OPINION OF THE COURT

A. LEON HIGGINBOTHAM, Jr., Circuit Judge.

INTRODUCTION

This is an appeal pursuant to 28 U.S.C. § 1292(a)(1) (1982) from an interlocutory order of the district court preliminarily enjoining the use and disclosure of certain trade secrets. Our scope of review is narrow. On a motion for a preliminary injunction "[u]nless the trial court abuses [its] discretion, commits an obvious error in applying the law, or makes a serious mistake in considering the proof, the appellate court must take the judgment of the trial court as presumptively correct." A.O. Smith Corp. v. FTC, 530 F.2d 515, 525 (3d Cir. 1976). Nonetheless, because we find that some of the district court's conclusions are without support in the applicable law of trade secrets, and that in some respects its order is so overbroad and vague as to constitute an abuse of discretion, we will vacate the order and remand for reformulation of the preliminary injunction.

I. [4] BACKGROUND

The unusually voluminous record upon which this preliminary injunction was issued includes nineteen days of testimony and well over 100 exhibits. Much of this material is of a specialized, technical nature. We will summarize the evidence in as much detail as is necessary to properly frame the numerous issues raised in this appeal, but with due regard for SI Handling Systems, Inc.'s proprietary claims.

A. SI Handling and CARTRAC

Appellee SI Handling Systems, Inc. ("SI"), founded in 1958 by its current Chairman and Chief Executive Officer L. Jack Bradt, is a Pennsylvania corporation with headquarters and principal manufacturing facilities located in Easton, Pennsylvania. SI employs approximately 300 persons and had sales of 20 million dollars in the fiscal year ended February 27, 1983. Through a number of subsidiaries and licensees SI's products are sold in much of the industrialized world.

SI is in the business of designing, manufacturing, and installing "materials handling systems". "Materials handling" is a generic term describing the transportation of materials, by any mechanized means, between locations in a factory or warehouse. Forklift trucks and conveyor belts are familiar examples of materials handling devices. A materials handling "system" connotes a combination of devices or components designed to integrate a number of warehouse or factory operations, in order to achieve greater automation and efficiency. SI, which at the outset made only manually-operated steel pushcarts, is today an industry leader with four sophisticated, highly automated product lines, each of which possesses the flexibility to be custom-designed for diverse systems applications.

This litigation involves only one of SI's product lines, known by the trade name "CARTRAC". CARTRAC was initially developed by a Swedish company which had limited success in marketing the product for light manufacturing applications during the 1960's. In 1971 SI purchased the worldwide rights to CARTRAC for 1.2 million dollars. SI's strategy with regard to acquisitions such as this is to identify products that can be further developed to meet market demands in a manner that secures SI a "unique proprietary advantage". Central to this strategy is the availability of patent or trade secret protection for SI's developments.

A somewhat detailed description of CARTRAC is helpful to understanding the issues raised in this appeal. CARTRAC is generically described as a "car-on-track" materials handling system. The track is a simple pair of steel rails, resembling railroad tracks. Materials are placed upon a carrier — the "car" — which transports them along the track. Propulsion for the car is provided by a cylindrical drive tube that is mounted in between the two rails and parallel to them. The car engages the tube via a urethane drive wheel mounted on the underside of the car in a pivoting, spring-loaded housing. When the tube is caused to spin by a drive belt connected to an electric motor, it imparts a force (or "thrust") to the drive wheel and causes it to turn. If the drive wheel's motion is perpendicular to the track the car cannot move and the energy expended is simply dissipated in spinning the drive wheel. If, however, the drive wheel is turned at an angle, a component of the thrust will be imparted in the direction parallel to the rails, thus enabling the car to move along the track. The car will accelerate as the angle of the drive wheel is increased, reaching a maximum velocity at 45 degrees. SI does not claim that these basic principles of CARTRAC propulsion are trade secrets.

SI's method of propelling the car along the track gives CARTRAC a number of capabilities not shared by other car-on-track systems. Among the advantages of the spinning tube approach are the capacity to operate different cars at different speeds at different points in the system ("asynchronous" operation), and to accelerate, decelerate, or stop an individual car at various points in the system with great precision and reliability.

Another feature, especially important in the context of this litigation, is the ability of the cars to automatically "accumulate": that is, to line up between work stations. Thus, where one work station in an integrated operation is shut down, the accumulated cars feeding materials to other stations will provide a ready bank of materials until the stalled station is again operational. Accumulation permits work stations "upstream" or "downstream" from the affected station to operate more or less continuously. Automatic "car-to-car" accumulation is accomplished in CARTRAC systems through the placement of devices at the front of each car that reduce the angle of the drive wheel as it approaches the rear of a stopped car. This method of accumulation utilizes the only significant patent that SI currently holds with regard to CARTRAC, the so-called "Jacoby" patent (No. 3,818,837).

In the years following SI's purchase of the rights to CARTRAC, SI further developed and refined the product for a variety of industrial applications. SI entered into a number of licensing agreements which provided for, among other things, the exchange of technical information relating to any improvements in the design or manufacture of CARTRAC. The most important of these agreements, in the context of this litigation, was the one SI entered into with Ishihawajima-Harima Heavy Industries Co., Ltd. ("IHI") of Japan. IHI succeeded in selling CARTRAC to Nissan Motors for use in its highly roboticized automobile assembly plants. Substantial technical development was necessary to bring CARTRAC to the point where it could interact effectively with robots, interaction that requires the capability to move very heavy loads (such as car bodies) at very high speeds and to stop them at very precise locations. By 1976 SI had begun to study the market for "automotive CARTRAC" in the United States. In 1977 an extensive transfer of technology from IHI began, and in conjunction with Unimation, Inc., a leading robot manufacturer, SI sold several CARTRAC systems to Chrysler.

Beginning in 1978, "SI mounted a campaign to convince General Motors that CARTRAC could provide the automated system necessary for it to retool and meet the Japanese automotive challenge." SI Handling Systems v. Heisley, 581 F. Supp. 1553, 1558 (E.D.Pa. 1984). This campaign included intensive engineering and sales efforts. In January 1981, after several small purchases by GM, the companies consummated what has been referred to as "the big buy" — an order for eight CARTRAC systems at a price of more than 17 million dollars. Since then SI has continued to sell CARTRAC systems to GM, and continues to view the GM market as its outstanding business opportunity.

Among the specific projects SI has undertaken to meet GM requirements is the development of inexpensive "two-way accumulation" cars. Such cars are desirable where a factory operation requires that cars be sent back and forth between two work stations to be loaded and unloaded cyclically. Two-way accumulation permits a bank of cars to "stack up" at both stations, without the necessity of turning the cars around so that they can accumulate in the "car-to-car" manner we have previously described. By mid-1981 GM had purchased three CARTRAC systems using two-way accumulation, but were dissatisfied with the price of $400,000. Though SI gave high priority to solving this problem, it was not until April 11, 1983, after this litigation was commenced, that SI was able to offer GM a two-way system at a price ($200,000) that was competitive with other methods of performing this particular materials handling task. Two other development projects that SI has undertaken at GM's behest figure in this litigation: "buffer" systems and "tugger" systems. Each of these systems would involve the use of a spinning tube car-on-track conveyor to solve a current GM materials handling problem.

SI does not place any restrictions on how purchasers of CARTRAC use it, and it has publicized the basic principles and capabilities of the product extensively in trade shows, articles, and advertisements. With regard to SI's confidentiality policies, the district court found:

SI has taken the usual and reasonable precautions at its Easton facility to preserve and protect the confidential nature of its development of the CARTRAC product. Entry into the building is limited and visitors are restricted. All visitors must have passes and be escorted throughout the operation when they are inside the building. Alarm systems are utilized when the business is closed. Certain documents of a highly proprietary nature are marked accordingly. Files containing sensitive documents are locked. Drawings are prestamped with a proprietary legend. In addition, the majority of employees at SI are required to sign an Employee Agreement which purports to limit disclosure by employees of confidential information. Licensees of SI are duly proscribed from disclosure of the licensed designs and processes received via their respective license agreements.

All Sales proposals for, and maintenance manuals provided with, CARTRAC systems contain restrictive use language. SI's purchase orders to suppliers also contain a restrictive provision.

581 F. Supp. 1561-62 (footnotes omitted). The district court also found that "[i]n supplying product to General Motors SI took precautions not to give GM any more technical detail than is necessary for successful operation and maintenance of the systems." 581 F. Supp. at 1558.

SI's CARTRAC competes in the market with other sorts of materials handling systems that can do the same jobs. It appears that two Japanese companies, Murada and Daifuku, have marketed car-on-track systems that utilize spinning drive tubes, though they may differ from CARTRAC in other significant respects. There was no evidence that they have sold their systems in the United States. Thus, CARTRAC had no competition from other spinning tube car-on-track systems in the U.S. automobile manufacturing market until the advent of appellants' ROBOTRAC product.

B. The Appellants and ROBOTRAC

Appellant Michael E. Heisley was an officer of SI from 1973 to mid-1978, serving as president for nearly all of that period. In that capacity he was involved in early discussions concerning SI's entry into the automotive manufacturing market, and he traveled to Japan to study IHI's technology. Heisley left SI at about the time of the early automotive CARTRAC sales to Chrysler. After leaving SI in 1978 Heisley formed appellant Heico, Inc. ("Heico"), an Illinois corporation. Appellant Philip L. Bitely, then SI's vice-president for finance, left the company and joined Heico shortly thereafter. Bitely had been with SI since 1971, and had also been involved in the early development of automotive CARTRAC. The record indicates that Heico is a diversified company employing over 300 persons with annual sales of approximately 25 million dollars. Heisley is the chairman of Heico and Bitely is its president. There is no evidence that, prior to 1982, Heico made any effort to develop a car-on-track system.

Appellant Thomas H. Hughes joined SI in 1976 as manager of field sales. In March 1977 he was promoted to vice-president of marketing, and in January of 1979 he became vice-president in charge of the computer controls division. It appears that in the latter two positions he had significant responsibility for the development of automotive CARTRAC. In July of 1981 Hughes left SI, with a number of other controls division employees, to form appellant Sy-Con Technology, Inc. ("Sy-Con"), a subsidiary of Heico. Sy-Con is located in Easton, Pennsylvania and is in the business of providing controls software for materials handling systems.

The watershed event leading to this litigation took place on April 17, 1982. On that date chief executive officer Bradt fired SI's vice-president of operations, appellant Richard O. Dentner. Dentner, it is fair to say, was the "father" of SI's automotive CARTRAC, with responsibility for coordinating both the engineering and marketing of this product. According to Bradt, he fired Dentner reluctantly, only after efforts to resolve tensions between Dentner and other SI officers had failed.

Dentner remained as vice-president of operations until May 9, 1982, and continued to receive severance pay until September 30, 1982. During that period he also did some consulting for SI on a per diem basis. Despite this continuing relationship with SI, Dentner was already making plans, in association with Heico, to market a competing spinning tube car-on-track system, ROBOTRAC.

Dentner's efforts are well-documented by exhibits produced from Heico's files. A document dated May 1, 1982, in Dentner's handwriting, contained detailed charts projecting costs and revenues, with the legend: "The above numbers were calculated using 20% less than SI for sell, using SI cost for engineering and installation and using SI manufacturing cost + 15% for Conco profit." Also introduced was a document, parts of which are dated May 4, 1982 and bear the initials of Dentner and Bitely, that appears to be a prospectus designed to attract venture capital. Under the heading "History" it reports: "In May 1982, Dick Dentner, Vice President of Operations, left SI Handling Systems to form ROBOTRAC Inc. He became concerned that CARTRAC was becoming overpriced because of an excessive overhead structure and absorption of large costs for three other non-profitable product lines."

The prospectus goes on to state that "[a]lmost all of the senior management personnel from SI Handling, who developed Cartrac, are part of the ROBOTRAC team." These individuals were identified as Dentner, Stanley K. Gutekunst (SI's Director of Engineering), Russell H. Scheel (SI's Director of Systems Engineering), and Barry Conklin (SI's Manager of Control Systems). Gutekunst and Scheel, key players in the development of automotive CARTRAC and appellants here, did not actually leave SI to join ROBOTRAC until February 1983, and Conklin remains an SI employee.

Under the heading "Strategy", the prospectus states that "General Motors, who bought 24 systems from SI Handling Systems, will be the prime prospect for the new company. This is because; . . . The team who sold, designed and installed the system at General Motors are [sic] now with ROBOTRAC Inc." Under the heading "Patent Position" the prospectus states:

In 1978, Mike Heisley, President of SI Handling Systems, had an in-depth study performed on the patent position of the CARTRAC system. This study was performed by the SI engineering organization and Ed Gonda the SI patent attorney. The result of that study clearly indicated that the Jacoby Patent No. 3,818,837 was the only significant patent. This patent covers car-to-car accumulation, which is an inexpensive but effective way to accumulate on a CARTRAC system.

In April 1982, Stan Dalton, patent attorney with Wagner, McCord, Wood and Dalton, reviewed the patent position of CARTRAC and confirmed that the previous patent position had not been improved.

The ROBOTRAC team reviewed several alternatives to the Jacoby patent and Mr. Dalton confirmed that they would not infringe. In addition, he indicated in an April 19, 1982 letter that possibly the Jacoby patent is invalid.

The ROBOTRAC team, because of their familiarity with CARTRAC, can — with minimum effort — bypass the SI patent position.

On June 26, 1982 Dentner authored a letter to Midwest Conveyor — a Kansas City materials handling company — captioned "Proposed Plan of Action for Midwest Conveyors Robotrac Product". Dentner proposed that Midwest establish a temporary office in Easton with the following "charter":

1.) Basic Concepts Designs of Robotrac Components.

a.) Basic design of some critical long lead items such as drive wheel assemblies, drive tube components, stop station components. These items require castings and will be needed for prototypes.

b.) Basic design of most recent GM body shop system components. . . . I feel it is important to get the basic designs completed while they are still fresh in our minds.

c.) Basic design of a car to car accumulation method that does not infringe upon the existing SI (Jacoby) patent.

. . . .

e.) Basic concept design of a Tugger system for use in GM Trim Line automation projects.

f.) Basic concept design of Buffer Systems for use between various operating lines of automotive assembly plants.

. . . .

. . . .

3.) Sales Contact with GM and a few other customers that we have been working with while employees of SI Handling.

The Midwest letter goes on to propose that the Easton office be staffed by Dentner, Gutekunst (who would start by mid-July of 1982 and relocate to Kansas City by July 1983), Scheel (who would start in mid-August of 1982 and relocate by July 1984), and Conklin (who would start in September of 1982 "if still interested"). The remainder of the letter details how the "charter" would be carried out.

Also introduced into evidence was a July 20, 1982 letter from Heisley to ACCO Industries, Inc. ("ACCO"), another materials handling company. In this letter Heisley proposed three ways in which "ACCO could have access to the ROBOTRAC product line at substantially less cost than starting with a development group of your own." The options were: (1) a joint venture (apparently between Heico and ACCO); (2) for Heico to start an engineering group, "located in Easton, Penna., and hire the personnel from SI Handling Systems necessary to provide ACCO with all of the engineering drawings to build ROBOTRAC"; or (3) for ACCO to hire "R. Dentner and 2 or 3 of his associates directly."

The record is largely silent as to events between July of 1982 and February of 1983. It appears that the overtures to Midwest Conveyor and ACCO proved fruitless. At some point, Dentner was installed as president of appellant Eagle Sheet Metal Manufacturing, Inc. ("Eagle"), an Illinois maker of computer room furniture that Heico had acquired, and this became his base for further ROBOTRAC activities.

On February 22, 1983, Scheel, Gutekunst, and appellant Barry L. Ziegenfus (an SI engineer who had significant responsibility for the development of automotive CARTRAC, and who was deeply involved in the exchange of technology with IHI) resigned from SI, and joined Heico. Appellant Frank V. Possinger, who had been fired as SI's manager of cost accounting in a staff cutback in June 1982, had by then also joined Heico. At about this time Ziegenfus persuaded George Bartha, an SI draftsman who is not a party to this litigation, to leave SI and work for ROBOTRAC. With this team in place, Heico finally announced the formation of its ROBOTRAC Division. In a February 25, 1983 letter to the Director of Purchasing at GM's Fisher Body Division, Heisley reported that "[t]he purpose of the ROBOTRAC division is to market, design, engineer, manufacture, install and service a high quality, low cost product similar to SI's CARTRAC." The letter listed all the individual appellants as "a staff uniquely qualified to design, manufacture, install and service CAR-ON-TRACK systems."

The district court described ROBOTRAC's subsequent progress, through the time of the hearing on SI's motion for a preliminary injunction:

On April 11, 1983, HEICO, through ROBOTRAC, submitted to General Motors an "unsolicited bid" for three spinning tube systems. Although unsuccessful in obtaining the bid, it is significant to note that the bid was even considered by GM. Part of the development effort expended by SI in securing General Motors as a customer was the selling of itself as a company that could accomplish the systems engineering necessary to fully automate its manufacturing operations. HEICO, on the other hand, went through no such process. Rather, it attained instant stature with GM by the means of hiring away the key SI people who sold General Motors on the system in the first place and then performed as required to install the systems.

Since the formation of the new division HEICO has been successful in securing one contract with General Motors for an installation at its Livonia plant. It has also bid on several other systems. Of note is its bid on what was termed the Buick City job. The request for quotation on this job was answered by SI and HEICO and the contract awarded to SI. This entire process took place during the time of the hearing. Two matters concerning this bid are important.

First, General Motors' request was conditioned upon responder's undertaking to deliver detailed drawings of all products supplied as well as title to all new developments created in the performance of the contract. SI took exception to these conditions; HEICO did not.

Second, General Motors referred to CARTRAC components in the request by specific CARTRAC terminology and by reference to MH drawing numbers. These MH numbers are General Motors' internal description for SI CARTRAC components.

581 F. Supp. at 1560-61 (footnote omitted). Significantly, ROBOTRAC's April 11, 1983 bid was for two-way accumulation systems in the $200,000 price range that GM had sought since 1981. This was the same date that SI, after two years of development effort, first offered two-way systems in this price range.

C. Proceedings in the District Court

On March 21, 1983 SI filed a complaint against the aforementioned corporate and individual appellants, alleging violations of the Sherman Act and the Racketeering Influenced and Corrupt Organizations Act, and stating a variety of common law claims sounding in contract and tort, including misappropriation of trade secrets. SI moved for a preliminary injunction against use and disclosure of its trade secrets on October 20, 1983, and a hearing on this motion commenced November 7. The hearing, prolonged by acrimonious disputes over discovery (which continued throughout the hearing), a motion for a temporary restraining order (made at the time of the bidding on the Buick City job), and fruitless settlement discussions, ended February 2, 1984 after 25 days of testimony and argument. A number of SI officers and employees testified on the company's behalf. Appellants presented only one witness — an expert on the subject of reverse engineering.

In an opinion handed down on March 1, 1984, the district court found that SI had made the requisite showings for preliminary injunctive relief, and in particular had demonstrated a reasonable probability of success on the merits of its trade secrets claim.

The district court's order enjoining appellants from use and disclosure of SI's trade secrets reads, in pertinent part:

1. all defendants are enjoined from the use and disclosure of any mechanisms developed by them or any of them to design around the Jacoby patent;

2. all defendants are enjoined from the use and disclosure of any detailed drawings of car-on-track materials handling products not revealed by the CARTRAC hardware;

3. the defendants Dentner, Bitely and Possinger are enjoined from engaging in any costing/pricing of a car-on-track materials handling system;

4. the defendants Scheel, Gutekunst and Ziegenfus are enjoined from concepting, designing, manufacturing, marketing or installing of any car-on-track materials handling system except as may be necessary to complete the General Motors Livonia Plant installation;

5. this injunction being preliminary in nature is limited in time until final hearing and the issuance thereafter of an order pertaining to plaintiff's application for a final and permanent injunction; subject, however, to earlier termination, upon appropriate application, if the confidential information in question comes into the possession of defendant by legitimate means.

This appeal followed.

II. [39] DISCUSSION

In considering a motion for preliminary injunctive relief, a court must carefully weigh four factors: (1) whether the movant has shown a reasonable probability of success on the merits; (2) whether the movant will be irreparably injured by denial of such relief; (3) whether granting preliminary relief will result in even greater harm to the nonmoving party; and (4) whether granting preliminary relief will be in the public interest. See Klitzman, Klitzman and Gallagher v. Krut, 744 F.2d 955, 958-59 (3d Cir. 1984); Continental Group v. Amoco Chemicals Corp., 614 F.2d 351, 356-57 (3d Cir. 1980).

Appellants concede nothing here. They challenge every one of the district court's specific trade secret findings as unsupported by the law or the evidence. They argue that none of the prerequisites for preliminary injunctive relief have been met. Finally, they argue that the district court's order is, on any view of the law and evidence, overbroad and vague. We will consider appellants' contentions at some length, but within the confines of the limited scope of our appellate review of the grant or denial of a preliminary injunction.

A. Probability of Success On The Merits

In exercising pendent jurisdiction over trade secrets claims, federal courts must apply state law. Rohm and Haas Co. v. Adco Chemical Co., 689 F.2d 424, 428-29 (3d Cir. 1982); Sims v. Mack Truck Corp., 608 F.2d 87, 95 (3d Cir. 1979), cert. denied, 445 U.S. 930, 100 S.Ct. 1319, 63 L.Ed.2d 764 (1980). In this case there is no dispute that under the choice of law principles of the forum state, Pennsylvania, it is the trade secrets law of Pennsylvania that we are to apply.

To be entitled to an injunction against use or disclosure of information, under Pennsylvania law, a plaintiff must show: (1) that the information constitutes a trade secret; (2) that it was of value to the employer and important in the conduct of his business; (3) that by reason of discovery or ownership the employer had the right to the use and enjoyment of the secret; and (4) that the secret was communicated to the defendant while employed in a position of trust and confidence under such circumstances as to make it inequitable and unjust for him to disclose it to others, or to make use of it himself, to the prejudice of his employer. See Reinforced Molding Corp. v. General Electric Co., 592 F. Supp. 1083, 1087 (W.D.Pa. 1984); Felmlee v. Lockett, 466 Pa. 1, 8, 351 A.2d 273, 277 (1976); MacBeth-Evans Glass Co. v. Schnelbach, 239 Pa. 76, 87, 86 A. 688, 691 (1913). In this case there is no serious dispute that the latter three elements were made out with respect to the trade secrets the district court found. Thus, our discussion will focus on the first element — the existence of trade secrets.

The Pennsylvania courts have adopted the definition of a trade secret given in the Restatement of Torts § 757 comment b. (1939):

A trade secret may consist of any formula, pattern, device or compilation of information which is used in one's business, and which gives him an opportunity to obtain an advantage over competitors who do not know or use it. It may be a formula for a chemical compound, a process of manufacturing, treating or preserving materials, a pattern for a machine or other device, or a list of customers.

See Ecolaire Inc. v. Crissman, 542 F. Supp. 196, 205 (E.D.Pa. 1982); Felmlee v. Lockett, supra, 466 Pa. at 9, 351 A.2d at 277; Van Products Co. v. General Welding and Fabricating Co., 419 Pa. 248, 258-59, 213 A.2d 769, 775 (1965). "Novelty is only required of a trade secret to the extent necessary to show that the alleged secret is not a matter of public knowledge. . . . A trade secret may be no more than a slight mechanical advance over common knowledge and practice in the art." Anaconda Co. v. Metric Tool Die Co., 485 F. Supp. 410, 422 (E.D.Pa. 1980). Matters which are fully disclosed by a marketed product and are susceptible to "reverse engineering" — i.e., "starting with the known product and working backward to divine the process which aided in its manufacture," Kewanee Oil Co. v. Bicron Corp., 416 U.S. 470, 476, 94 S.Ct. 1879, 1883, 40 L.Ed.2d 315 (1974) — cannot be protected as trade secrets. See Air Products and Chemicals v. Johnson, 296 Pa. Super. 405, 432, 442 A.2d 1114, 1128 (1982). Moreover, the concept of a trade secret does not include "[a] man's aptitude, his skill, his dexterity, his manual and mental ability, and such other subjective knowledge as he obtains while in the course of his employment, . . . the right to use and expand these powers remains his property . . . ." Pittsburgh Cut Wire Co. v. Sufrin, 350 Pa. 31, 35, 38 A.2d 33, 34 (1944).

Some factors to be considered in determining whether given information is a trade secret are: (1) the extent to which the information is known outside of the owner's business; (2) the extent to which it is known by employees and others involved in the owner's business; (3) the extent of measures taken by the owner to guard the secrecy of the information; (4) the value of the information to the owner and to his competitors; (5) the amount of effort or money expended by the owner in developing the information; and (6) the ease or difficulty with which the information could be properly acquired or duplicated by others. Restatement of Torts § 757 comment b (1939); International Election Systems Corp. v. Shoup, 452 F. Supp. 684, 706 (E.D.Pa. 1978), aff'd, 595 F.2d 1212 (3d Cir. 1979).

With these general principles before us, we will consider the factual and legal basis for each of the district court's trade secret findings:

SI's method of examining drive tubes for concentricity. This testing procedure, developed over a period of years by SI, is intended to ensure that the rotating drive tube, when installed, will not exceed the maximum allowable deflection as the moving car presses down upon it. This would seem to fit comfortably within the rubric of "processes of manufacture," MacBeth-Evans, supra, 239 Pa. at 85, 86 A. at 691, and therefore be protectible subject matter under the trade secret law of Pennsylvania. Cf. Sperry Rand Corp. v. Pentronix, Inc., 311 F. Supp. 910 (E.D.Pa. 1970) (methods of testing magnetic memory cores).

Appellants attempt to trivialize this procedure, arguing that it is one that would be used by any competent engineer and therefore cannot be a protectible trade secret. See Capital Bakers v. Townsend, 426 Pa. 188, 191-92, 231 A.2d 292, 294 (1967) ("`[T]rade secrets' which will be so protected must be particular secrets of the complaining employer and not general secrets of the trade in which he is engaged.") See also Trilog Associates v. Famularo, 455 Pa. 243, 252, 314 A.2d 287, 292 (1974). If so, we must wonder why the competent engineers of SI only began using this procedure after a number of years of experience manufacturing the product. A trade secret may be no more than "merely a mechanical improvement that a good mechanic can make." Schmidinger v. Welsh, 383 F.2d 455, 466 n. 14 (3d Cir. 1967) (quoting Restatement of Torts § 757 comment b (1939)), cert. denied, 390 U.S. 946, 88 S.Ct. 1031, 19 L.Ed.2d 1134 (1968). We cannot say that the district court's finding that this was a secret process of manufacture is clearly erroneous.

The dimensions, tolerances, and method of fit between drive tubes and drive plugs. Drive plugs are attached to each end of the drive tube and connect it to the ball bearings that support the tube while permitting it to rotate. The district court found that the diameter of these plugs, their tolerances ( i.e., the amount by which they may actually deviate from the stated diameter), and the manner in which they are fitted into the drive tube prior to welding are SI's trade secrets.

Appellants do not seem to dispute the district court's finding that this information could not be obtained by disassembling the product. Indeed, tolerances have previously been recognized as trade secrets par excellence, because they cannot be obtained by even the most precise measurements. See Williams v. Curtiss-Wright Corp., 681 F.2d 161, 164 (3d Cir. 1982); Ecolaire Inc. v. Crissman, supra, 542 F. Supp. at 203; Anaconda Co. v. Metric Tool Die Co., supra, 485 F. Supp. at 418; Business Organizations, Milgrim on Trade Secrets § 9.03[3][d][i] (1984) [hereinafter cited as "Milgrim"]. Rather, appellants argue that the tolerances, at least, can be ascertained by reference to a steel bar manufacturer's catalog. The district court rejected this contention, and because we see no necessary connection between the variations in the size of steel bars a manufacturer sells and the variations a particular user can allow for a particular application — to say nothing of the method of fit, which the district court found would be "obliterated" in any attempt to discover it, 581 F. Supp. at 1564 — we are not able to say that the district court's finding of fact was clearly erroneous.

Use of a nonstandard maximum angular misalignment in conjunction with certain grease pack specifications in bearings. SI's ball bearings are designed to allow for greater-than-usual angular misalignment of the shaft, resulting from the slight deflection of the drive tube as the weight of the car presses down upon it. SI also has special bearing lubrication requirements. These specifications required special negotiations with bearing manufacturers, and under the terms of SI's purchase orders the specifications were to be treated as confidential. The district court found that the precise numbers used are SI's trade secrets.

Again, we do not understand appellants to be contending that this is not the sort of information that is protectible under Pennsylvania trade secret law as "formulae" or "patterns", cf. Henry Hope X-Ray Products v. Marron Carrel, 674 F.2d 1336, 1341 (9th Cir. 1982) (applying Pennsylvania law); rather they ask us to overturn the district court's factual finding that this information is, indeed, secret. They point out that the bearing manufacturer used the prefix "J" in the part number, meaning, according to their catalog, "extra loose internal fit", and that this prefix could be found by inspecting the part. This does not adequately answer SI's evidence indicating that the exact specifications were the result of confidential negotiations and that the part was manufactured exclusively for SI — one who ordered by this catalog number simply would not obtain the same part. The district court's finding is not clearly erroneous.

Knowledge of the existence of alternate suppliers of parts at lower prices. The "parts" the district court referred to here appear to be the bearings discussed in the preceding paragraph. Out of 200 bearing manufacturers in the United States, SI found only two that would make bearings that meet its specifications. After purchasing bearings from one of these suppliers for six years, SI learned in 1982 that the other company could offer a lower price, and therefore switched suppliers. The district court deemed this comparative price information to be SI's trade secret.

To the extent that knowledge of alternate suppliers of these bearings, and their respective prices, was dependent on knowing the secret specifications, this information would seem to be secret as well. Milgrim § 2.09[8][c]. We do not, however, recognize this as an independent trade secret. Here the information SI wishes to enjoin appellants from using (the identity of the vendors and the price of their merchandise) is already in the hands of third parties — i.e., the bearing suppliers — who have every incentive, and every right, to disclose it to their customers. To prevent appellants from using this information would put an undue burden on the innocent vendors, as well as place an artificial constraint on the free market. As the Pennsylvania Supreme Court recognized in Van Products, supra, "material sources and costs" are "something that would be learned in any productive industry, . . ." 419 Pa. at 261, 213 A.2d at 776. See also Jewish Employment and Vocational Service v. Pleasantville Educational Supply Corp., 220 U.S.P.Q. 613, 626-27 (E.D.Pa. 1982). The district court erred as a matter of law in holding that knowledge of alternate suppliers of parts, and their prices, is protectible as a trade secret. (Language to the contrary in Sims v. Mack Truck Corp., 488 F. Supp. 592, 601 (E.D.Pa. 1980) is not supported by the cases cited therein and we reject that language.)

Efficiency factors gained from component experience. Design of a CARTRAC system requires knowledge of the efficiency of the various components of the drive tube system (a speed reducer gear box connected to the motor, the belt system, and the drive tube/drive wheel interface). The district court apparently accepted SI's contention that the efficiency of the drive tube/drive wheel interface is nonstandard and not readily measurable.

This is clearly the sort of technical data that, if secret, may be protected under Pennsylvania law. Appellants contend, however, that these efficiency factors may be obtained by reference to standard engineering principles, and that their expert engineering witness demonstrated this on the stand. The expert did indeed, when prompted, come up with the exact figure SI claims as its secret, but it appears he did so without factoring in the drive tube/drive wheel interface. Therefore we cannot say the district court's finding was clearly erroneous.

The nonstandard coefficient of friction used by SI in making calculations for system design. The coefficient of friction between the drive tube and drive wheel is one factor determining the car's acceleration and deceleration distances and times. In designing its systems SI uses an artificial figure incorporating a margin for safety. Because this design number will not be revealed by measurements of actual systems in operation, the district court found that it was SI's trade secret.

Again, appellants challenge this as a finding of fact rather than as a conclusion of law. Appellants' brief points to the admissions of an SI engineer that the downward force and forward thrust of the car can be measured, and argues that, with those two measurements, "another producer would not need to know Cartrac's design driving coefficient; it is simply a function of those two measurements plus a design safety factor." Appellants miss the point — it is the "design safety factor" that the district court found to be a trade secret. This finding is not clearly erroneous.

Knowledge of long lead times in component supply. In order to meet its tight delivery schedules, SI must begin work on certain parts well in advance of the planned delivery date for a system. The district court found that knowledge of these "critical, long lead items" was SI's trade secret.

We believe that this information stands in much the same position as supplier price information — except to the extent that the component itself is secret (and there is no indication that the district court was referring here to components that are not disclosed by the marketed product), knowledge of long lead times in component supply is "something that would be learned in any productive industry," Van Products, supra, and therefore is not protectible as a trade secret.

SI's knowledge of the key decisionmakers within General Motors. SI asserts that GM has over 100,000 white collar workers. According to the district court, "[i]t was only after peeling through layer and layer of personnel and following many false leads that SI was finally able to reach the real decisionmakers as to its product. The development of these contacts was a secret to SI and was of value to it." 581 F. Supp. at 1565.

SI seeks to bring this information within the category of "customer lists", subject matter that has long been protectible as trade secrets under Pennsylvania law. See, e.g., American Ice Co. v. Royal Petroleum Corp., 261 F.2d 365 (3d Cir. 1958); Morgan's Home Equipment Corp. v. Martucci, 390 Pa. 618, 136 A.2d 838 (1957). It is apparent to us, however, that this information — which involves only one customer that is well-known in the industry and that, the record shows, actively seeks to disseminate the information — is something quite different. Appellants use of this information is more akin to the exploitation of SI's "goodwill" than to the misappropriation of secret customer lists. Though this goodwill is obviously of great value to SI, we do not believe it is protectible as a trade secret. Rather, SI could have prevented this exploitation through reasonable covenants not to compete, which are enforceable under Pennsylvania law. See Bettinger v. Carl Berke Associates, 455 Pa. 100, 314 A.2d 296 (1974); Morgan's Home Equipment v. Martucci, supra, 390 Pa. at 628, 136 A.2d at 844. SI did not require such covenants from its employees, and cannot now through the medium of trade secrets law prevent them from exploiting their GM contacts. See Trilog Associates v. Famularo, supra. The words of the Pennsylvania Supreme Court in Spring Steels v. Molloy, 400 Pa. 354, 363, 162 A.2d 370, 375 (1960) are pertinent here:

It is not a phenomenal thing in American business life to see an employee, after a long period of service, leave his employment and start a business of his own or in association with others. And it is inevitable in such a situation, where the former employee has dealt with customers on a personal basis that some of those customers will want to continue to deal with him in his new association. This is so natural, logical and part of human fellowship, that an employer who fears this kind of future competition must protect himself by a preventive contract with his employee, . . .

SI's identification of GM needs for "two-way accumulation", "tugger", and "buffer" systems. The district court found that knowledge of each of these potential markets was peculiar to SI, and was its trade secret.

Whether or not market research may qualify as the sort of "compilation of information" that is protectible under trade secret law, see Sims v. Mack Truck Corp., supra, 488 F. Supp. at 600 (holding that under Van Products market study for front-discharging concrete mixers could not be a trade secret), this knowledge "of GM needs" clearly stands on a different footing. Where market research explores the needs of numerous, diverse buyers, the resulting profile is information that can only be obtained by others who undertake the same study. See Air Products and Chemicals v. Johnson, supra, 296 Pa.Super. at 419-420, 442 A.2d at 1121-22. Here the information relates to a single prominent buyer that is presumably well aware of its own needs, and which naturally would like to choose from among competing sellers. Where the claimed trade secret is information that is in the hands of a third party by whom it is not treated as secret, we think giving trade secret protection would seriously skew the workings of the marketplace. Indeed, it appears from the district court's findings that GM, at least with regard to its two-way accumulation needs, was soliciting bids and buying from a number of SI competitors offering non-"car-on-track" systems.

The two-way accumulation system developed by appellants for ROBOTRAC while they were still in SI's employ. Within six weeks of the formation of ROBOTRAC, appellants offered GM a two-way accumulation system at approximately the $200,000 price GM had sought. The district court found it "inconceivable" that the same personnel who were unable to report any progress on this front in 19 months with SI could make this offer unless "they did so by misappropriating the unmatured development effort paid for by SI. . . . For these reasons the two-way accumulation concept developed by defendants for ROBOTRAC is deemed to be the property of SI and is presumably a trade secret as well." 581 F. Supp. at 1565.

We do not question the district court's factual finding that appellants misappropriated information that was developed with SI's resources and, under the terms of their employment contracts, was SI's property. We cannot, however, agree with the district court's unsupported legal conclusion that this property "is presumably a trade secret as well". It is difficult to understand how information that was never revealed to SI can be its "trade secret" in the sense of information that is important in the conduct of one's business. Cf. Rohm and Haas Co. v. Adco Chemical Co., supra, 689 F.2d at 430. Moreover, there is no finding that if this development were to be marketed, it would be entitled to proprietary protection. See Milgrim § 2.02[1], at 2-14. In the event that appellants could patent this development, or are entitled to trade secret or some other type of proprietary protection, then under Pennsylvania law SI may be entitled to "shop rights" — i.e., a free license to use this technology. See Toner v. Sobelman, 86 F. Supp. 369 (E.D.Pa. 1949); Quaker State Oil Refining Co. v. Talbot, 315 Pa. 517, 174 A. 99 (1934). SI is not without remedies for breach of the employment contract and may have other causes of action as well; the trade secret injunction, however, is simply not the remedy for all employee breaches of faith. We believe that these remedies would be preferable to issuing an injunction that may result in appellants' system never reaching the market at all.

The contents of three pending SI patent applications. SI has applied for patents on several devices invented as part of the development effort leading to the GM "big buy". Scheel, Gutekunst, and Ziegenfus are named as inventors of one or more of these devices. The district court found that the information contained in these applications, which are treated as confidential by the patent office, and which would enable appellants to design around the prospective patents, is SI's trade secret. We see no defect in the district court's conclusion. See Milgrim § 2.06[2].

Methods developed by appellants, while they were still in SI's employ, for achieving car-to-car accumulation without infringing on the Jacoby patent. Documentary evidence showed that as early as April 1982 "the ROBOTRAC team" (then including three SI engineers) had reviewed with patent counsel accumulation devices that would not infringe on SI's patent. The district court determined that the accumulation configuration used on ROBOTRAC prototypes was SI's property and trade secret.

Appellants argue that, unlike the developments reflected in the pending patent applications, the Jacoby patent has been issued and the mechanism fully disclosed, and SI may not extend its legal monopoly to the point of preventing others from designing noninfringing devices that do the same job. We agree, see also Sims v. Mack Truck Corp., supra, 488 F. Supp. at 599, but note that the district court's theory with regard to this trade secret was actually somewhat different. The district court was concerned that the design of the noninfringing devices had occurred while appellants were under contract to turn their developments over to SI. Thus, these developments stand on the same legal footing as the two-way accumulation system discussed previously. For the same reasons given there, we do not think appellants' method of achieving car-to-car accumulation without infringing the Jacoby patent can be protected as SI's trade secret, though SI may well have other remedies for this breach of duty.

SI's CARTRAC costing and pricing information. Apparently the district court rejected SI's contention that the formulae it uses in pricing its systems are trade secrets, but agreed that the numbers SI actually uses are trade secrets that appellants have misappropriated.

As we understand the district court's finding, subsumed under "costing" and "pricing" information is a whole range of data relating to materials, labor, overhead, and profit margin, among other things. Thus, unlike the price of bearings, this is not information that is readily obtainable by anyone in the industry. We believe such information qualifies for trade secret protection. See also Ecolaire Inc. v. Crissman, supra, 542 F. Supp. at 206; cf. C. Albert Sauter Co. v. Richard S. Sauter Co., 368 F. Supp. 501, 506 (E.D.Pa. 1973) (Sherman Act case). Appellants contend, based on SI's admissions that these figures change over time, and the fact that prior to issuance of the preliminary injunction appellants had not won any contracts it competed for against SI, that the district court's finding of fact was clearly erroneous. In the face of ROBOTRAC documents making financial projections on the basis of SI figures, we are not prepared to reverse the district court's factual finding.

Nonstandard formulae for systems design developed by an IHI engineer (the "Tokunago formula book"). This book, entitled "CARTRAC Engineering Data, Theoretical and Practical Analysis", was supplied to SI under its technology exchange agreement with IHI. Though some of the formulae reflect common engineering knowledge, the district court found, based on expert testimony, that others were based on experimental data and assumptions not in the public domain.

Empirical formulae used in systems design are clearly at the very core of trade secret law protection. Appellants ask us to reject the district court's factual finding on the ground that there was no evidence that the appellants are using these trade secrets. We cannot agree. There was ample evidence that the Tokunago formulae were crucial to the design of CARTRAC systems, and that appellants, who were among the few who had access to this information, were building an identical system, from which the district court could surmise that these trade secrets were being used. Cf. Mixing Equipment Co. v. Philadelphia Gear, 436 F.2d 1308, 1314 (3d Cir. 1971) (applying New York law). In trade secret cases

[m]isappropriation and misuse can rarely be proved by convincing direct evidence. In most cases plaintiffs must construct a web of perhaps ambiguous circumstantial evidence from which the trier of fact may draw inferences which convince him that it is more probable than not that what plaintiffs allege happened did in fact take place. Against this often delicate construct of circumstantial evidence there frequently must be balanced defendants and defendants' witnesses who directly deny everything.

Greenberg v. Croydon Plastics Co., 378 F. Supp. 806, 814 (E.D.Pa. 1974). On this issue, more than any other, the failure of the appellants to testify in answer to SI's strong showing justifies the inference that their testimony would be unfavorable to their cause. See Stowe Township v. Standard Life Insurance Co., 507 F.2d 1332, 1337-38 (3d Cir. 1975); United States v. Cherkasky Meat Co., 259 F.2d 89, 93 (3d Cir. 1958).

SI's "know-how" in systems engineering. As the district court described it, this "know-how" is "the cumulative knowledge and experience necessary to design a materials handling system which answers a particular customer's needs in a unique way. As such it goes far beyond the general knowledge and skill which any employee might gain by working at SI. . . . SI's system engineering know-how — the knowledge of how to make a system work — on the other hand is grounded in the trial and error, experimentation and expenditures of investment dollars over the period of product development. This know-how belongs to SI, is a valuable secret to it and has unlawfully been appropriated by defendants." 581 F. Supp. at 1564.

In Van Products plaintiff sought trade secret protection for "know-how" in manufacturing deliquescent desiccant air driers "which it only achieved after years of success and failure, mistakes and corrections, testing and retesting, field testing, experimentation, study and analysis of problems in actual operation." 419 Pa. at 256, 213 A.2d at 774. The Pennsylvania Supreme Court wrote:

[T]he concept of "know-how" is . . . a very fuzzily defined area, used primarily as a short-hand device for stating the conclusion that a process is protectible. It covers a multitude of matters, however, which in the broad sense are not protectible, e.g., an employee's general knowledge and skill.

419 Pa. at 263-64, 213 A.2d at 777. See also Somat Corp. v. Combs, 40 Pa.D. C.2d 107 (Chester County 1966). In verbally formulating its findings the district court was careful to observe the distinction between protectible "know-how" and "general knowledge and skill", but we believe that, in substance, the matter the district court sought to include under this slippery heading was not protectible as a trade secret.

Mixing Equipment Co. v. Philadelphia Gear, supra, where we applied New York law, is instructive. There the plaintiff claimed as its trade secret "know-how" which was described as

proficiency at analyzing the various problems presented by its customers and designing equipment capable of producing the desired results. This ability, while dependent in part upon certain scientific and mechanical relationships known throughout the industry, in the main is based upon experience in the field. That experience, which has been acquired by Mixco during its 50 years of operation, is recorded in the form of graphs, charts, drawings, and other data sheets. . . .

436 F.2d at 1314. Interestingly, it was not the "ability" or "experience" that we recognized as qualifying for trade secret protection, but rather the compiled products of that ability and experience that had been recorded for repetitive use. (Milgrim, similarly, equates protectible "know-how" with "methods and techniques". Milgrim § 2.09[3].) Here we have accorded trade secret status to such compiled "know-how" as the method of testing for drive tube concentricity, efficiency factors, and the Tokunaga formulae. But the employee ability and experience that led to these developments, and presumably will lead to still further developments, does not belong to SI. "If this were not so an apprentice who has worked up through the stages of journeyman and master workman could never become an entrepreneur on his own behalf. Any such system of quasi-serfdom has long since passed away." Midland Ross Corp. v. Yokana, 293 F.2d 411, 412 (3d Cir. 1961) (applying New Jersey law). See also Sims v. Mack Truck Corp., supra, 488 F. Supp. at 600 (rejecting attempt to protect as "know-how" the "entire working knowledge of the Plaintiff in the field of manufacturing, servicing, and selling front-discharge mixers.").

As we understand the district court, the "know-how" it enjoined appellants from using is comprised of two things: (1) the ability to solve novel problems arising in CARTRAC applications; and (2) the experience necessary to avoid past mistakes and failures ("negative know-how"). Though we do not depreciate the value such "know-how" has to SI (indeed, the technology exchange agreements introduced at the hearing all provide that SI will make its employees and their "know-how" available to licensees), or the price at which it was obtained, we do not think that, after employees leave, SI can assert proprietary rights over their problem-solving ability or knowledge of mistakes to be avoided. Under Pennsylvania law, "the employee, upon terminating his employment relationship with his employer, is entitled to take with him `the experience, knowledge, memory, and skill, which he gained while there employed.'" Van Products, supra, 419 Pa. at 260, 213 A.2d at 776. It is also doubtful that under Pennsylvania law an employer can keep his employees' knowledge of past mistakes and failures as a trade secret. See Pressed Steel Car Co. v. Standard Steel Car Co., 210 Pa. 464, 470, 60 A. 4, 7 (1904); see also M. Jager, 1984 Trade Secrets Law Handbook § 5.05[5], at 133. We believe that the line Van Products draws is a salutary one that comports with common sense and human nature. We, too, would be loath to "act as a judicial eraser to blot out . . . knowledge and skill." 419 Pa. at 261, 213 A.2d at 776. Thus, we conclude that the district court erred in finding that the appellants' "know-how" was SI's trade secret.

Reverse Engineering. It is clear that under Pennsylvania law CARTRAC is not entitled to trade secret protection if it is susceptible to reverse engineering, regardless of whether appellants in fact went through such an exercise or, as would be more likely, relied on their memory. See Henry Hope X-Ray Products v. Marron Carrel, supra, 674 F.2d at 1341-42; Van Products, supra, 419 Pa. at 267-68, 214 A.2d at 779-80. Thus, we must review one more finding of the district court — its finding that CARTRAC could not be reverse engineered.

Appellants argue that this finding was clearly erroneous in that it ignored the detailed testimony of their expert engineering witness, Dr. Terrance Willis. Dr. Willis opined that a small team of engineers and draftsmen could take complete measurements of a CARTRAC system and reverse engineer it in ten to twelve weeks. In addition, "he offered that without the benefit of any measurements, but by visually inspecting the system and knowing its function, four engineers without CARTRAC experience, could reverse engineer the entire system in 16 to 18 weeks." 581 F. Supp. at 1568.

We do not believe the district court ignored Dr. Willis's testimony. Rather, it considered the testimony closely and concluded that "Dr. Willis did not have the expertise as to the scope of the field of his alleged endeavor . . . . The inconsistencies, lack of foundation and illogical result cause the testimony of this witness to be of no moment." Id. We have reviewed this testimony and we, too, detect the smell of the lamp about it. Even if our impression was to the contrary, we would be unlikely to reverse what was basically a credibility determination of the district court. Even if credited, Dr. Willis's testimony pales beside the substantial evidence that, prior to ROBOTRAC, nobody had marketed a comparable product in the United States, and the fact that ROBOTRAC — rather than hiring, or even contemplating hiring, engineers without CARTRAC experience and independently developing a product "in 16 to 18 weeks" — worked for nearly a year in putting together a team of former SI engineers. Cf. rohm and Haas Co. v. Adco Chemical Co., supra, 689 F.2d at 431-32. Though we assume, without deciding, that SI had the burden of proving that CARTRAC could not be reverse engineered, contra Henry Hope X-Ray Products v. Marron Carrel, supra, 674 F.2d at 1341 , we cannot say the district court's finding was clearly erroneous.

In summary, we affirm the district court's factual and legal findings that SI has shown a reasonable probability of success on the merits as to the following claimed trade secrets:

(1) SI's method of examining drive tubes for concentricity;

(2) the dimensions, tolerances, and method of fit between drive tubes and drive plugs;

(3) use of a nonstandard maximum angular misalignment in conjunction with certain grease pack specifications in bearings;

(4) efficiency factors gained from component experience;

(5) the nonstandard coefficient of friction used by SI in making calculations for system design;

(6) the contents of three pending SI patent applications;

(7) SI's CARTRAC costing and pricing information; and

(8) nonstandard formulae for systems design developed by an IHI engineer (the "Tokunago formula book").

We reverse the district court's findings that the following were SI's trade secrets: knowledge of the existence of alternate suppliers of parts at lower prices; knowledge of long lead times in component supply; knowledge of key decisionmakers within General Motors, SI's identification of GM needs for "two-way accumulation", "tugger", and "buffer" systems; the two-way accumulation system developed by appellants for ROBOTRAC while they were still in SI's employ; methods developed by appellants, while they were still in SI's employ, for achieving car-to-car accumulation without infringing on the Jacoby patent; and SI's "know-how" in systems engineering.

B. Equitable Considerations

In exercising its discretion in the issuance of a preliminary injunction, a district court must weigh, in addition to the movant's probability of success on the merits: (1) the threat of irreparable harm to the movant if relief is denied; (2) the balance of harms; and (3) the public interest. Continental Group v. Amoco Chemicals Corp., supra, 614 F.2d at 356-57. Appellants contend that the district court abused its equitable discretion because none of these factors militate in favor of a preliminary injunction here. We disagree. Though our holdings with regard to the individual trade secret findings require that the preliminary injunction be vacated, we see no barrier to issuance of a modified injunction on remand.

Appellants contend that there has been no showing that SI will suffer irreparable harm in the absence of a preliminary injunction. They rely on the fact that SI did not move for a preliminary injunction until seven months after filing their complaint. (SI attributes this delay to appellants' failure to comply with discovery.) We do not understand appellants to be asserting the defense of laches, but rather to be suggesting an estoppel — i.e., SI's conduct is inconsistent with a claim of immediate and irreparable harm. We do not, however, think it is important whether the movant considered a preliminary injunction necessary at the time of filing the complaint. The relevant inquiry is whether the movant is in danger of suffering irreparable harm at the time the preliminary injunction is to be issued. Here SI made an ample showing that appellants intended to use its trade secrets, and did not intend to take reasonable measures to preserve their secret status. See Union Carbide Corp. v. UGI Corp., 731 F.2d 1186, 1191-92 (5th Cir. 1984) (applying Pennsylvania law); Ecolaire Inc. v. Crissman, supra, 542 F. Supp. at 205; cf. Continental Group v. Amoco Chemicals Corp., supra, 614 F.2d at 358-59.

Appellants further contend that had the district court properly balanced the harm they would suffer if an injunction was granted against the harm to SI if an injunction was denied, it could not have issued a preliminary injunction. They claim that the harm appellants suffer is "catastrophic" and that for Scheel, Gutekunst and Ziegenfus it is "the equivalent of economic capital punishment". We do not find any evidence in the record that would support these claims. Heico, the parent of Sy-Con and Eagle, appears to be a corporation of roughly equal size to SI, with the advantage of greater diversity. Though it is clear that individual appellants Scheel, Gutekunst and Ziegenfus are prevented by the injunction from exploiting valuable expertise, we cannot find by judicial notice that they have been subjected to economic capital punishment. It appears that they continue to be employed by Heico, and we surmise that their considerable talents have found a productive outlet.

Finally, appellants ask us to hold that the district court failed to properly weigh the public interest when it stated that "the preservation of commercial morality far outweighs any negative impact on free competition which might result from any order emanating from this litigation." 581 F. Supp. at 1562. Appellants point to the numerous statutes and cases embodying the strong public policies favoring competition and economic mobility, apparently implying that these are overriding interests. Appellants do not, however, explain why this case is different from numerous other trade secret cases, all implicating similar "commercial morality" issues and thus these same public policy concerns, where injunctions have issued. See, e.g., United Insurance Co. v. Dienno, 248 F. Supp. 553, 557 (E.D.Pa. 1965). Appellants here are not concerned about raising the morality of capitalism; rather their sole concern is raising the profitability of their ventures predicated on utilizing a competitor's trade secrets. A classic statement of the competing policies appears in Wexler v. Greenberg, 399 Pa. 569, 160 A.2d 430 (1960) where the court found that trade secrets cases bring

to the fore a problem of accommodating competing policies in our law: the right of a businessman to be protected against unfair competition stemming from the usurpation of his trade secrets and the right of an individual to the unhampered pursuit of the occupations and livelihoods for which he is best suited. There are cogent socio-economic arguments in favor of either position. Society as a whole greatly benefits from technological improvements. Without some means of post-employment protection to assure that valuable developments or improvements are exclusively those of the employer, the businessman could not afford to subsidize research or improve current methods. In addition, it must be recognized that modern economic growth and development has pushed the business venture beyond the size of the one-man firm, forcing the businessman to a much greater degree to entrust confidential business information relating to technological development to appropriate employees. While recognizing the utility in the dispersion of responsibilities in larger firms, the optimum amount of "entrusting" will not occur unless the risk of loss to the businessman through a breach of trust can be held to a minimum.

On the other hand, any form of post-employment restraint reduces the economic mobility of employees and limits their personal freedom to pursue a preferred course of livelihood. The employee's bargaining position is weakened because he is potentially shackled by the acquisition of alleged trade secrets; and thus, paradoxically, he is restrained, because of his increased expertise, from advancing further in the industry in which he is most productive. Moreover, as previously mentioned, society suffers because competition is diminished by slackening the dissemination of ideas, processes and methods.

399 Pa. at 578-79, 160 A.2d at 434-35 (footnote omitted). See also Kewanee Oil Co. v. Bicron Corp., supra, 416 U.S. at 481-82, 485-87, 94 S.Ct. at 1886-87, 1888-89; M. Jager, 1984 Trade Secrets Law Handbook §§ 1.03-1.04. It was not necessary for the district court to engage in extended analysis of the public interest — extensive precedent supports an injunctive remedy where the elements of a trade secret claim are established. See, e.g., Belmont Laboratories v. Heist, 300 Pa. 542, 151 A. 15 (1930); Fralich v. Despar, 165 Pa. 24, 30 A. 521 (1894).

C. The Scope of the Injunction

The district court found:

It will be virtually impossible for the various individual defendants not to use their CARTRAC systems engineering knowledge if they continue to work in car-on-track materials handling systems. Further, continued use of it by them would inevitably result in disclosure to fellow employees and potential customers. For these reasons a limited injunction against disclosure and use of the information is not sufficient.

581 F. Supp. at 1569. It is clear that under Pennsylvania law a court of equity may fashion a trade secret injunction that is broad enough to ensure that the information is protected. See Air Products and Chemicals v. Johnson, supra, 296 Pa.Super. at 420-21, 442 A.2d at 1122-23; Milgrim § 7.08[1][b]. Because we have rejected several of the district court's most significant trade secret findings, it is necessary that this preliminary injunction be vacated. Because the terms of an injunction are normally committed to the discretion of the district court, rather than issue a modified order ourselves, we deem it advisable to remand to the district court for reformulation. In view of our holding, it will be necessary for the district court to reconsider its conclusion that it will be "virtually impossible" for appellants to work on car-on-track systems without compromising SI's proprietary information. In particular, the district court should consider whether, with regard to the eighth trade secret findings that we have affirmed, all appellants are equally situated, or that work by appellants on any car-on-track system — as opposed to merely spinning tube car-on-track systems — is threatening to SI's proprietary interests. The district court is in the best position to tailor an appropriate order.

For the guidance of the district court on remand, we set out here two additional concerns relating to the breadth and specificity of the original order.

Paragraph four of the preliminary injunction enjoins appellants Scheel, Gutekunst, and Ziegenfus from, among other things, "concepting" any car-on-track materials handling system. The term "concepting", though it was used by a number of witnesses and appears to be part of the jargon of the trade, is probably one that ought to be eschewed in judicial orders. Our concern is in part linguistic. As Theodore Bernstein has written, the tendency "to make the lesser seem the greater and to enfold the commonplace in the mantle of science or philosophy has had a debasing effect on the work concept." The Careful Writer 113 (1977). The coinage of the nonstandard verb "concepting" only makes matters worse. Moreover we doubt, whatever exactly is meant by "concepting", that it is something that can be enjoined by judicial fiat. A court of equity should not issue an order that it cannot enforce.

The fifth paragraph of the preliminary injunction provides that it will terminate upon disposition of the application for a final injunction, "subject, however to earlier termination, upon appropriate application, if the confidential information in question comes into the possession of defendant by legitimate means." To the extent this means that, should SI's trade secrets be publicly disclosed, the injunction would terminate, we think this is a salutary and necessary provision. Cf. Kewanee Oil Co. v. Bicron Corp., supra, 416 U.S. at 473-74, 94 S.Ct. at 1882-83. To the extent it means, as SI suggests, that appellants should have to retrace every step SI has taken in developing CARTRAC before they can market a similar product, we believe it would be a wasteful and unduly punitive requirement. A court of equity should not require one who knows that Los Angeles is west of Chicago to look at a map before going to Los Angeles. We endorse the current trend toward so-called "lead time" injunctions, whereby the trade secret injunction lasts only so long as is necessary to negate the advantage the misappropriator would otherwise obtain by foregoing independent development. See Anaconda Co. v. Metric Tool Die Co., supra, 485 F. Supp. at 431; Morgan's Home Equipment v. Martucci, supra, 390 Pa. at 636, 136 A.2d at 848; Uniform Trade Secrets Act § 2 Commissioners' Comment (1980); Milgrim § 7.08[1]. See generally Comment, Trade Secrets: How Long Should an Injunction Last?, 26 UCLA L.Rev. 203 (1978).

We recognize that with the advantage of hindsight it is much easier for appellate judges to find a deficiency in a decree than it would be to write a specific decree that does not offend the law when dealing with the somewhat nebulous field of trade secrets. We are aware of the difficult tasks that confronted the trial judge and of the conscientious efforts he made to take the appropriate path through this litigation jungle. In framing a decree a trial judge is not dealing with mathematical niceties, and thus he must have substantial discretion in defining the relief. Here, however, because the preliminary injunction is predicated on erroneous legal conclusions we must remand for a recasting of the decree and the redrawing of the line between permissible and impermissible conduct. Some of the conduct of the appellants was very near the line or within the zone of permissible conduct, while other conduct had no redeeming value and simply reflected appellants' pursuit of profit without regard to the harm caused to SI. As Justice Holmes observed in a somewhat different context, "the inevitable result of drawing a line" is "distinctions [that] are distinctions of degree; and the constant business of the law is to draw such lines." Dominion Hotel v. Arizona, 249 U.S. 265, 269, 39 S.Ct. 273, 274, 63 L.Ed. 597 (1919).

CONCLUSION

For the reasons set forth above, the order of the district court will be vacated and the case remanded for proceedings consistent with this opinion.