total braking force at each front tire and 10% at each rear tire is ideally balanced. All four tires would reach their limits of adhesion simultaneously, assuming a uniform coefficient surface, and thus enable the car to reach its full deceleration potential before experiencing wheel lock.

[20] Advances in electronic and microchip technology have made possible electronic "antilock" braking systems that have begun to appear recently on cars marketed in the United States. Such systems "sense" wheel speed independently for each wheel several times a second and "back off" line pressure if a lockup is imminent. No American-built 1980 model cars had them, and earlier efforts to apply more primitive mechanical "antilock" systems were not successful. See Paccar, Inc. v. NHTSA, 573 F.2d 632 (9th Cir.), cert. denied, 439 U.S. 862, 99 S.Ct. 184, 58 L.Ed.2d 172 (1978).

[21] The limit of adhesion, at which the maximum braking force is available at a tire/road interface, is the product of the coefficient of friction and the normal force at that interface (L=µ N).

[22] While the magnitude of their effect is disputed, variations in tire inflation pressure and tread wear are acknowledged to have at least some effect on tire/road coefficients, and a significant percentage of cars on the road have substantially worn and/or improperly inflated tires. Different tire compounds and tread patterns can also produce different coefficients. The importance of these factors in influencing the sequence of wheel lockup is apt to be greatest when the car is close to ideal brake balance.

[23] There are yet other complexities involved in trying to achieve ideal brake balance, such as the effects of speed, changes in the rolling radius of the tires, and steering input which not only reduce the friction forces available for braking but will induce as well a side-to-side inertial force transfer that may affect the sequence of wheel lockup.

[24] The adjective "premature," as it modifies "lockup," is also a colloquialism without scientific meaning, and expresses only a subjective judgment of the driver as to when he might expect his wheels — either front or rear — to lock up in any given stop. The further away a vehicle from the ideal brake balance, the greater the degree of "prematurity" perceived by the driver of the lockup of the wheels favored by the bias.

[25] Studies indicate that in perceived emergencies, to avoid a collision most drivers instinctively apply the brakes with force sufficient to lock all four wheels simultaneously, regardless of whether the car is front biased or rear biased.

[26] On a slippery surface (.5 peak µ/.3 slide µ) at an initial speed of 30 mph, an X-car sustaining front brake lockup will require 17' more (84' v. 67') to stop than with rear lock. At 55 mph, the rear lock condition improves stopping distance capability relative to front lock by 56'. Assuming a constant deceleration rate from those two initial speeds, the X-car with front lock would be traveling at 13.5 mph and 24.5 mph, respectively, at the points at which the X-car with rear lock would be completely stopped.

[27] Control of a car involves several parameters of which stability is but one. A stable system tends to resist control inputs, while instability tends to augment them. For example, a large ocean tanker is very stable underway, but, therefore, not very controllable.

[28] As the evidence vividly demonstrated, however, a moving vehicle may experience yaw with any combination of wheels locked, or as a result of the exertion of forces upon it that involve no braking at all. Rear brake lockup does not "cause," nor is it inevitably accompanied by, yaw. See Part VII, infra.

[29] It so happens that the laws of physics dictate that maximum deceleration rate in gravity units is essentially equal to the value of the peak tire/road coefficient.

[30] NHTSA insists that front biased cars, properly constructed, can pass FMVSS-105 at all loading conditions, but brake balance measurements made by both GM and NHTSA in the course of this case on several hundred competitive cars show that a substantial percentage of cars in service in America between 1977 and 1984 had some degree of rear bias and were intentionally so designed, including models manufactured by Ford, Datsun, Chrysler, Toyota, Honda, and Renault.

[31] Shortly before this trial concluded, it was reported to the Court that NHTSA had published a notice of proposed rulemaking to consider adoption of a new American braking standard, to become effective in 1991 as FMVSS-135. The proposed standard, which is still under consideration, would expressly require that 1992-model and later new cars be predominantly front biased.

[32] Although GM sold X-cars with nine separate combinations of brake components, the evidence shows that none of the configurations was immune from complaints. Even before the adverse national publicity in January, 1983, attendant upon NHTSA's initial defect determination (for which GM blames most of the outcry), GM itself had received some 700 complaints about 1980 X-cars; 269 involved automatic transmission cars, 237 of them from the population of automatics never recalled.

[33] The declarants were "randomly" selected, NHTSA said, from a sample of the 2,000-plus complaints NHTSA had in hand at the time.

[34] At the February 1, 1979, chief engineers' meeting, after production had commenced, representatives of Buick presented their three design modifications as the recommendations of the task force, i.e., use of a 350 × 27% proportioning valve, 4050/4050 rear linings, and finned rear brake drums. Whether these three changes truly represented the task force "consensus," NHTSA submits, is doubtful; it found no other task force document to corroborate Buick's characterization of it as such.

The three modifications were adopted as "running changes," but only after thousands of X-cars had been manufactured without the changes, and without retrofitting. GM began installing finned drums on production X-cars on August 15, 1979, by which time over 278,000 X-cars had been built with nonfinned drums. On August 27th, 27% proportioning valves came into use on production X-cars, nearly 294,000 X-cars having been equipped with the 41% proportioning valve in the meantime.

[35] In July, 1980, almost at the conclusion of the production of the 1980 model, GM redesigned the X-car's park brake system, allowing it to use 4050/4050 linings on even manual transmission X-cars and still be assured that they would comply with FMVSS-105. However, by July, 1980, GM had produced some 200,000-plus manual transmission X-cars with the Bendix 3198/3199 linings.

Installation of 4050/4050 linings as original equipment on automatic transmission X-cars began in March, 1979, after, however, it had produced nearly 32,000 automatics with the 4035/4050 linings.

[36] Both sides were not only required to produce to one another in advance the data underlying the test programs to be presented in court, they were expected as well to depose any expert witnesses by whom it would be presented, and those by whom it may have been collected if necessary, substantially before it was received into evidence. Consequently, they were prepared to, and did, make full use of one another's data to the extent it suited their purposes.

[37] "Current configuration" refers to X-cars equipped with 8032 front brake linings, 4050/4050 rear linings, with either 41% or 27% proportioner valves, and either smooth or finned rear brake drums, i.e., configurations remaining on the road post-recalls.

[38] Altogether, 302 "as received" X-cars (including some equipped with pre-recall rear linings and others with "aftermarket" linings from non-GM sources) and 528 "peer" cars were tested in some fashion by the parties. The "as received" measurements reflected the cars' brake balances as of the time they were acquired from the owners for testing, i.e., with such changes in brake balance from design intent as may have occurred while the cars were in customer service for any reason.

[39] Shortly before trial NHTSA evaluated the wheel lock sequence on 102 1980 X-cars and 41 1979-81 model peer cars with original equipment brakes. The tests were conducted on three road surfaces using electronic decelerometers, pedal force transducers, and outside observers to determine wheel lock sequence. On high traction asphalt roads, 35% of current configuration X-cars locked rear wheels first compared to 92% of the peer cars tested by NHTSA. On the lowest coefficient road, the figures were 25% and 58%, respectively. Moreover, the pedal force transducers indicated that the pedal effort needed to lock the second axle after the first axle had locked on the rear biased X-cars was less than that needed to lock the peer cars, i.e., the X-cars were closer to ideal brake balance.

[40] The three samples encompassed the initial 57 1975-84 state-of-the-art cars, all from the Arizona area, tested by GM without pre-screening, the 101 statistically random 1979-81 cars, from Michigan, also tested by GM, and the 41 1979-81 cars from Ohio tested by NHTSA without pre-screening.

[41] GM engineers devised graphical plots, called "cloud charts," to display the distribution of comparative braking efficiencies for the 103 current configuration X-cars and the several fleets of competitive cars tested. In each instance, a larger proportion of X-cars than peer cars appear therefrom to be front biased, and the most rear-biased X-cars were more efficient than many of the competitive cars.

[42] GM tested 32 X-cars with 4050/4050 rear linings in the rebuilt and burnished condition and all were front biased at all deceleration ranges except one which became slightly rear biased around .7g. NHTSA tested four such current configuration X-cars; all four were front biased in the lightly loaded condition. Smaller samples of competitive models were similarly rebuilt and burnished with new production linings; three of them proved to be rear biased in the lightly loaded condition at all decelerations.

[43] The evidence as to the effect of corrosion of the front brakes on brake balance being negligible, in mid-trial NHTSA shifted focus to certain properties of the 8032 semi-metallic lining material that might cause brake balance to creep rearward with use.

Two types of variation in 8032 output were observed by NHTSA engineers: the output is lowest when the brakes are cold, a characteristic of semi-metallic linings generally; and, over an extended journey with minimal braking, the sustained output can decline by a factor of nearly a third. The magnitude of the reduction in output under such conditions is, however, comparable to the decreases in output experienced with other linings and in other cars.

Vehicle tests of 12 competitive cars and five X-cars, tested with torque wheels, as received and then rebuilt and burnished, showed that competitive cars experienced in-use front output reductions equal to or greater than those of the X-cars. And even those current configuration X-cars with the lowest measured front specific torques were found nevertheless to be front biased.

[44] Similar "worst-case" projections were readily devised by GM witnesses for hypothetical competitive cars which, were they to exist, would be even more rear biased than the most flagrantly imbalanced "worst-case" X-car.

[45] Near the conclusion of the presentation of the last of the engineering evidence, NHTSA's Associate Administrator for Enforcement conceded that NHTSA was unable to offer an "engineering explanation" for the X-cars' elevated complaint rate.

[46] NHTSA's National Center for Statistics and Analysis maintains two automotive data bases: the Fatal Accident Reporting System ("FARS"), and the National Accident Sampling System ("NASS"). FARS is a census of all fatal motor vehicle accidents on public roads in the United States. NASS is a scientifically chosen sample of motor vehicle accidents which NHTSA investigates in depth for causal factors. Various states also maintain accident data bases extracted from state police reports, including Alabama, Georgia, South Carolina, Pennsylvania, Texas, Washington, Michigan, Maryland, New York and Idaho.

[47] Comparison of 1980 X-car accident rates was made to those of all 1980 model cars, a "peer group" of 20 1979-81 models matched to X-cars according to size and cost, and an "alternate peer group" of nine models from 1979-81 matched in driver demographics and environmental usage. The Ford Fairmont was the most closely matched peer car.

[48] As previously noted, supra note 2, Count V was severed for separate trial and all proceedings thereon stayed.

[49] The court of appeals reversed a summary judgment for the government, holding that even where a significant number of functional failures was admitted, it would still be open to the manufacturer to establish that the failures were attributable to unforeseeable misuse of the product or other owner fault rather than a product deficiency. Wheels, 518 F.2d at 427, 438. The court also held that for components whose useful life is less than that of the vehicle as a whole, the government would still retain the burden of negating "causes like age and expected wear and tear" for the functional failures. Id. at 438. It did not address the situation in which extrinsic causes, i.e., those without the control of either the manufacturer or owner/driver, were as likely as mechanical malfunction to be responsible for less-than-optimum vehicle performance, although it did make clear that substandard performance alone would not suffice to support a finding of the existence of a "defect." Id. at 437-38.

[50] In Wheels, the manufacturer admitted that broken wheels were "failures" but contested whether the breakage was due to vehicle or driver fault. In United States v. General Motors Corp., 561 F.2d 923, 924 (D.C.Cir.1977) ("Pitman Arms"), cert. denied, 434 U.S. 1033, 98 S.Ct. 765, 54 L.Ed.2d 780 (1978), the manufacturer disputed whether broken pitman arm steering failures caused an unreasonable risk of accidents. In United States v. General Motors Corp., 565 F.2d 754 (D.C.Cir.1977) ("Carburetors"), the manufacturer disputed whether the number of engine fires which could be anticipated as a result of dislodged carburetor fuel plugs was significant and an unreasonable risk; it conceded that dislodged carburetor plugs were failures. In United States v. Ford Motor Co., 421 F.Supp. 1239, 1242 (D.D.C.1976), ("Seatbacks") affirmed in part, appeal dismissed in part as moot, 574 F.2d 534 (D.C.Cir.1978), the manufacturer made a similar argument with respect to collapsed seatback brackets. And in United States v. Ford Motor Co., 453 F.Supp. 1240 (D.D. C.1978) ("Wipers"), the manufacturer admitted that detachments of windshield wipers from their pivot assemblies were failures, but contested whether their frequency of occurrence was significant and whether it presented an unreasonable risk.

Compare Center for Auto Safety, Inc. v. Lewis, 685 F.2d 656 (D.C.Cir.1982) ("Transmissions"), in which the court of appeals approved settlement of a case in which the manufacturer denied the defect and NHTSA was unable to establish a vehicle-oriented reason to explain why automatic transmissions would occasionally and unexpectedly slip from "park" to "reverse."

[51] The government appears to have begun this case on the surmise that GM had, intentionally or inadvertently, designed the X-car to be rear biased. When it became clear that GM's design philosophy was one of ideal balance, NHTSA proceeded to examine the various brake system components and their interaction to determine how they might cause X-cars to become rear biased, intermittently or permanently, over time. Then, confronted with the evidence that the X-car does not, generically, either originate with or develop more rear bias than its competitors, near the close of the case NHTSA was obliged to acknowledge that it was without an "engineering explanation" for the erratic performance reported to it by the complainant-consumers.

The closest the government comes to a description of the etiology of the phenomenon described by consumers is that X-cars (or some of them, at least) must become significantly rear-biased "in service" on the road, a progression which may be attributable in unspecified proportions to: (1) a decline in front brake effectiveness over time; (2) increase in rear brake effectiveness over time; (3) parking brake "drag;" and (4) the use of a duo servo brake mechanism on the rear axle. It has also, over the course of the case, at various times suggested that the proportioning valve, "aggressive" rear brake linings, non-finned rear brake drums, the power booster system, and front brake pin corrosion are implicated.

Ultimately, however, the government insists that it is not obliged to explain the consumers' mishaps; it is enough if a sufficient number of them actually happened.

[52] Numerous courts have expressed doubt that skidding alone is probative of a defective vehicle. See, e.g., Zidek v. General Motors Corp., 66 Ill.App.3d 982, 23 Ill.Dec. 715, 384 N.E.2d 509 (1978); Woods v. General Motors Corp., 423 So.2d 112 (La.Ct.App.1982); Provence v. Williams, 62 Tenn.App. 371, 462 S.W.2d 885 (Ct.App.1970).

[53] In closing argument government counsel suggested that retro-fitting 1980 X-cars with a new generation of GM semi-metallic front brake linings, known as 8034/8035's, might be appropriate.

[54] GM's February/March, 1983, recall notification did include the "hotline" number.

[55] Despite GM's belief that NHTSA acted with less than worthy motives, the Court finds that the United States was justified in commencing this action as precipitously as it did. While the true reasons for the X-car's unusually high prepublicity complaint rate may never be known, the prospect of leaving an entire generation of unpredictably uncontrollable mass-produced automobiles on the nation's highways while NHTSA worked to determine the cause was sufficiently alarming to induce it to abandon the administrative proceedings in favor of an immediate lawsuit. Nevertheless, the consequence was to propel an archetypal case for agency adjudication to court without benefit of a fully developed administrative record, and hasten it to an early trial before discovery could refine the issues. The government came into court with nothing more, essentially, than a reasonable suspicion, without the evidence to prove it. Perhaps it expected GM to capitulate once more, and, if not, sooner or later the evidence confirming its suspicion would materialize. Neither happened, and this decision is the result. Whether time has actually been saved, or the disposition achieved the proper one, are debatable.