A jury found truck driver Rodney Martin and the moving firm for which he worked, V.I.P. Transportation, Inc., liable for the injuries Joseph Roback and Wendy Rizzo suffered when Martin's truck rear-ended the automobile in which they were driving. Martin and VIP had filed a third-party complaint against Chicago Kenworth, Inc., Paccar, Inc., and AlliedSignal, Inc. based on a purported defect in the cruise control system of Martin's truck. That defect, according to the defendants, contributed to the collision by distracting Martin's attention from the roadway until it was too late for him to stop the truck and avoid colliding with the car in front of him. The district court entered summary judgment in favor of these third-party defendants prior to trial, however. Martin and VIP appeal both the grant of summary judgment and the jury's modest award of punitive damages. They contend that the district court overlooked factual questions regarding the extent to which Martin's negligence alone caused the accident and that it erred in excluding evidence from two experts that would have bolstered their claim that the cruise control defect was a proximate cause of the collision and the plaintiffs' injuries.1 We affirm the grant of summary judgment but vacate the award of punitive damages.

At approximately 8:00 p.m. on July 2, 1991, Roback and Rizzo were on their way to a concert at the World Music Theater in Tinley Park, Illinois along with Paul Horjes and Melissa Johnson. Roback was driving and had brought the car in which they were riding to a complete stop in the far right line of Interstate 80 westbound, waiting to exit the highway onto Harlem Avenue southbound. Concert traffic was so dense that Roback's car was at that point last in a line of vehicles a mile and a half long waiting to exit. Rizzo, who was sitting in the back seat of Roback's car along with Johnson, leaned forward between the front seats asking that the radio be turned up so that she could enjoy a favorite song. Within seconds, Martin's tractor-trailer crashed into the rear of the car, killing Johnson.

Although Horjes emerged from the vehicle relatively unscathed, both Roback and Rizzo suffered significant injuries. The bone structure around one of Roback's eyes was fractured and had to be reconstructed surgically. Rizzo's liver was lacerated and also had to be repaired in surgery. They each suffered a variety of other lacerations, and although both have recovered well physically, each bears visible scars and Roback, who suffered a moderate hearing loss in one ear, continues to experience tinnitus, a constant ringing in that ear.

Martin has been an over-the-road truck driver since 1979. Martin owns his truck and is thus self-employed. Since 1988, he has contracted his services to VIP, which does business under the more commonly known name of Allied Van Lines. VIP leases Martin's truck and pays him to drive it. VIP owned the trailer that Martin was pulling at the time of the accident.

In late 1988, Martin purchased a Model T600A truck from Kenworth and took delivery of that vehicle in early 1989. Martin paid approximately $100,000 for the new vehicle. The vehicle had an expected life of over one million miles; at the time of the accident, it had accumulated 250,000 miles. One of the options that had been installed on the vehicle at Martin's request was a cruise control system. That system was manufactured by AlliedSignal.

Martin had begun to experience problems with the cruise control and speedometer of his truck in May of 1991. While Martin was driving in northern California with the cruise control engaged and set at 55 miles per hour, suddenly "the truck just went haywire" (Tr. 362): the digital speedometer began to display rapidly fluctuating speeds of between 30 and 50 miles per hour, while at the same time the engine of the truck (which was normally quite quiet) alternately revved to very high RPMs, as if the truck were accelerating dramatically, and then backed off consistent with deceleration. This caused the truck to vibrate in an alarming fashion. Ultimately, Martin disengaged the cruise control by flipping the appropriate toggle switch on the dashboard of the cab. The erratic behavior of the truck ceased immediately. On the following day, Martin took the truck into a Kenworth dealership in San Leandro, California and reported the problem. A sensor unit on the vehicle was replaced. Martin then drove his truck across country to the east coast. While driving through the Southwestern desert, Martin attempted to engage the cruise control and on several occasions the truck displayed the same erratic behavior that he had experienced in California. When Martin returned to California from the east coast a short time later, he telephoned the San Leandro dealership to inform them that he was still experiencing difficulty with the truck, but that dealership was unable to work him into its repair schedule. When Martin returned to his Florida home in June, he took the truck to another Kenworth dealership in Orlando and described the recurrent malfunction. The sensor was once again replaced, and Martin was credited for the apparently faulty sensor installed in San Leandro. Shortly after the repairs, Martin left Orlando and drove to Greensboro, North Carolina. En route, he engaged the cruise control and, again, the same phenomenon occurred with the wildly fluctuating speedometer and racing engine. Martin turned the cruise control off. From Greensboro, Martin proceeded to Hanover, Maryland, where he spent the night. On the following day, July 1, he left Hanover bound for Chicago. He spent the evening of July 1 at a rest stop outside Delta, Ohio, near the Indiana border. (Martin's truck was equipped with a sleeper berth.)

At 3:00 a.m. on July 2, Martin began the final leg of his journey to Chicago. Rain and an abundance of state troopers counseled in favor of keeping the truck at a speed of 55, so Martin decided to engage the cruise control. After a stop at the Indiana toll booth, the system malfunctioned yet again. This time, however, Martin noticed that in addition to the speedometer behaving erratically, the accelerator pedal dropped to the floor and the truck steadily began to increase its velocity beyond the set speed of 55 miles per hour. Frightened by the runaway acceleration, Martin switched the system off and the trouble ceased. Martin continued the trip without incident and arrived in Chicago at the Pickens-Kane storage facility at 6:45 a.m. After unloading the trailer and having breakfast with friends, Martin drove his truck to the Chicago Kenworth dealer where he had purchased the truck.

Martin described the problem to the service manager at the Kenworth dealer and then waited for the rest of the day while his truck was examined and repaired. The repair order by the service writer at Chicago Kenworth, read as follows:

Customer complaint: Speedo will work erratically. Customer has had sender replaced and speedo replaced. Was told rear yoke on trans. may be causing the problem. Works fine after trans. is warmed up. Gets D152 code on dash which indicates a bad sending unit.

Martin Dep., Plaintiffs' Ex. 18. The first mechanic who worked on Martin's truck found that the sensor was working, but he also discovered that a bolt holding together the two halves of a 16-way plug located at the firewall of the truck had been stripped, resulting in a loose connection in the wiring. He tightened the connection before he went off duty at 4:00 p.m. He instructed the mechanic coming on duty after him to drill out and replace the stripped bolt. That mechanic did so and then checked the diagnostic computer in Martin's truck to see whether it would generate an error code. It did not, which apparently satisfied the mechanic that any problem had been corrected. He did not, however, take the truck for a test drive as he conceded he would have had the speedometer itself been repaired. Nor did he examine the rear yoke referenced in the repair order. The rear yoke is a part which connects the transmission of the truck to the drive line. The first mechanic conceded at trial that a loose yoke might result in erratic and false signals to the speedometer and cruise control system. The repairs that were made were summarized on an itemized bill provided to Martin. Martin paid the bill, hooked his truck up to the VIP trailer, performed a pre-trip walk-around inspection of the truck, and set out for the Illinois-Iowa state line, where he planned to spend the night.

Martin left the Chicago Kenworth facility at approximately 7:45 p.m. on July 2. He entered Interstate 294 southbound and immediately thereafter exited onto Interstate 80 westbound. Interstate 80 westbound has three lanes at that point, and Martin kept his truck in the far right lane. Traffic was light at that time, night had not yet fallen, visibility was good, and the pavement was dry. Shortly after he entered Interstate 80, Martin engaged the cruise control and set it at 55 miles per hour. Although Martin was aware that there was traffic in front of him, he did not notice that any of it had stopped, even as he approached mile marker 150, where Roback's vehicle was last in the long line of cars waiting to exit the Interstate at Harlem Avenue.

As he approached mile marker 150 (which was just a few miles beyond Interstate 80's intersection with Interstate 294), the cruise control and speedometer on the truck malfunctioned once again. The speedometer displayed wildly fluctuating speeds, the engine raced, and the cab began to shake. The accelerator pedal dropped to the floor, and Martin felt the truck begin to accelerate. He looked at the dashboard to locate the toggle switch to deactivate the cruise control. He attempted to turn the switch off. When he looked up, he saw Roback's vehicle directly in front of him. The vehicle was barely visible over the hood of his truck, which meant that Martin was already perilously close to the car. Martin slammed on the brakes immediately, but it was too late to avoid a collision. The tires of his truck left more than 170 feet of skid marks on the highway.

A jury found Martin and VIP negligent and awarded compensatory damages of $371,097.10 to Roback and $317,090.85 to Wendy Rizzo. It also ordered Martin to pay punitive damages of $171.71 to the plaintiffs.

Martin and VIP challenge two aspects of the proceedings below: the district court's grant of summary judgment to the third-party defendants, and the jury's award of punitive damages. Martin contends that factual questions concerning the degree to which his own negligent failure to monitor the traffic in front of him caused the accident and the injuries to the plaintiffs render the district court's grant of summary judgment improper. He challenges the punitive damages award based on the district court's decision to exclude from the trial the proffered testimony of two expert witnesses, William Rosenbluth and Paul Olson.

A. Summary Judgment for Third-Party Defendants

Martin and VIP sought to hold Paccar and AlliedSignal (the manufacturers of the truck and the cruise control, respectively) liable for purported defect in the cruise control system and Chicago Kenworth liable for its failure to correct the problem just prior to the accident. These third-party defendants in turn sought summary judgment arguing, among other things, that Martin's own negligence was the sole cause of the collision with Roback's car. Their argument stemmed from the following exchange that occurred during Martin's deposition:

Q. Based on what you now know upon reflection, do you think that if you had touched the brake pedal when you first noticed the erratic operation of the cruise control, you could have brought the vehicle to a stop and avoided this accident?

A. No, sir, I do not.

Q. Why not?

A. It happened so fast I don't believe I could have.

Martin Dep. 432. As he later did at trial, Martin explained at his deposition that when he noticed the engine begin to surge in speed, he looked at the dashboard to locate the toggle switch to disengage the cruise control; immediately thereafter, when his eyes returned to the road, he saw the Roback vehicle in front of him and applied the brakes. Martin Dep. 192-93, 196, 298-300, 374-75. Although Martin customarily used the toggle switch on the dashboard to disable the cruise control when it malfunctioned,2 he also could have touched the brake pedal to disengage the cruise control; more importantly, he did not have to turn the cruise control system off in order to apply the brakes and slow the vehicle down. Seizing on Martin's admission that he could not have avoided the collision even if he had begun to apply the brakes as soon as he noticed the malfunction recur, the third-party defendants argued that Martin's own negligent failure to monitor traffic and to maintain a safe distance between his vehicle and those in front of him was the sole cause of the accident.

Judge Holderman agreed that on the record before him, the third-party defendants were entitled to summary judgment:

There is no evidence in the record indicating that Martin's assessment is inaccurate. No other witness has Martin's unique knowledge of the crash and what led to it. There is no objective data regarding stopping distances nor any schematic maps contradicting Martin's admission that had he touched the brake pedal when he first noticed the erratic operation of the cruise control, he could not have brought the vehicle to a stop and avoided the accident. Further, Martin's statement ... that he could not have stopped if he applied the brakes immediately was not based on an unrefreshed recollection of the accident. His opinion was apparently based on a subsequent visit to the site on a trip after the accident. Moreover, it is not disputed that Martin was a driver of considerable experience that had logged several thousand miles on the truck involved in the accident. He was doubtlessly familiar with the truck's handling characteristics.

[T]here is some ambiguity in Martin's admission, given the lack of specifics and other hard data. Based on the record before the court, however, there is no competent evidence to refute Martin's assessment regarding his inability to stop the truck in time. Therefore, Martin's negligence has been shown by the undisputed material facts to be the only proximate cause of the accident.

July 6, 1994 Mem. Op. at 12-13, 1994 WL 327414 (record citations omitted).3

Martin initially argues that the remark relied upon by the court to grant summary judgment was isolated and ambiguous, and does not unequivocally support the view that any defect in the cruise control and the speedometer did not contribute to the accident. It is true that this exchange was isolated in the sense that the discussions immediately preceding and following it concerned other topics, and no one at the deposition ever asked follow-up questions that might have confirmed or clarified Martin's response. Judge Holderman himself pointed this out. July 6, 1994 Mem. Op. at 5. Yet, it is not as if the question put to Martin were unclear or Martin's statement itself an aside; he was asked a direct question and he gave a direct answer. Martin himself cites no other testimony contradicting or modifying his answer on this occasion. The isolated nature of the exchange does not, therefore, render his answer immaterial.

Nor do we find Martin's answer to be inherently ambiguous. The defendants suggest that it may have taken a moment for Martin to realize that the surge in the engine noise was a recurrence of the cruise control/speedometer malfunction. Thus, when Martin said that he could not have stopped the truck even if he had applied the brakes when he first noticed the problem, he may have meant that it was too late to do so by the time he realized the cruise control was malfunctioning; in other words, the time it took to interpret the sounds he was hearing and the erratic speedometer display as a malfunction, however brief, may have deprived him of the time and distance he needed to bring the truck to a safe stop. (Defendant's Br. at 43-45.) Yet, the defendants cite nothing in the summary judgment record that supports this interpretation of Martin's seemingly unequivocal admission.4 Nor, upon review of the summary judgment briefing below, do we find that the defendants urged such an interpretation, despite the fact that Martin's statement was the basis for the third-party defendants' argument that Martin's own negligence was the sole cause of the accident. The defendants did argue that Martin's statement had been taken out of context and was inconsistent with the record as a whole (R. 195 para. 37), as Judge Holderman acknowledged (March 29, 1994 Mem. Op. at 4 n. 3, 1994 WL 113105), but they never proffered the construction that they do now with any clarity. Consequently, we deem any argument based on a contrary interpretation of Martin's statement to have been waived.

The defendants also argue that even if a defect in the cruise control system and/or the speedometer was not a proximate cause of the accident, it may have been a proximate cause of the plaintiffs' injuries. Their theory is that had Martin's attention not been diverted from the roadway by the defect, Martin might have begun to apply the brakes sooner than he did so that even if the collision would still have occurred, it would have occurred with less force and the plaintiffs' injuries would therefore have been less serious. Yet, as the third-party defendants have pointed out, this was not argued below either. The defendants concede that they did not go into any detail on this alternative argument below, but suggest that it is an elementary point encompassed within the arguments that they did make. It is true that here and there in the summary judgment briefing below, the defendants adverted to their burden to show that the product defect and negligence attributed to the third-party defendants resulted in "injury" or "damage." E.g., R. 194 at 3, 6; R. 196 at 9. But nowhere did Martin and VIP suggest that the conduct of the third-party defendants might have been a proximate cause of the injuries Roback and Rizzo suffered, if not a cause of the accident itself. The third-party defendants themselves drew no distinction between the collision and the plaintiffs' injuries; they argued unequivocally that any difficulties with the cruise control and speedometer did not cause the accident and that they were therefore off the hook entirely. At that juncture it was the defendants' duty to articulate the alternate theory that they propose now, however obvious they may believe it to have been. It certainly was not Judge Holderman's responsibility to divine it. We deem the argument waived.

B. Punitive Damages: The Exclusion of Defendants' Experts

As we have noted, the district court excluded the testimony of two experts tendered by the defense, William Rosenbluth and Paul Olson. Judge Holderman initially barred them on the plaintiffs' written motion in limine; he later reaffirmed his ruling and denied the defendants' motion to reconsider after hearing both witnesses testify at a pretrial hearing. Martin contends that the exclusion was erroneous and that it deprived him of a fair trial insofar as the award of punitive damages is concerned. The plaintiffs, we note, have not appeared in this appeal to defend the punitive damages award. We should also point out that although the award is small, it has important ramifications for the defendants in terms of insurance coverage.

William Rosenbluth is an engineer who uses a computerized data acquisition system that he refers to as the DATAQ to gather data on the performance of various systems within an automobile or truck. Using the DATAQ, Rosenbluth can, for example, monitor a vehicle's wheel speed, how far the accelerator is depressed, engine RPMS, pounds of air or force in the brake system, and so on. He can then use that data to illustrate how a particular system performed while the vehicle was in operation and thus identify abnormalities. He claims to have investigated some 480 vehicle systems, using the DATAQ to do so on 75 to 80 occasions. He was retained by the defense in this case to determine and report whether there was any demonstrable malfunction of the vehicle control systems in Martin's truck, specifically and principally in the cruise control system.

Essentially, Rosenbluth took Martin's truck for a 90-mile drive and used the DATAQ to document how the vehicle performed. He paid particular attention to the engine throttle, the position of the accelerator pedal, the acceleration of the cab of the truck, the wheel speed, the output frequencies of the cruise control and speedometer sender units, and the RPMs of the engine. He also used a video camera to record the display of the electronic speedometer during the test. During the drive, Rosenbluth had the operator of the truck engage the cruise control on several occasions. The data gathered during the drive revealed erratic signals from the speed sensor of the truck. The result in some instances was that the cruise control, based on faulty data suggesting that the truck was either slowing down or speeding up too much, opened and closed the engine throttle even as the actual truck speed remained more or less stable. At the same time, as the videotape revealed, the speedometer reflected dramatic fluctuations in speed of between 36 and 50 miles per hour over impossibly short periods of time (three seconds, for example). In other instances, the actual speed of the truck would fall off, and then the cruise control would violently but briefly attempt to correct the fall-off by opening the throttle to its maximum setting. At no time during the testing did Rosenbluth observe the truck actually exceed the set speed limit to a significant degree. Also, the cruise control always disengaged when the driver touched the brake pedal. Rosenbluth offered no opinion as to the specific cause of the malfunction he observed. His purpose was merely to confirm that there was a malfunction of the kind that Martin had described.

Paul Olson is a part-time consultant, principally as to issues concerning the visual perception of automobile drivers. He formerly worked for the Transportation Research Institute at University of Michigan. According to Olson, people perceive relative velocities poorly. Their major cue in assessing relative velocities is a change in image size. As a driver approaches the car in front of him, for example, it appears progressively larger to him and the driver thus realizes that his car is traveling at a higher rate of speed than the other. At greater distances, however, a driver is unable to perceive the change in image size and thus cannot determine as readily how quickly he may be gaining on the vehicle in front of him; this only becomes apparent as the vehicles come closer together. Distraction of the driver compounds the difficulty of perceiving relative velocities, as the driver's focus is diverted from the visual cues by which he can assess his own speed in relation to that of other vehicles.

Having reviewed Martin's deposition testimony regarding the accident, Olson estimated that Martin was distracted for not less than about three seconds. Normal visual interrogation of the dashboard would take about one second, by Olson's account; interrogation in the context of a problem would logically take longer. The normal time it takes a driver to react when danger is perceived is also about one second. Assuming that Martin's truck was traveling at 60 miles per hour (88 feet per second), a three-second distraction would have cost him about 267 feet in which to stop the truck. In Olson's opinion, Martin almost certainly could have stopped the truck short of a collision if he had not distracted by the cruise control/speedometer malfunction.

Both Rosenbluth and Olson were proffered as expert witnesses. Expert testimony is available when "scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue." FED. R. EVID. 702. The decision whether to permit such testimony is committed to the discretion of the district judge. E.g., United States v. Anderson, 61 F.3d 1290, 1297 (7th Cir.), cert. denied, --- U.S. ----, 116 S. Ct. 543, 133 L. Ed. 2d 446 (1995); Taylor v. Illinois Cent. R.R. Co., 8 F.3d 584, 585 (7th Cir.1993).

We find no abuse of discretion in the decision to exclude Olson. Arguably his testimony might have been helpful to explain how drivers perceive relative velocities, a task most of us perform by habit without thinking about how we do it. But any juror who drives would readily have understood without the benefit of Olson's testimony that a driver needs to watch a vehicle in front of her in order to assess how quickly she is approaching it, and that if the driver is distracted she may not realize until too late that she needs to slow down. See Taylor, 8 F.3d at 585-86. Olson's additional opinions as to how long Martin's eyes likely were diverted from the road and how much reaction time this cost him border on speculation.

The exclusion of Rosenbluth is another matter. Plaintiffs challenged his testimony under the framework of Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S. Ct. 2786, 125 L. Ed. 2d 469 (1993), arguing principally that it lacked any scientific basis. Judge Holderman agreed, pointing out that Rosenbluth was not a professional engineer, that he could not identify the source of the malfunction that he observed, and that the DATAQ that Rosenbluth used in his analysis had never been subjected to meaningful peer review. We reject the notion that Rosenbluth's analysis was not sufficiently "scientific" to satisfy the dictates of Rule 702 and Daubert, however. Documenting the malfunction of a vehicle by gathering and compiling data during a test run is hardly a novel methodology. In a basic sense, Rosenbluth was no different than an eyewitness who may have observed Martin's truck malfunction on other occasions. Arguably, however, his testimony would have been more reliable because his observations were quantified. The only thing apparently unique to Rosenbluth's approach was the DATAQ, in the sense that he put together the hardware and designed the software, and (with the exception of a doctoral student) only he had ever used them. But Rosenbluth used standard components to assemble the DATAQ, and he certainly could have been interrogated about the way in which his software worked. His data were subject to examination and independent verification. We see no way in which Rosenbluth's testimony did not qualify for admission under Rule 702.

We also think the exclusion of Rosenbluth's testimony prejudicial to Martin. Judge Holderman, in addition to finding Rosenbluth's testimony insufficiently grounded in science, believed Rosenbluth's testimony to be irrelevant. Based on the pre-trial briefing, he expected no dispute that Martin was looking down at the dashboard just prior to collision; why Martin may have been doing so was immaterial, in the judge's view. He noted, for example, that the malfunction had occurred on several occasions prior to the date of the accident; thus, Martin could not claim to have been startled by the recurrence. Sept. 30, 1994 Mem. Op. at 12-13. Even so, Martin argues that Rosenbluth's testimony was admissible to confirm Martin's description of the malfunction, and we agree. As the district judge pointed out in granting the third-party defendants' motion for summary judgment, "[n]o other witness has Martin's unique knowledge of the crash and what led to it." July 6, 1994 Mem. Op. at 12. Martin's explanation for his failure to notice Roback's vehicle until it was too late to avoid a collision, and the credibility of that explanation, were key issues. As a defendant charged with liability for a horrible accident, one might readily believe that Martin had a motive to fabricate. It is true that, for the majority of the trial, the plaintiffs did not appear to quarrel with the notion that the truck had malfunctioned in the way Martin described. Nevertheless, during closing arguments, the plaintiffs' attorney suggested in no uncertain terms that Martin had made the whole thing up. Tr. 898 ("The cruise control, I submit, is the problem that arose after the accident ....") (emphasis supplied). At the same time, as reflected in the jury instructions, Martin's failure to keep his eyes on the road was one of the circumstances that the plaintiffs claimed was evidence of "willful and wanton conduct," entitling them to punitive damages. Tr. 941-42. "Willful and wanton conduct" was defined for the jury as "a course of action which shows an utter indifference to or conscious disregard for the safety of others." Tr. 938. We think it obvious that the reason for Martin's distraction had a significant bearing upon whether or not Martin's failure to monitor the traffic in front of him was willful or wanton. Given that Martin's credibility on this subject had been placed in question, the exclusion of Rosenbluth's testimony was particularly prejudicial.

The plaintiffs did point to another ground on which Martin's conduct might be deemed willful and wanton. A theme that was repeated throughout the closing argument in particular was that Martin had continued to drive his truck on the public highways with reckless abandon, knowing that it was defective. See, e.g., Tr. 721. As plaintiffs' counsel put it:

[T]hink what we have here. We have a man who chooses to put everyone's life in danger by driving a 42,000 pound battering ram down the highway. That's what we have. And that, I would submit, is reckless disregard for everybody but him.

* * * * * *

[Y]ou may consider sending a message not just to Mr. Martin, but to all those truck drivers on the road who choose to haul down the road thinking only of themselves and driving rigs that aren't safe, and that's called punitive damages, not just to punish Mr. Martin, but it has the more important effect of saying to the public in general and to the society we live in, "Nobody gets on the road and does this. Nobody."

If your brakes don't work, if your cruise control doesn't work, if your rear view mirror doesn't work, you don't take this vehicle and kill people, and maim people. You don't do that.

Tr. 898, 903-04. Now, after reviewing the trial transcript, we have found very little evidence to support the notion that Martin took what he knew to be a defective, "42,000 pound battering ram" on the road. As our summary of the facts reveals, Martin had repeatedly attempted to have the problem with the cruise control and speedometer fixed. Repairs were made on each occasion that Kenworth mechanics apparently believed were adequate. Although in retrospect it appears that the repairs did not address the problem, we are given no reason to believe that Martin knew this in advance or was indifferent to the efficacy of the repair work. Thus, so far as the evidence reveals, when Martin left the Chicago Kenworth facility on July 2, he believed, and reasonably so, that the situation had been resolved by replacing the stripped bolt and tightening the loose electrical connection (a problem that had not been discovered or addressed previously). The subsequent malfunction of the cruise control and speedometer immediately in advance of the collision, which occurred literally within minutes of Martin's departure from the Kenworth facility, was the first indication that these repairs were, in fact, inadequate.

Our point is not that there was no basis whatsoever for punitive damages in this case. It is simply that the exclusion of Rosenbluth's testimony substantially prejudiced the defendants as to one of the grounds asserted for punitive damages, and we cannot be sure that the jury would have awarded punitives on another ground, particularly in light of the evidentiary weakness we have cited. The award of punitive damages must be vacated and the case remanded for a new trial on Count II of the plaintiffs' complaint, should they remain interested in seeking punitive damages.

For the reasons discussed above, the grant of summary judgment in favor of the third-party defendants is affirmed. The award of punitive damages against defendant Martin is vacated and the case remanded for a new trial on Count II of the plaintiffs' complaint. The parties shall bear their own costs on appeal.

AFFIRMED IN PART, VACATED IN PART, and REMANDED.