Full opinion text
FINDINGS OF FACT AND CONCLUSIONS OF LAW PHILIP PRATT, District Judge. This action arises from the performance of a contract between plaintiff Eaton Corporation (“Eaton”) and defendants Magnavox Company and Magnavox Government and Industrial Electronics Company. Eaton has filed a six-count complaint in connection with the performance of this contract. Eaton claims that Magnavox’s conduct constituted: 1. A breach of contract; 2. A breach of express warranty; 3. A breach of implied warranties; 4. Misrepresentation; 5. Negligence; and 6. The manufacture of a defective product. Eaton originally sought both damages and declaratory relief. In a memorandum opinion and order issued on October 29, 1980, this Court denied without prejudice Eaton’s claims for declaratory relief. A trial was had before the Court on January 4 through 25, 1983 on Eaton’s remaining claims for damages. The Court now issues its findings of fact and conclusions of law. Fed.R.Civ.P. 52(a). I. FINDINGS OF FACT A. Introduction This dispute revolves around “controllers”. A controller is an electronic minicomputer which consists of many small components soldered to a printed circuit board. As manufactured by Magnavox, the printed circuit board was encased in epoxy and enclosed in a metal housing. The controller is but one part of an “anti-lock” system which was manufactured by Eaton. An anti-lock system is designed to regulate electronically the application of brakes on a vehicle axle, in order to prevent the brakes from “locking”. To accomplish this objective, a “wheel speed sensor” on each wheel sends an electronic signal to the controller. The controller processes these signals and sends a different signal to a solenoid and valve. The valve then regulates the functioning of the air brakes on the axle. When the controller detects that certain abnormal conditions exist, it sends a signal to a warning lamp which is installed in the dashboard of the vehicle. When the warning lamp is activated, the vehicle’s driver is informed that these abnormal conditions are occurring. Between 1973 and 1976, Magnavox manufactured controllers pursuant to its contract with Eaton. Eaton, in turn, sold its anti-lock systems to original equipment manufacturers (“OEMs”) of trucks. After some of the OEMs complained to Eaton that the anti-lock systems were not functioning correctly, Eaton concluded that the controllers built by Magnavox did not conform to Eaton specifications in at least two material respects. The OEMs notified the National Highway Traffic Safety Administration (“NHTSA”) of the problems they had encountered, and Eaton instituted a recall of all of the controllers which had been installed in vehicles. After Magnavox refused to accept responsibility for the alleged defects in the controllers, Eaton filed this action. B. Formation of the Contract In September, 1972 Curtis Eibling of Eaton contacted Richard Hoffman of Magnavox, to inquire as to whether Magnavox would be interested in manufacturing controllers for Eaton. Magnavox officials decided that they were interested in the project, and a series of negotiations ensued. In addition to Eibling and Hoffman, others involved in these negotiations were Warren Matthey of Eaton and J.T. Smith and William Nank of Magnavox. On March 1, 1973, Eaton first issued a purchase order to Magnavox for the purchase of controllers. This purchase order, number 102-15-012, included all of Eaton’s standard terms and conditions. Three of the paragraphs included in these standard terms stated as follows: 5. INSPECTION AND ACCEPTANCE: ... Acceptance of all or any part of the goods shall not be deemed to be a waiver of Buyer’s right either to cancel or to return all or any portion of the goods because of failure to conform to this contract, or by reason of defects, latent or patent, or other breach of warranty, or to make any claim for damages, including manufacturing costs, damage to materials or articles caused by improper boxing, crating or packing, and loss of profits or other special damages occasioned the Buyer. Such rights shall be in addition to any other remedies provided by law. 7. SELLER’S WARRANTIES: Seller hereby warrants that the whole of the goods furnished hereunder shall be of merchantable quality and fit for Buyer’s purposes and that they shall conform with Buyer’s instructions, specifications, drawings, and data. Seller hereby further warrants that the whole of the goods furnished hereunder shall conform to all representations, affirmations, promises, descriptions, samples, or models forming the basis of this contract. Seller agrees that these warranties shall survive acceptance of the goods. Said warranties shall be in addition to any warranties of additional scope given by Seller to Buyer. None of said warranties and no other implied or express warranties shall be deemed disclaimed or excluded unless evidenced by a purchase order change notice or revisions issued and signed by Buyer. 12. INDEMNIFICATION: Seller further agrees to indemnify and save Buyer harmless from any and all losses, liabilities, damages, suits, actions, proceedings, subrogations, and expenses, including court costs and reasonable attorneys’ fees, related in any way to this contract, or the services performed or goods delivered under this contract, except for goods manufactured entirely to Buyer’s specifications, which are claimed or made by any person, firm, association, or corporation, including employees, workmen, servants, or agents of the Seller and his subcontractors arising from any cause or for any reason whatsoever. Seller further agrees, upon receipt of notification, to promptly assume full responsibility for the defense of any and all such suits, actions, or proceedings which may be brought against Seller or against Buyer. In the event Buyer’s machinery or equipment is used by Seller in the performance of any work that might be required under this contract, such machinery or equipment shall be considered as being under the sole custody and control of Seller during the period of such use by Seller. On March 2, 1973 Nank (M) responded to this purchase order. Part of his response was a document entitled “Amendments to Eaton Corporation Purchase Order Terms and Conditions”, which stated in part: Clause Comment 5 Delete the fourth paragraph in its entirety and substitute the following: “Buyer’s acceptance of part of the goods shall not be deemed to be a waiver of Buyer’s right either to cancel or to return any other portion of the goods because of failure to conform to this contract. However, Buyer’s acceptance of goods will be considered to be final except as to latent defects, fraud or gross mistakes amounting to fraud.” 7 & 12 Delete pending final negotiations on specifications, acceptance test details and MTBF. It is our present intent to warrant against defects in workmanship and material for a period of twelve months after delivery and to indemnify Eaton against non-conformances to specifications but in no instance will Magnavox accept responsibility for consequential damages. Eaton’s first purchase order did not result in the formation of a contract between Eaton and Magnavox. On June 27, 1973 Nank (M) sent Eaton a second proposal for the manufacture of controllers. Nank’s letter stated that this proposal was made “subject to the comments submitted on 2 March 1973 pertaining to Terms and Conditions.” On July 20, 1973 Eaton issued a second purchase order, number 102-15-012. Attached to this purchase order were the same standard terms and conditions which had been part of Eaton’s first purchase order. Nank returned a signed copy of this purchase order to Eaton on July 25, 1973. In an attached letter, he repeated the amendments to Eaton’s standard terms which he had requested earlier. After reciting these requested amendments, he stated: “Please let me know if these changes are acceptable.” The issuance of this second purchase order also failed to result in the formation of a contract. The contract was finally executed as the result of a meeting on March 1, 1974. Present at this meeting were Mat-they (E), Nank (M), and Edward Byer of Magnavox. At this meeting Eaton can-celled its first two purchases orders and issued a third, number 102-15-066. This purchase order required Eaton to purchase from Magnavox “a minimum of 35% of its aggregate controller requirements for the two-year period beginning with the effective date of FMVSS 121 providing that all units meet Eaton-Brake Division specs and that all prices remain reasonably competitive in comparison to domestic competition.” Attached to the purchase order was a copy of Nank’s (M) letter of July 25, 1973, in which he had repeated Magnavox’s requested changes in the terms of the purchase order. Under the portion of the letter where Nank had stated: “Please let me know if these changes are acceptable,” Matthey (E) wrote: “OK W.A. Matthey 3-1-74”. Therefore, the terms of the contract between Eaton and Magnavox consisted primarily of Eaton’s standard terms, as modified by the requests made by Nank (M) on March 2 and July 25, 1973. The only exception is that in February, 1977, the parties agreed to extend the warranty period from twelve to fifteen months. See Exhibit 38 to deposition of James W. Fa-hey. All of the controllers manufactured by Magnavox were made pursuant to this purchase order and were subject to these terms. During the contract negotiations, the parties never discussed the meaning of Magnavox’s promise to “indemnify Eaton against nonconformances to specifications”, nor did they define the term “consequential damages”. They never discussed what Magnavox’s liability would be if Eaton were required to institute a recall of the controllers for any reason. Magnavox was aware, however, that the controllers were to be used in Eaton’s anti-lock systems, that they would be installed on trucks, and that they would be difficult to repair once installed. C. Selection of Parts and Production of the Controllers During the early negotiations which took place between September 1972 and March, 1973, Eaton officials supplied Magnavox with a list of the components which Magnavox was to use in manufacturing the controllers. In this list, Eaton specified that all except one of the operational amplifiers (“op amps”) should be “LM 1458N”s, a component which was manufactured by National Semiconductor. During these early meetings, Eibling (E) showed Hoffman (M) a sample of the controllers then being manufactured for Eaton by Sylvania Corporation. Hoffman took a photograph of this sample controller. He later found that instead of LM 1458Ns, this sample contained RC 4558DNs, a different op amp which is manufactured by Raytheon Company. Nevertheless, the last parts list which Eaton submitted to Magnavox prior to Magnavox’s initial bid on the project called for the use of six components described as “LM 1458 Dual Operational Amplifier or approved equivalent.” In calculating the costs of producing the controller, Magnavox used a price it had obtained for “LM 1458N”s manufactured by National. Several additional parts lists were exchanged by Eaton and Magnavox during contract negotiations, and indeed throughout the production of the controllers by Magnavox. On March 16, 1973, Eibling and E. James Lane of Eaton told Magnavox that Magnavox should seek alternate vendors for each of the components to be used in manufacturing the controllers, in order to reduce the costs of production. On April 25, 1973, Magnavox received another parts list from Eaton which called for National LM 1458Ns. At some point during this period, Harry Gill visited Magnavox. Gill was a design engineer for Raytheon, and he had invented the RC 4558DN. During his visit, he explained the differences between the RC 4558DN and the RC 1458DN. The primary difference, he explained, was that the 4558’s input transistor had a PNP input bias current while the 1458’s input transistor had an NPN input bias current. Gill testified as to the foregoing when he was deposed by the parties in 1982. He could not recall who was present at his visit at Magnavox. He testified that at some time between February, 1972 and April, 1973. Raytheon had modified the design of the 4558 so that it had a PNP input bias. He testified that prior to that time, the 4558 had an NPN input bias. Raytheon modified the design because an op amp with a PNP input bias could be manufactured at a lower cost. Gill further testified that in “ninety-nine percent of the applications”, an op amp would perform the same functions regardless of whether its input bias was PNP or NPN. Gill deposition at 72. He did not testify, however, that he informed Magnavox personnel to this effect when he visited them in 1973. On May 29, 1973, Magnavox prepared its own parts list for the controller. This list called for the use of “LM 1458N”s and listed as manufacturers of this component Motorola, Raytheon, Signetics, and National. As mentioned above, one of Magnavox’s early proposals to Eaton was contained in a letter from Nank (M) to Mat-they (E), dated June 27, 1973. Attached as a part of Magnavox’s proposal was a parts list prepared by Magnavox. This list called for “LM 1458N”s and listed as manufacturers Motorola, Raytheon, Signetics, National, MIL, and Fairchild. Between May and July, 1973, Magnavox conducted tests on the components which it was receiving from its vendors, including the op amps. Among other tests, Magnavox tested the input bias current of the op amps, to determine whether these components met their manufacturers’ specifications. ' As contract negotiations were continuing between May and August, 1973, Magnavox delivered to Eaton twenty prototype controllers. Magnavox used both National LM 1458Ns and Raytheon RC 4558DNs in manufacturing these prototypes. Between September, 1973 and January, 1974, Magnavox delivered to Eaton approximately two hundred “pre-production” controllers. Like the prototypes, these pre-production models contained both 1458s and 4558s. Both Eaton and Magnavox anticipated that Magnavox would use more than one source to obtain the op amps to be used in the controllers. In mid-1973, Magnavox concluded that it preferred to use one part number to refer to all of these op amps, whatever their source. Accordingly, in August, 1973 Hoffman (M) prepared a part drawing number 619937-101. Thereafter, Magnavox frequently used this number to refer to the op amps. Magnavox preferred this method of numbering for two reasons. First, Eaton had specified that the op amps should function over an extended temperature range, and that they should be “burned in” before they were installed in the controllers. A designation of an “LM 1458N”, standing alone, did not incorporate these special requirements. Magnavox’s drawing, on the other hand, did incorporate these requirements. Second, Magnavox’s internal policies called for the use of one number for all equivalent parts, even if they were furnished by more than one manufacturer. Therefore, Magnavox wanted to use number 619937-101 to refer to all of the op amps used in the controllers, regardless of who had manufactured them. Hoffman (M) selected the parts which were included in this drawing. In addition to parts manufactured by Signetics, Motorola, and Microsystems, he selected the National LM 1458N and the Raytheon RC 4558DN. After making an investigation, Hoffman had determined that these five parts were equivalent and that each would meet Eaton’s requirements. As a part of this investigation, Hoffman compared the specifications contained in National and Raytheon catalogs. He did not compare the schematic drawings contained in these catalogs, or the input bias current of the two devices. Hoffman testified at trial that he considered an op amp to be nothing more than a “black box”, and that he was not interested in its internal circuitry so long as it met its performance specifications. In September, 1973, Fred Williams of Eaton visited Magnavox. Hoffman (M) testified that Williams was given all of Magnavox’s internal part drawings during this visit. Neither Hoffman nor any other witness, however, could recall specifically whether Magnavox’s drawing number 619937 was shown or given to Williams at this time. On September 28, 1973, Williams wrote in an internal Eaton memo that Eaton employees were checking all available sources for controller components, to ensure that all relevant specifications were acceptable. On October 18, 1973, Eaton internally released for “tooling” one design for its controller. The parts list which was released by Eaton that day called for “National LM 1458N”s. In December, 1973, Richard Hossman' replaced Hoffman as Magnavox’s project engineer in charge of the anti-lock program. On July 31, 1974, Williams (E) wrote to Nank to authorize Magnavox to begin production of the controllers. In his letter, Williams stated that “all correspondence related to any Magnavox supplied components of Eaton Skid Control Systems dealing with changes to components or circuits must be either authored by or countersigned by Fred Williams.” Plaintiff’s Exhibit 36A. In October, 1974, Magnavox began production of the controllers. Throughout the entire pre-production phase of their business relationship, Eaton and Magnavox had many informal contacts. Many contacts were made via telephone, without any written documentation as to their content. When changes in design were documented in writing, the documentation often occurred only after the changes had taken place. On November 4, 1974, William Deibel of Eaton wrote to Smith (M) and noted that “both of us may have relaxed some of the basic business controls normally followed in less hectic circumstances.” Exhibit 7 to Deibel deposition. Deibel requested that the parties develop “a more formal policy regarding our commitments to you lest any misunderstandings develop ...”. Id. Deibel later testified that this letter was prompted by Eaton’s concern that Magnavox had often performed work without prior authorization from Eaton, and then billed Eaton for such work. Deibel testified that Eaton officials believed that they should not have been required to pay for such work, but that they did so to ensure Magnavox’s future cooperation. On November 22, 1974, Hossman (M) wrote to Krawczak (E). and enclosed a revised copy of Magnavox’s parts list for the controllers. This list called for the “LM 1458N”s and listed Motorola, Raytheon, Signetics, National, MIL, and Fairchild as manufacturers of this component. On December 5, 1974, Eaton and Magnavox employees met. At this meeting it was mentioned that Magnavox was using Raytheon op amps exclusively in manufacturing the controllers. It was not mentioned, however, which model Raytheon op amps were being used. On January 14, 1975, Hossman wrote to Robert Medler at Eaton to propose a different method of burning in the op amps. In this letter, Hossman stated that Magnavox’s part number 619937-101 represented a burned-in “LM 1458N”. Hossman also stated that Magnavox had produced 14,000 controllers as of that date, and had used 94,000 “LM 1458Ns”. On January 20, 1975, Hoffman wrote to Krawczak and stated that Magnavox had performed experiments to achieve cost reductions in the controllers. He stated that in one of these experiments, Magnavox had replaced the “LM 1458 Op Amps” in the controller. On January 24, 1975, Eaton prepared a revised parts list. This list called for the use of either “LM 1458 N (NAT)” or “RC 1458 DA (RAY)”. On February 14, 1975, Byer wrote a memo to Smith, Nank, Hoffman, and Harley Meadows at Magnovox in which he stated: “Warren warned us again to build only the Eaton design and specifications and to not make any arbitrary or unilateral revisions.” Plaintiff's Exhibit 237. On March 13, 1975, Williams (E) sent a then-current Eaton parts list to Hossman. He asked Hossman to make whatever revisions were necessary in order that the list would accurately reflect Magnavox’s use of components. On March 27, 1975, Hossman returned a “marked up” copy of this list to Williams. On the page of the list which included the op amps, Hossman drew a line through the National and Raytheon part numbers and wrote “615268 Mag.” In a letter attached to the list, Hossman stated that “Magnavox reserves the right to utilize parts and materials from sources in addition to those listed on the Eaton parts listing, provided they are equivalent to the listed vendor’s product. In any case, where such substitutions are required Eaton will be notified prior to making the change.” Plaintiff’s Exhibit 53. On April 1, 1975, Nank wrote to James Fahey of Eaton to convey the same message. He attached a copy of the parts list which Hossman had marked up, and he stated that Magnavox reserved the right to substitute equivalent parts. He also stated that Eaton would be notified prior to any such change. On April 8, 1975, Hossman sent a second marked-up copy of Eaton’s parts list to Williams. On this list, Hossman did not draw a line through the National and Raytheon part numbers. He did, however, add the Magnavox number 619937-101. On June 18, 1975, Eaton changed its parts list to reflect this change. For the first time, Eaton listed “619937-101 (MAG)” as its approved part number. On April 10 and 11, 1975, Williams (E) went to Magnavox to meet with Hossman and other Magnavox personnel. Williams wrote a memo to Krawczak (E) on April 17 to summarize his activities during this visit. He stated that he had received the parts list which was marked up by Magnavox, and added: “the part specifications covering the parts which we list by Magnavox part number were received.” Plaintiff’s Exhibit 55. Williams also stated that he would review these specifications and inform Hossman of any discrepancies. At trial, Williams testified that he did not receive Magnavox’s 619937 drawing at this time. Hossman could not recall whether this drawing was given to Williams during this visit. On April 24, 1975, Fahey (E) responded to Hossman’s letter of March 27. Fahey stated that no changes or substitutions were to be made to the Eaton parts list without written approval from Eaton. On July 29, 1975, Hossman wrote to Krawczak (E) regarding the alternate method of burning in the op amps which he had mentioned earlier. He stated that in order to evaluate this method, tests had been performed on “400 standard production RC4558DN units.” Defendants’ Exhibit I. The clear implication of the letter, however, was that Raytheon had performed these tests on its “standard production RC4558DN units.” Hossman did not state that Magnavox was using these particular op amps in producing the controllers. In fact, Magnavox was using the Raytheon 4558s in production. On August 11, 1975, Raytheon’s director of marketing wrote to Magnavox. He informed them that Raytheon had discovered that some of its recently-manufactured 4558s were defective. He requested that Magnavox return all of its unused 4558s for replacement. Magnavox returned approximately 40,000 4558s to Raytheon pursuant to this request. Magnavox did not inform Eaton of this development, however. In August, 1975, Eaton decided that Magnavox should use a second source for op amps, in addition to Raytheon. Eaton requested Magnavox to perform certain tests on the Fairchild 1458 op amp, to determine whether it would be an acceptable alternative. After Magnavox performed these tests and obtained favorable results, Fahey (E) wrote to Byer on December 3, 1975 authorizing Magnavox to “add Fair-child 1458 quad-amps ... to your approved vendor list.” Defendants’ Exhibit K5. Fa-hey added: “These may be used along with or as an alternate to Raytheons presently in use.” Id. In November, 1975, Williams (E) began efforts to assemble a complete set of the Magnavox part drawings which were in use in the anti-lock program. He requested Magnavox to send him a 619937 drawing, among others. On December 4, 1975, Eaton received a complete set of the Magnavox drawings, including the 619937. Plaintiff’s Exhibit 61I. As discussed above, this drawing called for the use of Raytheon RC 4558DNs, among other parts. Williams testified at trial that he did not pay close attention to these drawings when he received them, but merely forwarded them to Krawczak (E). On March 23, 1976, Krawczak (E) sent Hossman (M) a document entitled “Magnavox Approved Sources Listing”. This list included part number “619937(MAG)”, which was described as a “Dual 741 Op-Amp LM1458”. Defendants’ Exhibit R8. The approved vendors for this part were Raytheon, National, and Fairchild. This was apparently the last parts list prepared by either Eaton or Magnavox during Magnavox’s production runs one through twenty-six. The last controllers which were manufactured during these runs were delivered to Eaton on April 30, 1976. Eaton had, by this time, purchased 42,947 controllers from Magnavox. The total purchase price Eaton paid for these controllers was $2,628,833.65. On May 24, 1976, Fahey (E) wrote to Byer (M) regarding the transition to production run twenty-seven of the controllers. Fahey stated that Eaton wanted to know if Magnavox or Sylvania had any surplus parts which would not be used in the run twenty-seven controllers. Fahey stated that Eaton had on hand “8,620 Raytheon RC 4558 Op-Amps that are burned-in”, and asked Byer if Magnavox could use them. Defendants’ Exhibit N5. Fahey later testified that he believed these were surplus parts given to Eaton by Sylvania. By way of summary, the Court makes the following findings. Magnavox used three types of op amps in producing the controllers. These were the Raytheon RC 4558DN, the National LM 1458N, and the Fairchild 1458P1. The majority of the op amps used by Magnavox were Raytheon RC 4558DNs. Eaton informed Magnavox numerous times that the National 1458 was approved for use in the controllers. Moreover, Eaton expressly authorized Magnavox to use the Fairchild 1458, in a letter of December 3, 1975. Eaton never expressly authorized Magnavox to use the Raytheon 4558. Eaton certainly knew that Magnavox was using Raytheon op amps. Eaton did not know, however, that Magnavox was using the RC 4558DN rather than the RC 1458DN. In several of its parts lists, Eaton had approved the use of Raytheon 1458s. Prior to receipt of Magnavox’s 619937 drawing, Eaton had no indication that Magnavox was using the 4558 rather than the 1458. Once Eaton received the Magnavox 619937 drawing, it had notice that Magnavox considered the Raytheon 4558 and the National 1458 to be equivalent parts. Eaton received a copy of this drawing on December 4, 1975; there has been no showing that Eaton ever received a copy of this drawing before that date. Yet even the notice provided by receipt of this drawing was ambiguous and somewhat contradictory. Hossman had earlier told Eaton that 619937-101 represented an “LM 1458N” which had been burned in. In addition, this drawing included other part numbers which had never been approved by Eaton and were never used in producing the controller. Finally, Eaton never expressly authorized Magnavox to use the Raytheon 4558s. In contrast, Eaton did expressly authorize the use of National 1458s, Fair-child 1458s, and the Raytheon 1458s which Magnavox never used. Moreover, it was the clear understanding of all of the parties that Magnavox could not substitute any other parts unless it had received Eaton’s prior approval. Accordingly, it was improper for Magnavox to use Raytheon RC 4558 DNs in manufacturing the controllers. D. Eaton’s Discovery and Investigation of Defective Controllers Shortly after production of the controllers began, Eaton discovered that many of them contained faulty solder connections. On November 19, 1974, Medler wrote a report in which he stated that Eaton technicians had found many cracked solder joints in the controllers. In a meeting at Magnavox on December 5, 1974, the parties discussed the fact that faulty soldering had caused a number of controller failures. On January 14, 1975, Hossman (M) wrote to Krawczak (E) regarding a controller which Eaton had returned to Magnavox. Hossman stated that this controller had many cracked solder joints, and that the leads from some of the components had broken. Hossman noted that the lead breakage occurred on those components that “stood up” when inserted in the printed circuit board. Hossman believed that the breakage was caused by the expansion and contraction of the “potting” material during the manufacturing process. On February 5, 1975, Eaton concluded that the most common cause of controller failures was cracked solder joints of large components. This conclusion was the result of tests which Eaton had performed on failed controllers; Eaton found that one of the components most often affected was a capacitor designated as “C 104” in the controller design. In March, 1975, Tom Gee of Eaton performed tests on “unpotted” controllers obtained from Magnavox. Gee found that several of the solder joints were stressed and that some of them were broken. After reporting the results of his tests, Gee stated: “I have not developed any explanation of how these joints can fail without any potting.” Plaintiff’s Exhibit 73. On March 13, 1975, Williams and Krawczak visited Magnavox, and observed the manufacturing process there. In a report of this visit, Williams expressed dissatisfaction with Magnavox’s soldering techniques. He stated: “I informed Magnavox that it was of utmost importance that they cure their solder problems.” Plaintiff’s Exhibit 71. In May, 1975, Hossman (M) discovered one source of these solder problems. C 104 was located at the very edge of the printed circuit board. Hossman found that in order to fit the controller into its housing, Magnavox employees were manually pushing this large component towards the center of the board. Hossman found that the resulting stress on the C 104 solder joints could cause them to break. Although Magnavox had already shipped to Eaton approximately 22,000 controllers which had been manufactured in this way, Hossman did not inform Eaton of his discovery. In September, 1975, Eaton received three controllers as a return from one of its customers. Eaton found that each of these had a broken solder joint at C 104, and it sent two of them to Magnavox for further analysis. At a meeting at Magnavox on September 15, 1975, Eaton officials again informed Magnavox that some of its customers had experienced problems with the controllers that were caused by cracked solder joints at C 104. On September 22, 1975, Byer wrote an internal memorandum to Smith, Hoffman, Hossman, Meadows, and three other Magiiavox employees. He directed that Magnavox should resolder certain components, including C 104, which required “particular care in assembly.” Plaintiffs Exhibit 84. Byer added: “These are the components which have exhibited repeated solder joint fracture. This is our problem, and we will have to absorb the cost.” Id. Hossman (M) wrote to Krawczak (E) on September 26, 1975 regarding the two controllers which Eaton had sent to Magnavox for analysis. For the first time, he informed Eaton of the manufacturing problems which he had discovered in May. He described the movement of C 104 during the potting process and added: “This movement exerted a large force on the solder joint at the negative lead and could lead to intermittent connections.” Plaintiffs Exhibit 82. Hossman stated that Magnavox had taken corrective action in May to eliminate this problem. On September 80, 1975, Krawczak wrote a memo to other Eaton officials to convey the contents of Hossman’s report. Krawczak stated that the controllers manufactured prior to May, 1975 could be expected to exhibit unusually frequent C 104 failures. Krawczak believed at that time that the C 104 failures would generally be detected by the “supertester”, a device which tested all of the electrical functions in the controller at the end of the manufacturing process. This belief proved to be incorrect. The problem caused by the solder failures was intermittent; even a controller which passed the “supertester” could malfunction at a later time. Eaton continued to receive returns of controllers from its customers. In some instances, customers complained that the controller had failed to function, yet the warning light had not been activated. Eaton evaluated these returned units in order to determine the cause of the failures, and found that faulty solder connections at C 104 were a common cause. Eaton then began a major investigation into the causes of the controller failures. The head of this investigation was David Klien, who was a senior project engineer at the Eaton Research Center. Klien conducted a series of tests on controllers which had been returned to Eaton by its customers. Klien’s tests involved primarily the performance of the controllers’ failsafe functions. As designed by Eaton, the controller had two different failsafe functions. The short-term failsafe occurred when the controller detected abnormal conditions which were external to the controller. The long-term failsafe occurred when the controller detected abnormal conditions within the controller itself. In either instance, the anti-lock system was designed to disable itself. At the same time, the warning light in the vehicle’s dashboard was to be activated, and the affected axle was to be returned to manual, non-regulated braking. The short-term failsafe was to occur within one second of the onset of the abnormal conditions; the long-term failsafe was to occur within eight to ten seconds. In performing tests on the failed controllers, Klien found that when C 104 was “open”, the controller would develop an abnormal condition called a “constant solenoid”. Although the controller was designed to failsafe when this condition occurred, Klien found that some of the returned controllers did not. These controllers either failsafed only after an extended period of time, or did not failsafe at all. When the controller did not failsafe, this meant that the controller was not operating, yet the axle had not been returned to manual braking. As a result, the vehicle would have no brakes whatsoever on tjie affected axle. Klien then discovered that many of the controllers contained Raytheon RC 4558DN op amps rather than National LM 1458Ns. After performing further tests, he found that this substitution had a significant effect on the performance of the controllers. He found that when a controller contained op amps with an NPN input bias current, such as the National 1458, an open C 104 would cause the unit to failsafe in less than one second. When that occurred, the warning light would activate and the axle would be returned to manual braking. He found that when a controller contained op amps with a PNP input bias current, like the Raytheon 4558, it performed quite differently. In that event, the controller would produce an abnormal signal which was sufficient to cause the “constant solenoid” but was often insufficient to activate the failsafe function. Klien found that 74% of the controllers containing PNP input op amps would experience a “normal” long-term failsafe and disable the anti-lock system in six to twelve seconds. He found that 6% of these controllers experienced an “extended” long-term failsafe, disabling the anti-lock system within fifteen seconds to one minute after the onset of the constant solenoid. Klien found that the remaining 20% of these controllers did not failsafe at all. As described above, this meant that there would be no brakes on the affected axle. Klien reported the results of his tests in a memorandum to Robert Sokalski of Eaton on July 14, 1977. After further analysis of the problem, Eaton found that it could not identify which of the controllers manufactured by Magnavox contained op amps with a PNP input bias current. Moreover, Eaton found that it could not predict which of these op amps would experience the “normal” long-term failsafe and which would not. Eaton then attempted to develop a means of repairing the controllers which had already been installed in vehicles, in order to eliminate this problem. Eaton found that there was no practicable method of performing such repairs. Fahey (E) called Byer (M) on January 25, 1978. Fahey expressed his understanding that Magnavox’s warranty was to be “indefinitely extended if the vendor installs the wrong parts which results in deficient performance.” Plaintiff’s Exhibit 87C. Byer responded to this call in a letter of February 3, 1978. He disagreed with Fa-hey's understanding of the warranty terms, but added: “Magnavox’s responsibility with respect to use of incorrect or unapproved parts is also covered in the original warranty under ‘non-conformance’.” Id. In March, 1978, Eaton informed all of its customers of the failsafe malfunctions it had discovered in the controllers. By this time, Eaton knew of at least thirty-five verified occurrences in which a customer’s brakes had failed in the manner described above. On March 6, 1978, John Rodewig of Eaton wrote to Smith (M) to revoke Eaton’s acceptance of all controllers which Magnavox had manufactured in runs twenty-six and earlier. Rodewig contended in his letter that Magnavox had used non-specified op amps in manufacturing the controllers, and that this constituted a latent defect in the controllers. Rodewig and Smith met to discuss the problem on March 20, 1978. On March 23, Smith wrote to Rodewig. He contended that Magnavox was not responsible for the alleged defects in the controllers. He stated: “[0]nly approved parts, including the one in question, were installed in Eaton controllers.” Plaintiff’s Exhibit 93. E. Eaton’s Recall of the Controllers On March 29, 1978, a representative of Mack Trucks wrote to NHTSA to inform the agency that it was recalling many of its trucks in order to replace the controllers which it had purchased from Eaton. .This letter stated that the “constant solenoid” condition had caused brake failures in twenty of Mack’s trucks, and that the potential threat to safety had caused Mack to institute this recall. On March 30,1978, Miles Tuttle of Eaton wrote to Mack Trucks, offering to provide replacement controllers and reimburse Mack for reasonable replacement and administrative costs. On April 7, 1978, NHTSA sent a telegram to Rodewig. The agency informed Eaton of the letter it had received from Mack Trucks. The agency asked Eaton to supply its analysis of the defect in the controllers, and a list of the OEMs who had purchased these controllers. On April 11 and 14, 1978, Tuttle responded to this telegram. He informed NHTSA of the problem, explained Eaton’s analysis of the problem, and listed the OEMs who had purchased the controllers. On June 30, 1978, NHTSA issued a Consumer Advisory announcing the recall of all of the controllers which Magnavox had manufactured for Eaton in runs one through twenty-six. Because Eaton had reasonably concluded that the controllers could not be repaired, Eaton then arranged to replace all of the controllers which were being recalled. Tuttle found that Eaton had shipped 42,947 of the defective controllers to its customers. He then spoke to representatives of the OEMs who had participated in similar recalls. He found that Eaton could expect approximately sixty-five percent of its customers to respond to the recall notices and request replacement controllers. Accordingly, he concluded that Eaton would not need to order 42,947 replacement controllers in order to fulfill the' demands of the recall program. Eaton purchased controllers from Magnavox for use in the recall program. These controllers were of the run twenty-seven design. See n. 22 supra. In addition, they were supplied by Magnavox in their unpotted form. While Magnavox had sealed the original controllers in their housings, Eaton performed this function for the replacement controllers. Eaton then made arrangements with its customers to replace the defective controllers. Eaton shipped the new controllers to its customers. Eaton employees performed some of the labor required to replace the defective controllers; employees of the OEMs performed some of this work as well. Once the replacement controllers were installed, the defective controllers were returned to Eaton. Eaton did not return these defective controllers to Magnavox, but rather stored them in an Eaton warehouse. Eaton considered these defective controllers to be worthless, and did not attempt to repair them. Eaton devised an accounting method which it used to determine its costs in executing the recall program. As costs of the program, Eaton included the cost of purchasing and manufacturing the replacement controllers, the reasonable administrative costs which were charged to Eaton by its customers, the cost of paying Eaton employees to investigate the problem and assist its customers in replacing the controllers, and other freight and handling expenses. Eaton’s calculation of its total recall costs is comprised of three figures. The first is Eaton’s cost of obtaining the replacement controllers which were used during the recall program and the administrative costs of the program. Eaton actually used only 7260 replacement controllers to replace the defective controllers. The cost of these controllers was $336,070.35. In November, 1980, Eaton disposed of 5,250 replacement controllers which it had obtained from Magnavox. These controllers had not yet been used in the recall program, and Eaton concluded that they would not be required. The cost of these controllers was $236,551.86. Adding the other costs of administering the recall program, Eaton found that its costs for executing the recall were $1,023,615.18. The second cost calculated by Eaton was the value of the 5,612 replacement controllers that were not used in the recall campaign and were still in Eaton’s inventory when the recall program ended in 1982. Using the October, 1979 cost figures, Eaton found that the cost of these controllers was $261,853.10. The third cost calculated by Eaton was the cost of investigating the problem in the controllers. This consisted of the salaries paid to Eaton employees during the time they worked on this project. Eaton calculated that this cost was $117,445.00. Adding these three figures together, Eaton found that the total cost of the recall program was $1,402,913.28. Magnavox has not challenged the accuracy of any of these figures. II. CONCLUSIONS OF LAW A. Breach of Contract Eaton claims that once it learned of the problems which had developed in the con-, trailers, it revoked its acceptance of all of the controllers manufactured by Magnavox in production runs one through twenty-six. Complaint at ¶ 33. Eaton further claims that it is entitled to recover damages as a result of its revocation. The Court finds that Eaton’s attempt at revocation was ineffective, and that it cannot recover damages under this theory. Eaton attempted to revoke its acceptance of the controllers on March 6, 1978, when Rodewig wrote to Smith. This occurred nearly two years after the last of these controllers was delivered to Eaton. Although the mere passage of time did not prevent Eaton from revoking its acceptance, the intervening events which occurred during this time did preclude Eaton from making an effective revocation of its earlier acceptance. Under the Uniform Commercial Code (“U.C.C.”), six elements are required to establish an effective revocation. The buyer must establish that: 1. The nonconformity in the goods substantially impairs their value; 2. His acceptance of the goods was caused by his reasonable assumption that the nonconformity would be cured, the difficulty of discovering the nonconformity, or the seller’s assurances; 3. The revocation occurred within a reasonable time after the buyer discovered or should have discovered the ground for it; 4. The revocation occurred before any substantial change in the condition of the goods which was not caused by their own defects; 5. He has given notice to the seller; and 6. If he has taken physical possession of the goods, he has held them with reasonable care for a sufficient period of time to permit the seller to remove them. Mich.Comp.Laws Ann. §§ 440.2602, 440.-2608; see generally J. White and R. Summers, Uniform Commercial Code at § 8-3 (2d ed. 1980). Without addressing the other four elements, the Court finds that Eaton failed to establish two of these elements. First, Eaton attempted to revoke its acceptance only after there had been a substantial change in the condition of the controllers. In United States v. Crawford, 443 F.2d 611 (5th Cir.1971), the court interpreted Georgia’s enactment of the U.C.C., which is identical in this respect to Michigan’s enactment. In that case, defendant had purchased a number of fuel filter separation units for installation at a U.S. Naval Air Station. After defendant installed the units at the Air Station, the seller brought an action for the unpaid portion of the purchase price. The court held that defendant had accepted the goods as a matter of law, and could not avoid payment on the ground that he had revoked his acceptance. The court stated: A buyer who has accepted goods may under certain conditions revoke his acceptance. Georgia UCC § 2-608. However, once Crawford had accepted and installed the units supplied in place on the Albany site, any subsequent attempt at revocation was ineffective. 448 F.2d at 613. Accord, Management Assistance, Inc. v. Computer Dimensions, Inc., 546 F.Supp. 666 (N.D.Ga.1982). In the case at bar, Eaton first sought to revoke its acceptance of the controllers only after the controllers had been sold by Eaton to its customers, had been installed in trucks throughout the country, and had been used by the OEMs for extended periods of time. The sale, installation, and use of the controllers constituted a “substantial change in condition of the goods” which was not caused by any defect in the controllers, and it therefore precluded Eaton from subsequently revoking its acceptance. Mich.Comp.Laws Ann. § 440.2608(2). Second, Eaton did not and could not hold the controllers in order for Magnavox to remove them. Under Mich.Comp.Laws Ann. § 440.2608(3), a buyer who revokes his acceptance “has the same rights and duties with regard to the goods involved as if he had rejected them.” These duties are set forth, in part, in Mich.Comp.Laws Ann. § 440.2602(2)(b). Under this section, a buyer who has taken physical possession of the goods has a duty after revocation to hold the goods for a reasonable time in order to allow the seller to remove them. See White and Summers, supra, at 316. In the case at bar, Eaton attempted to revoke its acceptance of 42,947 controllers, despite the fact that only 7260 of the controllers were returned to Eaton by its customers. Assuming arguendo that Eaton fulfilled its duty with respect to the 7260 controllers which were returned by its customers, it is clear that Eaton did not hold for Magnavox the remaining 35,687 controllers which were in the possession of its customers. It would have been impossible for Eaton to do so, because these controllers were not within Eaton’s possession or control. In sum, Eaton sought to revoke its acceptance of the controllers only after the controllers had undergone a substantial change in condition, and Eaton did not hold for Magnavox the controllers which it had allegedly revoked. Eaton’s attempted revocation was therefore ineffective, and its claims in Count I must fail. B. Breach of Express Warranty Eaton claims that Magnavox breached an express warranty it issued in connection with the manufacture of the controllers. Eaton further claims that the recall of the controllers was necessitated by the breach of warranty, and it seeks to recover its damages. The court finds that Eaton is barred from recovering any damages under this theory; although Magnavox did breach its express warranty, Eaton failed to supply Magnavox with timely notice of the breach as required by Mich. Comp.Laws Ann. §§ 440.2607 and 440.2714. Magnavox issued two express warranties in its agreement with Eaton. First, Magnavox warranted against defects in workmanship and materials for a period of twelve months after delivery. Second, by agreeing to “indemnify Eaton against nonconformances to specifications”, Magnavox warranted that the controllers would conform to the specifications issued by Eaton. Magnavox breached the second warranty by using the Raytheon 4558 op amps in manufacturing the controllers. As noted above, Eaton issued several parts lists as part of the specifications for the controllers. These parts lists repeatedly called for the use of the National LM 1458N. The only other op amps which were approved by Eaton were the Raytheon RC 1458DN and the Fairchild 1458P1. Eaton never expressly authorized the use of the Raytheon RC 4558DN. Eaton did approve the use of Magnavox’s part number 619937, but the Court cannot regard this as approval of the use of the Raytheon 4558s. Magnavox had earlier informed Eaton that 619937 designated a burned-in LM 1458N, so the receipt of the 619937 drawing provided ambiguous notice as to the other part numbers contained therein. Moreover, even this ambiguous notice was not received by Eaton until December 1975, shortly before Magnavox completed manufacturing the controllers in runs one through twenty-six. Magnavox argues that Eaton waived any breach of this warranty by accepting prototype and pre-production controllers containing the 4558s without notifying Magnavox that the use of 4558s was improper. Even if Eaton was or should have been aware that these controllers contained 4558s, however, its failure to complain was not a waiver of the breach of warranty. The prototypes and pre-production controllers contained both National 1458s and Raytheon 4558s. They were produced and shipped to Eaton during a period of time in which there were a number of design changes in the controllers; they were not regarded by either party as a definitive model of the controllers. The shipment and receipt of these controllers is, at best, ambiguous evidence of Magnavox’s intent to use 4558s in producing the production controllers. Eaton’s parts lists, on the other hand, were wholly unambiguous. Every parts list issued by Eaton called for the National 1458; none called for the Raytheon 4558. In sum, Magnavox breached its express warranty in using the Raytheon 4558s, and Eaton never waived its rights to recover for this breach. Eaton claims that Magnavox also breached its warranty against defects in workmanship by supplying controllers with defective solder joints. This warranty, however, lasted for only twelve to fifteen months after delivery of the controllers. As the last controller manufactured during runs one through twenty-six was delivered to Eaton in April, 1976, Magnavox’s liability on this warranty had completely terminated by July, 1977. Magnavox cannot rely on defective solder connections alone to justify the subsequent recall of all of the controllers; as discussed below, the recall was necessitated by the combination of these faulty solder connections with Magnavox’s use of the 4558s. Moreover, Eaton made no attempt to establish that any of the particular controllers recalled had experienced solder failures within the fifteen months following shipment. Accordingly, Eaton failed to establish that Magnavox’s faulty solder connections, standing alone, constituted a breach of express warranty justifying a recall of all of the controllers produced. Even though Magnavox’s use of the Raytheon 4558s was a breach of warranty, Eaton must establish other elements under the U.C.C. in order to recover damages for the breach. Under Mich.Comp.Laws Ann. § 440.2714, a buyer can recover damages for breach of warranty only if he has accepted the goods and given the notice required under Mich.Comp.Laws Ann. § 440.-2607(3)(a). Under the latter section, “the buyer must within a reasonable time after he discovers or should have discovered any breach notify the seller of breach or be barred from any remedy.” The buyer must provide reasonable notice in order to recover for a breach of warranty. Steel & Wire Corf. v. Thyssen, Inc., 20 U.C.C.Rep. 892 (E.D.Mich.1976). Moreover, what constitutes a “reasonable time” is to be determined under the facts of each case. Mich. Comp.Laws Ann. § 440.1204(2). Eaton argues that it had repeatedly notified Magnavox that many of the controllers had faulty solder connections, and relies on this as adequate notice of the controllers’ defects. Yet the “defect” which led to the recall was not the mere presence of solder failures in some of the controllers. Eaton knew of many such solder failures, but never considered this defect of sufficient magnitude to require a recall of all of the controllers until it discovered that Magnavox had used 4558s in manufacturing the controllers. Prior to this discovery, Eaton’s practice upon receiving controllers with faulty solder connections was to return the controllers to Magnavox for repair or replacement. After Eaton learned that Magnavox had used 4558s, on the other hand, Rodewig (E) informed Smith (M) on March 6, 1978 that this non-conformance was sufficiently serious that Eaton was revoking its acceptance of all of the controllers. Eaton plainly still did not regard the solder failures as a serious problem, as Rodewig informed Smith in his letter that the controllers “as designed” — i.e. with National 1458s — would function properly even when C 104 failed. Magnavox’s use of the 4558s, therefore, was the most important factor in the breach of warranty. Eaton never informed Magnavox that its use of 4558s was considered a breach of warranty until Rodewig’s letter of March 6, 1978. Eaton claims that it was not required to notify Magnavox that the controllers contained 4558s, because Magnavox was already aware of this. This argument misses the point of the notice requirement in Mich.Comp.Laws Ann. § 440.2607, however. Eaton was required to inform Magnavox not only that the controllers contained 4558s, but also that Eaton considered this to be a breach of warranty. Magnavox never received notice to this effect until Smith received Rodewig’s letter which impliedly stated that the use of 4558s was a breach of warranty. Eaton further argues that Magnavox should be estopped to claim that Eaton’s notice was untimely, because Magnavox had led Eaton to believe that it was using the National 1458s in manufacturing the controllers. Even if this argument is accepted, however, Eaton undisputedly knew by July 14, 1977 that Magnavox had used 4558s in the controllers and that this caused problems. After having completed his investigation, Klien issued an internal Eaton memorandum on that date which stated: [T]he Raytheon devices specified and used by Magnavox are not true 1458 devices even though the other five source’s amplifiers, specified by Magnavox on their drawing number 619937, are 1458 type amplifiers. The device number specified by Magnavox for Raytheon is the RC4558DN which is an operational amplifier with specifications similar to the RC1458DN which Raytheon also produces. The major difference between these two devices is that the 1458 amplifiers use NPN input transistors while the 4558 devices use PNP input transistors. This difference totally explains the breakdown voltages depicted in WFM7 and WFM8. (Emphasis supplied.) Plaintiff’s Exhibit 90A. Eaton therefore knew that the controllers contained 4558s; Eaton had long known that some of the controllers had faulty solder connections. Klien’s discovery was the precise basis for Eaton’s later claim that the use of 4558s constituted a nonconformity to Eaton’s specifications, yet Eaton waited nearly eight months before notifying Magnavox that its use of 4558s was considered improper. Eaton has offered no justification for this eight-month delay in providing notice to Magnavox, except for its claims that it was trying to “bracket” the population of controllers containing 4558s, to evaluate possible methods of repair, and to discuss the situation with its customers. Yet Eaton knew by July, 1977 exactly what was causing the controller failures. While Eaton’s activities after that date may have been “reasonable”, they did not obviate Eaton’s duty under the U.C.C. to notify Magnavox of the problem. Rather than notifying Magnavox of the problem, however, Eaton attempted to obtain Magnavox’s unwitting agreement to extend the controllers’ warranty period. As noted above, Fahey called Byer on January 25, 1978 seeking confirmation of his “understanding” that Magnavox’s warranty was to be indefinitely extended if Magnavox installed the “wrong parts.” One of the policies behind the notice requirement in Mich.Comp.Laws Ann. § 440.2607 is to afford the seller an opportunity to prepare for negotiation and litigation. Thyssen, supra; White and Summers, supra, at 422. Eaton’s attempt to extend the warranty period before giving notice of a known defect was inconsistent with that policy, and is evidence of Eaton’s bad faith in attempting to resolve the disputes arising under the parties’ agreement. Although no prejudice to Magnavox was shown in connection with Eaton’s delay in providing notice, this is not required for a finding that the notice was untimely. Ashley v. Goodyear Tire & Rubber Co., 635 F.2d 571 (6th Cir.1980) (applying Michigan law). Under the circumstances, the Court finds that Eaton did not notify Magnavox of the controllers’ defects within a reasonable time after their discovery. Under Mich. Comp.Laws Ann. § 440.2607(3)(a), Eaton accordingly cannot recover any damages for Magnavox’s breach of its express warranty. C. Breach of Implied Warranties Eaton claims that Magnavox breached an implied warranty of merchantability, in that some of the controllers did not function properly. Eaton further claims that Magnavox breached an implied warranty of fitness for a particular purpose, by using the Raytheon op amps. Complaint at ¶ 50. Eaton claims that its recall expenses were caused by these breaches, and it seeks to recover these expenses. The Court finds that Eaton’s failure to provide timely notice bars any recovery under this theory as well. Every sales contract governed by the U.C.C. includes an implied warranty of merchantability, unless it is effectively disclaimed. Mich.Comp.Laws Ann. § 440.-2314. Where, as in the case at bar, the seller knows the particular purpose for which the goods are required and also knows that the buyer is relying on the seller’s skill and judgment, an implied warranty that the goods are fit for such purpose is also created unless effectively disclaimed. Mich.Comp.Laws Ann. § 440.-2315. Magnavox did not effectively disclaim these warranties. Under Mich.Comp. Laws Ann. § 440.2316(2), such a disclaimer must meet the following requirements: [T]o exclude or modify the implied warranty of merchantability or any part of it the language must mention merchantability and in case of a writing must be conspicuous, and to exclude or modify any implied warranty of fitness the exclusion must be by a writing and conspicuous. Magnavox’s attempted disclaimer of these warranties stated merely: Delete pending final negotiations on specifications, acceptance test results and MTBF. It is our present intent to warrant against defects in workmanship and material for a period of twelve months after delivery and to indemnify Eaton against non-conformances to specifications but in no instance will Magnavox accept responsibility for consequential damages. The attempted disclaimer did not “mention merchantability”, nor was any part of its language conspicuous. It was therefore ineffective. Magnavox contends that these implied warranties were deleted by agreement of the parties, even though they were