Strand v. Librascope, Incorporated, 197 F. Supp. 743 (E.D. Mich. 1961)
September 14, 1961
John STRAND, individually and doing business as Strand Engineering Company, Plaintiff,
LIBRASCOPE, INCORPORATED, a foreign corporation, Defendant.
United States District Court E. D. Michigan, S. D.
*744 Langs, Molyneaux & Armstrong, Detroit, Mich., R. F. Molyneaux and Neil A. Patterson, Detroit, Mich., of counsel, for plaintiff.
James D. Tracy, Dykema, Wheat, Spencer, Goodnow & Trigg, Detroit, Mich., Theodore H. Lassagne, Glendale, Cal., of counsel, for defendant.
LEVIN, Chief Judge.
The plaintiff, Strand, a citizen of Michigan, alleges breach of express and implied warranties and fraud and deceit in the sale to him by the defendant, Librascope, a California corporation, of certain components for an electronic digital computer to be built by Strand. These components, which were to be part of the magnetic drum system of the computer, were manufactured by Librascope and designated by it as "MH-10R read/record heads." Jurisdiction rests on diversity of citizenship.
Librascope contends that it breached no express or implied warranties and that, even if it did, its liability is restricted by the following provision which was incorporated in its contract of sale with the plaintiff:
"Librascope instruments and components are warranted to be free from defects in material and workmanship impairing the normal use and service for which they are intended. The liability of Librascope under this warranty is limited to repairing or replacing any instrument or component returned to it within a period of ninety (90) days after delivery to the original purchaser, with all transportation charges prepaid, and found by Librascope to be defective, and in no event shall Librascope be liable for collateral or consequential damages. This warranty is in lieu of any other warranty, express, implied, or statutory (except as to title) and no agreement extending it will be binding upon Librascope unless in writing and signed by an officer of Librascope, Incorporated."
The plaintiff alleges that Librascope made material misrepresentations which induced him to enter into and perform the contract and that a party committing a fraud may not protect himself by a provision in the contract of sale limiting the injured party's right of recovery. J. I. Case Threshing Machine Co. v. Feezor, 1910, 152 N.C. 516, 67 S.E. 1004; Ganley Bros. v. Butler Bros. Bldg. Co., 1927, 170 Minn. 373, 212 N.W. 602, 56 A.L.R. 1; Bryant v. Troutman, *745 Ky.Ct. of App.1956, 287 S.W.2d 918. The defendant denies that any fraud or deceit was committed and, in addition, declares that the action must be considered only as one for breach of warranty, since the fraud allegation was not pleaded with the particularity required under Rule 9(b) of the Federal Rules of Civil Procedure, 28 U.S.C.A. However, in paragraph 14 of the amended complaint, the plaintiff, after detailing at length the course of conduct between the parties, alleged that the defendant's activities constituted fraud and deceit. Also, during the pretrial hearings and the trial itself, it was apparent that the plaintiff was relying upon his allegation of fraud and deceit as his primary basis for relief. Under these circumstances, a further amendment to the pleadings may be permitted to conform to the proofs. Rule 15(b).
Does California or Michigan law govern? Because of important differences in the law of fraud and deceit in these states, this question should first be examined.
California requires, as an essential ingredient of an action for fraud and deceit, a proving of "scienter," i. e., that the party making the representation knew it to be false when made or asserted its truth without reasonable grounds for so believing. California Civil Code, §§ 1572, 1573, 1709, 1710; Gagne v. Bertram, 1954, 43 Cal. 2d 481, 275 P.2d 15; Pinney & Topliff v. Chrysler Corporation, D.C.S.D.Cal.1959, 176 F. Supp. 801.
Michigan applies a similar doctrine if no contractual relationship exists between the party making the representation and the deceived party. But, if material representations which are false in fact are made by one party to another in a transaction between them, are relied upon and cause damage, and the loss inures to the benefit of the representing party, there is a cause of action for fraud and deceit, regardless of whether the party making the representations knew them to be false. Kolinski v. Solomon, 1942, 303 Mich. 710, 7 N.W.2d 117; Essenburg v. Russell, 1956, 346 Mich. 319, 78 N.W.2d 136. The rationale is that a party representing a material fact to a person in privity of contract with him is under a legal obligation to determine its truthfulness prior to making the representation. The loss sustained by the plaintiff need not necessarily coincide with a benefit received by the defendant. Aldrich v. Scribner, 1908, 154 Mich. 23, 30-31, 117 N.W. 581, 18 L.R.A.,N.S., 379.
Some representations and some alleged partial disclosures occurred in California. However, Michigan is the forum state, and some alleged misstatements by Librascope were received by mail in this state. The accepted choice-of-law rule regarding fraud and deceit is that the law of the "place of wrong" governs. Restatement of Conflict of Laws, § 377. This theory, which emphasizes the consequence to the plaintiff rather than the defendant's conduct, declares that in a fraud and deceit action the "place of wrong" is not the state where the misrepresentations are made but rather where the loss is sustained. Boulevard Airport, Inc. v. Consolidated Vultee Aircraft Corporation, D.C.E.D. Pa.1949, 85 F. Supp. 876; Goodrich, "Handbook of the Conflict of Laws," § 93 (1949). Pinpointing the place of damage could present a difficult question in certain situations, but in this case it is *746 quite clear that, if the plaintiff did part with assets in reliance upon the alleged misrepresentations, he did so in Michigan. The defendant could also foresee that any harm caused by its conduct would occur in Michigan. Rheinstein, "The Place of Wrong: A Study in the Method of Case Law," 19 Tulane Law Rev. 4, 30-1 (1944). The law of Michigan governs.
The testimony in this case is of a highly technical nature, and I find it necessary, for an understanding of the issues and applicable law, to make reference to the evidence with some particularity, even though such particularity will result in an unsually long opinion.
Late in 1954, the plaintiff was requested by Radioactive Products, Inc. to manufacture three separate "two thousand channel time analyzers" (a type of computer) for use with the neutron velocity selectors of Knolls Atomic Power Laboratory, Columbia University, and Yale University, respectively. Delivery of two computers was to be made in December, 1955. The third was to be delivered in February, 1956.
The heart of the computer that Strand agreed to build was a magnetic drum memory system. This is a cylinder carefully machined from magnesium about eight inches in diameter and four inches long, on which is deposited a very thin and uniform coating of iron oxide. The drum system works in much the same manner as the common tape recorder. The information which it stores is recorded on it magnetically by some of the read/record heads which are the subject of this action. Other read/record heads simultaneously read information, and still others read and record on other tracks along the axis of the cylinder. In addition, data read by some of the read/record heads could be transferred to punched cards for storage or further computation. Because of the simultaneous operation of many of the heads, it was necessary that the operation of any one not interfere electrically with the proper operation of the others.
On May 12, 1955, the plaintiff requested a quotation from the defendant on two hundred read/record heads and three completed drum assemblies for these three computers. The letter gave general specifications and stated that the computers were necessary for a mechanical drum system required for a "special purpose application." The letter explained:
"* * * because of the nature of the system it is important that the writing operation of any head shall not cause any cross-talk on any reading heads, either on the same track or on adjacent tracks. Furthermore, the spacing between each pair of read-write heads must be adjustable by at least plus or minus 1/32 inches. The total spacing between pairs of read-write heads is not critical but should be as small as possible, consistent with the mechanical design of the assembly."
Librascope, in May, 1955, was an experienced and dominant concern in the electronics industry. It was offering for sale certain heads, designated MH-15A, MH-10A, and MH-10P, for utilization in digital computers. Librascope also had manufactured nine prototype models of a head termed MH-10R to be used in its own computer, then in the developmental stages, known as the LGP-30. The MH-10R heads were designed to permit placement in very close physical proximity to one another. In fact, the active part of each head could be within three-tenths of an inch of the centers of the adjacent heads. However, in order *747 to accomplish this result, it was necessary to change the design of the magnetic core pieces used in the R-type head from the proven and successful design used in the A-type (and P-type) head. This change made the problem of "noise" much more severe in the new heads, as Librascope must have realized, at least from the performance of the heads in their own LGP-30 computer.
On the basis of preliminary testing of the MH-10R heads prior to their installation in the LGP-30 computer, the Librascope engineers concluded, in the spring of 1955, that the MH-10R heads were adequate for any system substantially similar to the LGP-30, the use for which Librascope had tested the heads.
Librascope at one time contemplated the furnishing of MH-10A and MH-10P heads to fulfill Strand's specifications. If an A-type and P-type head are accurately attached so that they will read or record from a single track on the drum, they will perform the recirculating functions of a pair of MH-10R heads. In the opinion of the Librascope engineering staff, under Strand's requirements one P-type and one A-type head would have been needed in each recirculating line. However, this suggestion was rejected by Librascope, and on June 21, 1955, the newly developed MH-10R head was specified and a price quoted by Librascope. At the trial the quotation was explained by Librascope on this basis: The MH-10P head had special mounting which occupied more space than the MH-10R head; A-type heads also could not be placed as closely together as the R-type heads. Finally, the cost element was important. In 1955, the A-type head cost approximately $30; the P-type, $80 to $100, while the price quoted for each MH-10R head was $32.50. The quotation, on the back of which appeared the express warranty set forth above, was for the MH-10R heads even though it was known at that time that those heads, under comparable circumstances, had a signal-to-noise ratio lower or poorer by some margin than A-type heads. In a letter mailed the same day as the quotation, Librascope declared that the MH-10R heads were of a different construction from the MH-10A and MH-15A heads but did not state that, from their experience in the operation of the LGP-30, the adoption of the new construction actually resulted in increased "noise" from the MH-10R heads.
On June 23, 1955, the plaintiff sent a purchase order, which made reference to the quotation of June 21, 1955, to Librascope for two hundred MH-10R heads at the quoted price. Strand had decided to build the drum assemblies himself, using these heads, rather than buy the whole assembly from Librascope. One sample head was to be sent immediately, seventy others in the next sixty days, and the balance in one hundred twenty days.
Strand's order was the first ever filled by Librascope for MH-10R read/record heads.
Librascope, on June 24, 1955, sent one MH-10R head to Strand as a loan only to be returned upon the request of Librascope, or upon the completion of Strand's testing. At this time, Librascope still had in its possession only the nine prototype MH-10R heads from its model shop maufacturing facility.
Because the head loaned to Strand in June was not functioning properly, Librascope replaced it with another sample head on August 12, 1955. Strand immediately tested the new head for output voltage (signal strength), found this satisfactory, and so notified the defendant. *748 Strand then commenced more extensive testing of the new head and discovered excessive "noise" and "shoulder effects" (a type of distortion) in the output of the head. Strand also noticed variations in the amplitude of the signals. Personnel from Strand communicated with Librascope officials, who stated that, as far as Librascope could determine, the head was satisfactory, and suggested the possible presence of a mistake in the building of the mechanical drum system or associated circuitry.
Because the "noise" problem was not solved through further experimentation, and numerous telephone conversations with Librascope being of no avail, Strand, on September 15, 1955, went to Librascope's offices in California. At the time of the trip, he was "not suspicious of the Librascope heads" but felt, as a result of the assurance of the Librascope engineering staff, that something in the overall system was incorrect, perhaps the circuitry, the oxide, or the drum.
Strand took along the sample head he had been testing. Librascope examined the head with its equipment, and the test showed the head to be functioning within specifications. The wave appeared perfectly symmetrical upon the oscilloscope. This examination was different from the test performed by Strand in Ann Arbor, for the test device used by Librascope was not designed to demonstrate the absence or presence of "noise." James L. Cass, the project engineer responsible for the development of the LGP-30 computer, stated that he did not know whether Strand was informed of the nature of the testing procedure and its limitations. According to Cass, Strand had merely referred to a general "noise" problem. Strand testified, however, that although he did not describe specific kinds of "noise" he told Cass of the presence of variations in amplitude and "shoulders." In any event, it is clear that the meaning of the test was not discussed.
Cass testified that he could not recall whether he specifically informed Strand that the MH-10R heads were or were not known to be a source of "noise," although he realized the heads were responsible for some "spurious signals." He regarded this as a typical occurrence in the developmental stages of a complex computer.
Strand was told by Librascope engineers that the MH-10R head would be used by Librascope on the LGP-30 computer scheduled for completion in December, 1955. Responsible representatives of Librascope assured Strand that he would have no trouble with the head because Librascope was relying on the MH-10R head for its own LGP-30 computer.
Since Librascope personnel had suggested that his difficulties were probably caused by his circuitry or oxide coating, Strand asked for examples of the circuitry used by Librascope. Cass replied that this information was "proprietary" and would not be furnished. Strand was allowed to see the circuitry on an etching board, but he was given only a brief period within which to examine it and was therefore unable to determine the actual circuitry used. Strand was also informed that information about the oxide coating applied to the drum was likewise "proprietary."
Cass testified that he informed Strand at that time that the clipping level on the LGP-30 was as high as 50 to 75 per cent. Strand testified to the contrary, stating that he first learned that heads sold to his company were being installed *749 in a computer with a 50 to 75 per cent clipping level during the course of this trial. He testified that information about the clipping level was precisely the data he was seeking when he went to Librascope, for a 50 to 75 per cent clipping level is not standard in a large digital computer. A 15 to 30 per cent level is customary.
Edwin M. Couleur, an engineer attached to Librascope's sales department in 1955, testified that, at the time of Strand's visit, Cass declared that, because Strand did not give Librascope the entire order for the drum assemblies, "he could go fish for himself" insofar as any help was concerned.
Following his return from California, Strand undertook an extensive series of investigations to determine the cause of the "noise." He made the testing apparatus more elaborate by mounting a pair of heads (loaned to him at that time by Librascope for the purpose) in a single track to put together a recirculating loop. His staff examined circuitry, the effects of oxide thickness on the drum, and the mechanical positioning of the heads, all to no avail.
In November, 1955, at a time when only ten of the two hundred heads of Strand's order had been shipped to Ann Arbor, two important events occurred. The LGP-30 computer, containing seven MH-10R heads, failed to function correctly at the International Automation Exhibition in Chicago on November 15, 1955. In fact, an LGP-30 computer did not work satisfactorily until March, 1956, and the computer was not offered for sale to the general public until the fall of 1956.
On November 21, 1955, William Reinholtz, the engineer responsible for the construction of the MH-10R head and one of the engineers who discussed Strand's difficulties with him during his visit to California in September, 1955, suggested to Librascope officials that the "noise" and "cross-talk" in the MH-10R heads might be eliminated by grinding the tips of the pole pieces of the head at two places in a certain manner.
After some experimentation, and after the solution of the problems involved in translating Reinholtz' ideas into production designs, an order was prepared by Librascope directing the incorporation of the new designs into the production of MH-10R heads. The engineering order was approved on the condition that it was not to apply to Librascope's internal job order number 2138. This order was issued on June 24, 1955, and included the entire order for Strand and fifty other heads destined either for sale to other customers or for internal use by Librascope.
Cass testified that the Strand production order was excluded from the production of the modified heads because Strand had been pressing for immediate delivery of his order, and Librascope had no knowledge that the heads shipped to Strand were unsatisfactory. He also said that it was very difficult to halt a job lot during the production run, even though the grinding change was the final operation to change the heads physically.
Meanwhile, in Ann Arbor, Strand was receiving some of the unmodified MH-10R heads and was spending a great deal of time and effort in trying to make his computer work with them. It was not until February, 1956, that he was able definitely to establish that the difficulty was caused by the heads. By this time, (in fact, as early as September, 1955) it was too late to install heads of any other design without completely rebuilding the computer. The apparatus for physically mounting the heads was such that no other heads (either from Librascope or from other manufacturers) could be made to fit, and Strand still was under the impression that the MH-10R head was functioning as well as it could be made to do.
At this point, Strand had tried 30 per cent and 50 per cent clipping circuits*750 without success. In fact, even with 50 per cent clipping, the computer would not operate for more than a minute. He then tried a circuit that threw away 80 per cent of the signal coming from the heads. This clipping circuit was completed and installed in the computer in early April, 1956, but was also not very successful.
By April 12, 1956, Strand had received the two hundred heads in the order and still made no written complaint to Librascope, although he had commented orally to Librascope officials about what he assumed was his failure to obtain satisfactory results from the circuitry or oxide coating on the drum. Strand made payments on January 3, February 13, February 22, and March 5, but did not pay for the shipments of heads invoiced March 14 and April 12.
The first formal request by Strand to substitute any heads was made in June, 1956. He asked replacements for twelve completely inoperable heads. The replacements gave a far better signal, with lower signal-to-noise ratio, than did the best of the original lot of two hundred heads. Strand did not notify Librascope of the excellent quality of the new heads until July because the company was on a "crash program" to complete the first computer, which was already several months behind schedule on its delivery date.
On July 24, 1956, Strand telephoned Couleur at the office of his new employer (Couleur's employment relationship with Librascope terminated on July 13, 1956) and stated that the replacement heads had performed satisfactorily and were, in fact, the first Librascope heads delivered to him which fulfilled his specifications in every respect. Couleur at that point disclosed that Cass and Reinholtz had not considered the MH-10R head, as originally constructed, to be suitable for use with the LGP-30 computer. Couleur told Strand to call Richard Hastings, the supervisor of the Librascope commercial department, to discuss obtaining replacement heads.
Couleur also testified that he had told Hastings in May, 1956, that Strand was having trouble with his heads and that, if the modified heads operated satisfactorily, Strand's MH-10R heads should be exchanged for the newer heads. According to Couleur, Hastings just stated a terse "thank you" and ushered him out of the office.
Strand attempted to call Hastings at Librascope, but Hastings was unavailable, and he talked instead to a sales engineer. Strand told him that the twelve replacement heads were "perfectly satisfactory," that the original two hundred were not satisfactory, and that he hoped an equitable solution could be reached between the parties. Subsequent to this telephone conversation, an exchange of letters occurred between Librascope and Strand. Librascope offered a substantial price preference to Strand in procurement of replacements but refused to acknowledge that the original heads were deficient.
Strand did not make a formal and express demand for satisfactory read/record heads until October 19, 1956, in a letter in which he also stated that he intended to hold the defendant fully responsible for all losses sustained because of the defective heads. On October 24, 1956, Librascope replied that the record of the transaction disclosed no justification whatever for Strand's demand and revoked its offer to sell replacement heads at a substantial discount.
Dr. Max L. Yeater, professor of physics at the Rensselear Polytechnic Institute, was a physicist at the Knolls Atomic Power Laboratory in Schenectady, New York, in 1955 and 1956. He, together with colleagues, conceived the plans for the two thousand channel time analyzer.
*751 On January 10, 1957, Dr. Yeater wrote the president of Nucor Research Inc., the successor corporation to Radioactive Products, Inc., that the magnetic drum systems supplied by Nucor and Strand had received extensive tests at the Knolls Atomic Power Laboratory and at Columbia University and that the tests were quite discouraging. He stated that, despite the elaborate clipping circuits which were added to the system to combat the difficulties, it was not possible to maintain the critical adjustments required to derive usable signals from the head. He concluded that the read/record heads were defective and failed to meet specifications and so would have to be replaced. He further stated in the letter that the original circuitry appeared to be excellent but that the additional clipping circuits did not correct the signal from the heads to the extent that the system could be used for its intended application.
At the time Dr. Yeater wrote his letter of January 10, 1957, neither of the two computers already delivered had received the modified heads, with the exception of eleven of the twelve replacement heads installed in the first computer.
Following Dr. Yeater's letter, Nucor Research, Inc. ordered enough heads with the new grinds from Librascope to outfit completely the three computers. The heads were delivered to Strand and installed in all three computers. With the new heads, Strand was able to modify the clipping circuits to a level of about 20 to 30 per cent. The computers then operated for many hours without error, to the complete satisfaction of the users.
Although Nucor ordered the new heads, it felt it was Strand's obligation to deliver the computers in satisfactory operating condition. Therefore, the company billed Strand for the replacement heads.
An LGP-30 computer was demonstrated in the courtroom. Tests were made upon some of the heads originally delivered to Strand and used by him in his computer, purportedly to show that they would, in fact, operate in the computer. There is doubt as to whether such tests may be considered as evidence:
"Experiments in Court. Experiments to show the quality or operation of a substance, a machine, etc. are often excluded because of the dissimilarity of circumstances or because of probable confusion of issues; and for this reason the exhibition of such experiments before the tribunal may of course be forbidden." 4 Wigmore on Evidence, § 1154a, (3rd Ed.1940).
However, since I conclude that the demonstration was not a valid test of the machine or the heads, I need not rule on it as a matter of admissibility.
In the LGP-30 computer, there are three pairs of MH-10R heads, but in the Strand computer, there are sixty such heads. For this reason, Dr. Yeater stated that the demonstration in the courtroom would not be a fair test of the heads. He testified that, in order to simulate an ordinary half-hour test on the Strand analyzer, the LGP-30 computer would have to operate many weeks. It was Dr. Yeater's expressed view that, even though the heads tested in the courtroom may have worked in the single recirculating line of the LGP-30 in which they were tested, it would not indicate that the same heads would be satisfactory in Strand's more complex computers.
Dr. Norman R. Scott, professor of electrical engineering at the University of Michigan and an instructor and consultant in the area of digital computers, testified that the demonstration conducted on the LGP-30 computer in the courtroom was less a test of the heads themselves than of the entire system in which the heads were imbedded. He explained that to obtain an adequate test the examiner must look for the worst possible configuration which is likely to arise. If the system is not perfect, given a long enough period of operation, it will fail, and the probability of failure rises as the number of components in the system increases. Dr. Scott testified that a single track on the computer designed by Strand could operate at a possible 7,200,000 *752 "alterations" per minute while only about 10,000 "alterations" per minute were present on the single recirculating track demonstrated on the machine in the courtroom.
Thus, the courtroom demonstration would have to operate for 720 minutes, or 12 hours, to show that it would be merely possible for the Strand computer to use the head being tested for one minute. And to show that the head would operate reliably for a period of one minute in the Strand computer, the courtroom demonstration would have to run for a number of days without failing. Further, since the proper operation of the computer requires that all sixty of the heads in the recirculating lines function correctly simultaneously, even the worst of the heads in it must meet a still far more stringent reliability test. As summarized by Dr. Scott: "In other words, I feel it is rather dangerous to try to extrapolate from these tests and to conclude that, since the heads worked a few times with a given problem, they are perfectly satisfactory in general."
It is quite clear that the courtroom demonstration, though an interesting display of computer technology, was by no means a valid test of the operation of the MH-10R heads in the Strand computer, the use for which they were intended. It is also abundantly clear that these heads would not work in the Strand computer since at best, with severe clipping and very delicate adjustment, it would work for only short periods of time, and then not satisfactorily.
The court concludes that an examination of the entire transaction between Strand and Librascope demonstrates beyond doubt that the MH-10R heads supplied Strand by his order of June 23, 1955, were defective in their design and construction, in that they were not suitable for his intended use.
The first issue to be considered in determining if fraud in the transaction was present under Michigan law is whether a material misrepresentation was made to the plaintiff by Librascope.
In the area of electronic digital computers, research and development are advancing at an almost unbelievable pace. Research, development, and manufacturing companies necessarily have certain complex and technical information within their own exclusive possession. When Strand made inquiry concerning the availability of read/record heads to fit his specifications, he relied upon the superior knowledge and expertise of Librascope, as Librascope well knew. The transaction entered into between Librascope and Strand did not thus present the classic common-law situation of parties "bargaining at arms length." The facts vital to Strand's needs were exclusively and peculiarly within the possession of Librascopefacts which, if disclosed to Strand, might have dissuaded him from entering into the transaction. Furthermore, in this case, greater reliance was placed upon the knowledge, honesty, and good faith of the manufacturer than in the ordinary contractual relationship, since the potential purchaser was not in a position to make a prior investigation of the product.
Under these circumstances, the law did not compel Librascope to divulge all its confidential or proprietary information to Strand, but standards of fair dealing did require a complete disclosure of the state of the development of this new product, since such disclosure was necessary *753 to prevent Strand from drawing conceptions or inferences known by Librascope to be unwarranted.
Many statements made by Librascope to Strand during June, 1955, were materially misleading because he was not informed of unfavorable aspects importantly qualifying these statements. In the letter to Strand on June 21, 1955, Librascope failed to state that the new construction of the MH-10R heads created "noise" problems. Also, Librascope warranted the heads to be free from defects, which would presumably include "noise," without disclosing that such representations were based upon incomplete testing of the LGP-30 computer, which contained only seven MH-10R heads in contrast to the more than sixty in each Strand computer. Strand was not told that a combination of A-type heads and P-type heads would have been suitable. He should have been given the opportunity to compare the increased cost of utilizing heads known to function adequately against the risks inherent in ordering a newly developed, but less expensive, product. The failure to disclose an alternative known to be reliable is, under the circumstances in this case, tantamount to a misrepresentation. 3 Restatement of Torts, § 529 Comment a; Pohl v. Mills, 1933, 218 Cal. 641, 24 P.2d 476. Enough has already been established to hold that, to avoid misunderstandings, fair dealing made it incumbent upon Librascope to inform Strand that his order was the first one placed for the new MH-10R heads and that these heads were experimental and were still being tested in conjunction with the prototype model LGP-30 computer.
In Michigan, even without a fiduciary relationship, a party is under a duty to use diligence in making a complete disclosure of facts where partial disclosure may convey false impressions and mislead the plaintiff. Such half-truths or non-disclosures are considered to be concealment of facts and, therefore, misrepresentations. Groening v. Opsata, 1948, 323 Mich. 73, 34 N.W.2d 560. See also Equitable Life Insurance Co. of Iowa v. Halsey, Stuart & Co., 1941, 312 U.S. 410, 425-426, 61 S. Ct. 623, 85 L. Ed. 920; 1 Harper and James, "The Law of Torts" pp. 586-8 (1956). This does not mean that a company engaged in research and development must always completely disclose all facts concerning its products. However, under the singular circumstances of this case, where reliance was placed upon the expertise of an established concern that had the required information in its exclusive possession, the manufacturer was under an obligation to indicate the state of development of the new product in order to avoid misleading impressions created by its partial disclosure and its unqualified statements.
During Strand's visit to California in September, 1955, Librascope's incomplete testing device, coupled with the positive assurances from responsible Librascope officials that the head was satisfactory and that the fault existed in another area, created the false impression that the Librascope product was satisfactory, at the very time the heads were known to be a source of "spurious signals." This course of conduct constitutes a material misrepresentation.
I believe the testimony of Strand that Cass did not tell him at any time that the LGP-30 computer was being operated at a 50 to 75 per cent clipping level. If Cass' testimony that this information was given to Strand is correct, it would seem ludicrous for Strand to return to Ann Arbor and waste nearly six months of precious time in designing and constructing inadequate clipping levels of 30 and 50 per cent. Had Librascope officials told Strand, after assuring him that the MH-10R head was satisfactory because it was being installed in the LGP-30 computer, that the LGP-30 computer had a clipping level of 50 to 75 per cent, Strand would have undoubtedly realized that the unsatisfactory results he was obtaining *754 were largely due to the uncorrected MH-10R head.
The failure to disclose the information about the LGP-30 clipping level, in the context of the representations made to Strand, was a deliberate concealment of a material fact.
The contract between Strand and Librascope had been executed. However, none of the heads had been delivered to Strand for installation in the computers, and he was still in a position to demand heads of the necessary quality. Therefore, such misrepresentations, although not inducing the execution of the contract, were material in preventing Strand from returning the heads within the time period specified in the agreement, and in addition, were responsible for much of the cost incurred by Strand in attempting to make his computer work properly.
It has been contended that the statements of Librascope officials that the heads were satisfactory and that the fault existed with the circuitry or oxide coating were expressions of opinion upon which no reasonable reliance may be based. This is not the place to discuss the differences between the talismanic words "opinion" and "fact." It seems to me that the statements of Librascope were intended to represent the facts and were not in the format of an opinion. However, even if they were meant to be "opinions," Librascope held itself out as an expert, had exclusive, superior knowledge, and implied either that it knew facts to justify the opinions or knew of no facts to dispute them. Under these circumstances, it was reasonable for Strand to rely upon Librascope's statements as representations of facts which were not inconsistent with the expressed opinions. Gagne v. Bertran, 1954, 43 Cal. 2d 481, 275 P.2d 15; Kefuss v. Whitley, 1922, 220 Mich. 67, 189 N.W. 76; Fensom v. Rabb, 1950, 190 Va. 788, 58 S.E.2d 18; Seavey, "Caveat Emptor as of 1960," 38 Tex.Law Rev. 439, 442 (1960); Keaton, "Fraud: Misrepresentations of Opinion," 21 Minn.Law Rev. 643, 647-8, 652 (1937). The statements by Librascope were certainly not within the realm of "puffing" or "trade talk" upon which a purchaser may not rely. See Hayes Construction Company v. Silverthorn, 1955, 343 Mich. 421, 426, 72 N.W.2d 190.
The general rule is that a party to a business transaction is under an obligation to exercise reasonable care to disclose to the other party, before the transaction is consummated, any subsequently acquired information which he recognizes as rendering untrue, or misleading, previous representations which, when made, were true or believed to be true. 3 Restatement of Torts, § 551(2) (b); Incledon v. Watson, 2 F. & F. 841, 175 Eng.Rep. 1312 (1862); Hush v. Reaugh, D.C.E.D.Ill.1938, 23 F. Supp. 646, 652; 1 Harper and James, "The Law of Torts" p. 589 (1956). It seems to me that the doctrine of continuing representations should also govern a situation where the contract has been entered into but substantial performance consisting of shipments or payments is in futuro. In this case, at the time when only ten of the two hundred heads comprising Strand's order had been shipped to Ann Arbor, the LGP-30 computer failed to function properly at the automation exposition, and the "noise" problem in the R-type heads was, in effect, eliminated by Reinholtz' improvement. At this point, a duty was imposed upon Librascope to inform Strand that the source of the difficulty with the heads had been discovered and eliminated and that its representations in September, 1955, that the LGP-30 computer operated satisfactorily and that the "noise" in Strand's computer was due to circuitry or oxide coating, were incorrect. Had he been given this information, Strand could have returned the defective heads and would not have expended additional time and effort in chasing the "will-o'-the-wisp" of circuitry and oxide coating and would have been able to deliver his computers at a time nearer to the specified contract dates.
Under the Michigan doctrine of innocent misrepresentations, there is no requirement for an intent to deceive. *755 The intent required is to induce the action, Rosenberg v. Cyrowski, 1924, 227 Mich. 508, 511, 198 N.W. 905, an element certainly present in this action.
In Michigan, in order to sustain an action for misrepresentation, the plaintiff must show that the misrepresentations were material, as measured by an objective standard, and that he did, in fact, rely upon them. Hall v. Johnson, 1879, 41 Mich. 286, 2 N.W. 55. The express misrepresentations, concealments, and partial disclosures in this case were material. Strand had stated in his letter of inquiry that he wished to avoid "cross-talk." The presence of "noise" necessitating a high clipping level could ruin the value of the entire computer. Strand relied upon the misleading representations of Librascope. He testified that, had he known in May, 1955, of "noise" problems with the MH-10R heads, he would have attempted to secure either a different combination of read/record heads or heads from another producer. He also stated that, had he been told of the 50 to 75 per cent clipping level used in the LGP-30 computer, he would have realized immediately what apparatus was required in his computer, and, had he discovered that Librascope had eliminated the "noise" problem in the MH-10R heads, he would have demanded the modified heads. The misrepresentations were intentional, and, since they were not so incredulous as to place Strand on notice, his failure to discover their falsity is no defense. Salvati v. Cusolito, 1950, 98 Cal. App. 2d 582, 220 P.2d 800; Smith v. Werkheiser, 1908, 152 Mich. 177, 115 N.W. 964, 15 L.R.A., N.S., 1092. In addition, Librascope's misrepresentations, since they were relied upon by Strand, actually prevented Strand from determining that the MH-10R heads were the cause of the "noise," Strand only discovered that fact after he had exhausted the investigatory procedures suggested by Librascope.
Librascope contends that, as early as December, 1955, and certainly no later than February, 1956, employees of the Strand Engineering Company concluded that the heads were functioning improperly. Since Strand continued after those dates to receive the read/record heads and to make concurrent payments, Librascope urges that this factor indicates either that Strand had not relied upon the assurances of Librascope officials during the period of May through September, 1955, or that Strand waived any possible claim of fraud. The Michigan Supreme Court has stated that, in an action for damages, fraud is not waived by continuing payments after the discovery of the fraud. Elson v. Harris, 1959, 356 Mich. 175, 96 N.W.2d 767; Nowicki v. Podgorski, 1960, 359 Mich. 18, 101 N.W.2d 371. In addition, it should be remembered that the design of the drum assembly for the Strand computer had been "frozen," and it would have been prohibitively expensive for Strand to redesign its entire computer in order to utilize heads from another manufacturer. Strand, in February, 1956, was behind in his contract delivery date and, in addition, had no knowledge that Librascope had discovered the shortcomings of its MH-10R head and could correct them. Finally, in February, 1956, Strand's employees had not yet determined that it was impossible to eliminate the "noise" by use of more severe clipping.
The misrepresentations by Librascope's employees were made during the course of their employment with actual or apparent authority and would also bind the defendant even though the employees were not expressly authorized to make such statements. 1 Restatement of Agency 2d, § 265. Moynes v. Applebaum, 1922, 218 Mich. 198, 187 N.W. 241; Groening v. Opsata, 1948, 323 Mich. 73, 34 N.W.2d 560.
*756 Fraud will neither be presumed nor lightly inferred but must be proved by a preponderance of the evidence. However, it is sometimes stated that fraud must be established by "clear and convincing" evidence, and this phrase, or one of like import, has been held to specify a higher standard of proof than required by the doctrine of the preponderance of the evidence. 9 Wigmore on Evidence, § 2498 (3rd Ed.1940). The Michigan Supreme Court has ruled in some cases that the burden is on the asserting party to establish fraud by a "preponderance of the evidence," in other decisions by "clear and satisfactory proofs." Compare, for example, Kirk v. Vaccaro, 1955, 344 Mich. 226, 231, 73 N.W.2d 871; Goodrich v. Waller, 1946, 314 Mich. 456, 461, 22 N.W.2d 862; and Fahey v. Pell, 1945, 310 Mich. 280, 281, 17 N.W.2d 183, with A & M Land Development Co. v. Miller, 1959, 354 Mich. 681, 686, 94 N.W.2d 197, and Groening v. Opsata, 1948, 323 Mich. 73, 77, 34 N.W.2d 560.
It seems to me that the phrase "clear and satisfactory" as applied to the quantum of proof of fraud merely indicates that fraud, which ordinarily must be proved by circumstantial evidence, cannot be founded upon doubtful, uncertain, or conjectural evidence. Both the "preponderance" and "clear and satisfactory" phrases appear in the Michigan decisions without any indication of varying meanings. However, it is not necessary to hold that the "preponderance" doctrine prevails because, under either that standard or the "clear and satisfactory proofs" criterion, I hold that the plaintiff has sustained the burden in this action of establishing fraud by the defendant and that Strand sustained damages from the purchase of defective heads.
Prior to the trial, an order was entered, pursuant to Rule 42(b) of the Federal Rules of Civil Procedure, reserving the question of the extent of damages for a subsequent hearing. Within ten days, the court will notify counsel of an appointment for such a hearing.
An appropriate order may be presented for signature.
 Librascope, Incorporated was a subsidiary of General Precision Equipment Corporation for nineteen years prior to January 4, 1960. On that date, Librascope and three other subsidiaries of the same parent were merged into a new corporation called General Precision, Inc., a Delaware corporation, which is itself a subsidiary of the same parent, General Precision Equipment Corporation. All liabilities of Librascope were assumed by the new corporation. Although an amendment permitting the name change was granted during the trial, for convenience and uniformity, the defendant will be designated "Librascope" throughout this opinion.
 Historically, the action on the case for deceit afforded a remedy for many contract breaches. Not until Palsey v. Freeman, 3 Term.Rep. 51, 100 Eng.Rep. 450 (1789), was deceit recognized as a distinct tort and definitely disassociated from actions for breach of warranty. Thereafter, the two lines of recovery slowly diverged. Ames, "The History of Assumpsit," 2 Harv.Law Rev. 1, 8, 10-15 (1888); Prosser, "Handbook of the Law of Torts," 522-3 (2nd Ed. 1955). In Michigan, when alleging fraud or deceit, a plaintiff still may bring an action either of assumpsit or trespass on the case. 20 M.S.A. § 27.651, Comp.Laws 1948, § 611.1.
 This doctrine also prevails in several other states. See 23 Am.Jur. § 132 (1939).
 Such interference is known as "cross-talk." It may be understood by considering a "write" head as a miniature radio transmitter putting out a signal which is picked up by a "read" head. The signal so picked up will interfere with and distort the information which the "read" head is reading from its own track. "Cross-talk" is the term describing this particular type of disturbance or effect and also describing the unwanted signal (or "noise") resulting therefrom.
 See note 4, supra.
 "Recirculating functions" here means the reading of information from one point on a track by a "read" head, and the rewriting of the information (after modification) by a "record" head at precisely the same point on the rotating drum, slightly later. This requires that the distance between the two heads on the given track be adjustable with exactitude in order to insure that the "record" head is over the correct point on the drum at the moment the head writes a given signal.
 A clipping circuit is a specially designed and constructed circuit which eliminates the smaller signals coming from the heads. This, in effect, throws away "noise" if it is smaller than the desired signals. Clipping is a last resort to improve a signal contaminated by noise and is a palliative rather than a removal of the original cause of the trouble.
The percentage of the clipping level indicates that that proportion of the signal is excised. Thus, on a 50 to 75 per cent clipping level, one-half to three-quarters of the signal is removed.
 See note 6, supra.
 See note 7, supra.
 As the clipping level is increased, more of the signal is discarded, together with the "noise," and the performance thereby becomes more marginal.
 An "alteration" as here used means one reading operation of a "read" head plus the recording operation of the corresponding "write" head. Since there are 2000 "bits" of information to be read and recorded on each track of the Strand computer, and the drum on which the tracks are recorded revolves at 3600 revolutions per minute, there are 2000 times 3600, or 7,200,000 alterations per track per minute. There are thirty recirculating tracks on the drum; hence, 216,000,000 alterations per minute in the operation of the computer. This amounts to 432,000,000 head operations which must occur without error for the computer to operate one minute without failing.
 But see Dean Prosser's conclusion: "The law appears to be working to the ultimate conclusion that full disclosure of all material facts must be made whenever elementary fair conduct demands it." Prosser, "Handbook of the Law of Torts" p. 535 (2nd Ed. 1955).
 Even in jurisdictions requiring "scienter," the law seems to be tending to the doctrine that the intent needed is to induce action. See Gagne v. Bertran, 1954, 43 Cal. 2d 481, 275 P.2d 15; 1 Harper and James, "The Law of Torts" p. 533 (1956).