COYNE, Justice.

Respondent Electro-Craft Corporation (“ECC”) sued appellants Controlled Motion, Inc. (“CMI”) and CMI’s president, John Ma-honey, (a former employee of ECC) for misappropriation of trade secrets. ECC claimed that CMI and Mahoney improperly copied the designs of ECC’s electric motors. The district court found that misappropriation had occurred and also found appellants in contempt for violating a temporary restraining order. We reverse the order for judgment based on misappropriation and affirm the order finding appellants in contempt.

THE PRODUCTS

ECC, a Minnesota corporation, manufactures D.C. iron core motors, moving coil motors, brushless motors, and other products. ECC has four plants, three of which manufacture motors and motor parts. The total employment of those three plants is around 500.

CMI is a Minnesota corporation which manufactures moving coil motors. John Mahoney is the president and founder of CMI. Mahoney was formerly national sales manager for ECC. While at ECC, Mahoney established ECC’s customer relationships with Storage Technology Company and IBM, customers for the motors involved in this lawsuit.

ECC and CMI manufacture high performance D.C. motors, called “servo” motors, which are able to start and stop at least 30 times per second. These motors are useful for such high technology applications as computer disc drives and printers and industrial robots. In this action ECC claims misappropriation of trade secrets with respect to one moving coil motor and one brushless motor.

Moving coil motors, also known in some forms as shell-type armature motors or basket armature motors, represent a major development in high performance motors. In traditional D.C. motors, the rotor (moving component) consists of an iron core with wire wrapped around it in coils; the stator (stationary component) consists of permanent magnets, which create a magnetic field through the coils, causing motion when current is passed through the coils. In a moving coil D.C. motor, the coils of wire are not wrapped around a core to form the armature or rotor; the coils themselves form the armature. The wire coils are formed into a cylindrical basket which is made rigid by a combination of adhesives and other materials such as fiberglass. Since the moving part of the motor is a hollow shell, the inertia of the motor is reduced while the torque remains high. The motor may thus start and stop quickly without fighting as much inertia as is present in the iron core motors.

Working moving coil motors have been produced for many years. However, the early motors were low performance motors because of problems in making the coil structure rigid. These problems have been solved today because of developments in adhesives.

Iron core motors and moving coil motors, described above, use brushes and commutators to transmit current to the proper coils in the rotating armature. As the armature *894turns, different segments of the commutator (connected to different coils) come into contact with the stationary brush; thus, different coils receive current. These brush-type motors have two major disadvantages. First, these motors involve sliding contact between the brushes and the commutator; therefore, parts tend to wear out quickly, and some arcing occurs in the commutators, causing interference. Second, heat buildup is a problem in moving coil motors. The wires, which heat up as current is passed through them, are on the inside and surrounded by magnets, which trap the heat. Thus, moving coil motors must have a complex cooling system to draw away the heat.

Brushless motors solve these two problems by having the wire coils stationary on the outside, with the magnets mounted on the rotor. Thus, the heat can easily be drawn away from the coils on the outside. Furthermore, the coils do not move, so brushes are not required to transmit current, and no sliding contact exists to cause breakdowns. The brushless motor contains a sensing device to activate the proper coils to cause rotation. Brushless motors were developed more than 20 years ago for military and aerospace projects, but these were very expensive. Until six or seven years ago, the only commercial brushless motors were low performance models.

According to Mr. Edward Kelen, president of ECC, ECC was one of only three significant producers of small moving coil motors in 1980. Also, although seven companies produced brushless servo motors, ECC had more than half of the private sector domestic market. Motors made for the producer’s own use and motors produced for military or foreign markets were not included in Kelen’s evaluations, but it seems clear that ECC’s share of the overall servo motor market is substantial. Moreover, Kelen testified that the total brushless motor market is growing rapidly; in five years (from 1980) it was projected to grow from two million dollars to fifty or sixty million dollars per year.

ECC began work on moving coil motors in 1966. At this time Honeywell had already applied for a patent on a moving coil motor. ECC began making primitive moving coil motors in 1967; ECC developed its 1030/1040 motor to meet the needs of users of a comparable Honeywell motor, the 33 V.M. The two motors are nearly identical, and Designer Erland Persson had a 33 Honeywell V.M. in his possession when he designed ECC’s 1030/1040. Persson testified that he had problems with the adhesives and supporting materials for the 1030/1040 and that it took ECC from six to eight months to develop the armature for the motor.

About 1974, Robert Schept, Engineering Manager at ECC, designed the successful ECC 1600 moving coil motor. Schept studied various other motors on the market to try to determine the best combination of dimensions for the motor. Subsequently, Schept designed the 1125 moving coil motor for ECC. He did this by scaling down the ECC 1600 while keeping its performance basically the same. Schept testified that it took four to five months to design prototypes for the 1125 and around a year to start actual production. The 1125 is now used for a variety of applications; for each application ECC produces a different model 1125 with different dimensions. The model involved in this case is the 1125-03-003, designed for a specific computer system built by Storage Technology Co.

ECC’s brushless motors are of more recent origin than that of ECC’s moving coil motors. Nonetheless, a long process of trial and error, using new developments in technology and costing approximately two million dollars, has been involved in the development of workable models. ECC initially sent prototypes to several customers. Finally ECC worked with IBM to develop a prototype motor for an IBM printer. This model was a very successful enterprise for ECC, since ECC became the only known source of motors for the IBM project. For around five years, ECC has also been working with Ford Motor Company on a new application for ECC brushless motors. This application involves using larger ECC mo*895tors in automated assembly lines. ECC has produced several prototypes for Ford and hopes to begin production in 1983.

THE EVENTS

In May of 1980, John Mahoney, while employed by ECC, began to explore the' possibility of starting his own business. Mahoney already had many contacts in the business, including people at Storage Technology and at IBM — ECC customers for the ECC 1125-03-003 and brushless motors. Mahoney had also guided development of the IBM project and the Ford project. On June 12, 1980, Mahoney hired an attorney as counsel for the proposed new business, and counsel helped Mahoney prepare a prospectus which was circulated to prospective investors. Mahoney met with several prospective investors during June and July but apparently received no investments before August 1980.

The prospectus indicates that Mahoney proposed to compete with ECC in its IBM and Ford applications. Mahoney planned to complete prototypes for IBM in twelve weeks and obtain IBM approval in another week. Mahoney planned to try eventually to enter the market for the Ford systems. The prospectus projected revenues in the third month from sales of the prototype brushless motors but projected no research and development expenses for the first few months.

In June of 1980 Mahoney met with several of his fellow ECC employees about their joining the new business. On August 6, 1980, Mahoney resigned from ECC. Maho-ney and ECC’s president, Kelen, met briefly regarding trade secrets and Mahoney told Kelen not to worry. On September 16, 1980, four other ECC employees resigned in order to work for Mahoney’s company, now called CMI. The four employees were:

(1) William Craighill, a mechanical engineer who worked with ECC on the design of the ECC 1125 and the brushless motor for IBM. Craighill had previously worked for Control Data Corporation. At Control Data, Craighill did not design D.C. electric motors but worked on systems which he testified were similar to D.C. electric motors.

(2) James West, previously Quality Assurance Manager at ECC for electric motors. West was acting plant manager at one ECC plant for six months.

(3) William Anderson, who had worked for ECC as a technician for about two years. Anderson now works for Honeywell.

(4) Lynn Klatt, Buyer’s Assistant at ECC, who was familiar with ECC’s vendors and with the parts used in ECC’s motors.

All of these employees, as well as Maho-ney, had signed confidentiality agreements1 when hired by ECC. None of these agreements, however, included a non-competition clause. When these four employees left ECC, ECC’s management conducted exit interviews. The employees were asked to sign acknowledgment forms which outlined the areas that ECC considered confidential; only Anderson signed the acknowledgment.

On September 17, 1980, Mahoney met with some IBM representatives about developing a CMI motor as an alternative to the ECC brushless motor. The IBM representa-, fives said that IBM would not deal with such a new company and that Mahoney should try again in 6 months.

On September 18, Mahoney traveled to Colorado to meet representatives of Storage Technology Co. Mahoney received the specifications for a moving coil motor to *896meet Storage Technology’s application, at that time supplied by the ECC 1125-03-003 and a Honeywell motor. Mahoney delivered prototypes of a CMI motor, the CMI 440, to Storage Technology on December 15, 1980; the motor was finally approved on March 1, 1981.

The evidence is conflicting as to how CMI produced the 440. The CMI 440 is almost identical in dimensions and tolerances2 to the ECC 1125-03-003. William Craighill, the former ECC employee who developed the CMI 440, testified that he did not copy, nor even possess, an ECC 1125 motor when he designed the CMI 440. Craighill claimed that he used only a similar Honeywell motor, the Storage Technology specifications, and his own calculations to develop the CMI 440. On the other hand, circumstantial evidence pointed to the conclusion that CMI employees copied the ECC 1125. The similarity of the motors suggests copying, although the motors are not absolutely identical. Furthermore, the manufacturing processes, adhesives, and other materials are nearly identical. Expert testimony differed as to how long it should have taken CMI to “reverse engineer” the motor by taking apart an ECC 1125, measuring the parts and testing the material, and putting the plans together. A CMI expert estimated that it should have taken two to three months to develop a prototype motor. An expert for ECC estimated that the process would take at least six months to a year. Kelen estimated it would take a year.

CMI has, by reverse engineering, also been able to market remanufactured ECC and Honeywell motors. In remanufactur-ing, CMI replaces the parts to a broken motor, rebuilds it and rewarrants it.

THE ACTION

On September 26, 1980, about six weeks after Mahoney’s resignation, ECC sued CMI and Mahoney (hereinafter referred to together as “CMI”) claiming that CMI misappropriated ECC’s trade secrets. On October 1, 1980, Judge William Posten issued an order which enjoined CMI from accepting orders from ECC’s principal “Customer” (not named in the complaint but later shown to be IBM) with respect to “brush-less” motors. After a preliminary hearing Judge Lindsay Arthur of the Hennepin County District Court granted a temporary injunction in favor of ECC on April 16, 1981. CMI was enjoined by this order, pending decision at trial, from “directly or indirectly selling or soliciting sales of any low inertia, DC, electric servo motor for which any component not available on the open market has dimensions closer than 10% to Plaintiff’s 1125 motor.” It was later established that CMI had continued to produce and sell electric motors after April 16, 1981. On July 9,1981, Judge Arthur issued an order holding CMI in civil contempt for violating the temporary injunction. CMI was ordered to pay ECC $50.00 damages for each offending motor which CMI had sold since April 16, 1981, plus the premiums on ECC’s cost bond. The $50.00 per motor damages were later rescinded for lack of evidentiary foundation, but CMI remained liable for the bond premiums.

A trial was held before Judge Arthur, without a jury, from June 15, 1981 to July 7, 1981. By order of October 19, 1981, Judge Arthur found that CMI had misappropriated ECC’s trade secrets and enjoined CMI from producing or selling any “brush-less or low inertia electric motor or tachometer” with dimensions within 10% of the dimensions of ECC’s 1125 motor or ECC’s brushless motor produced for IBM. The injunction was to be in effect for 12 months after the expiration of the last stay of execution of the order. The court also awarded ECC $50.00 in exemplary damages (but no compensatory damages) for each offending motor sold.

CMI appealed separately from (1) the temporary injunction order of April 16 and the contempt order of July 9 (No. C7-81-894) and (2) the final order of October 19 (No. CO-81-1188). ECC filed a notice of *897review of the district court’s final order, seeking compensatory damages, attorney fees, and further injunctive relief, and seeking recovery against John Mahoney individually. The appeals have been consolidated for consideration.

THE ISSUES

Review of the district court’s order is made difficult by the complexity of the issues and the sheer volume of technical evidence. We must affirm the final order if we can ascertain that (1) ECC has pro-tectable trade secrets — i.e., that the district court correctly enumerated the necessary elements of trade secret status and found those elements to exist, and that those findings are reasonably supported by the evidence; (2) that ECC’s trade secrets, if they exist, have been misappropriated by CMI; and (3) that the relief granted in the final order was appropriate. In reviewing the contempt order we must determine whether the district court properly held CMI in contempt and whether the relief granted to ECC was appropriate.

A. Trade Secret Status.

The Uniform Trade Secrets Act, Minn.Stat. §§ 325C.01-325C.08 (1982)3 allows the protection of certain types of information through an action for misappropriation. Misappropriation is defined as improper acquisition, disclosure, or use of a “trade secret.” Minn.Stat. § 325C.01, subd. 3. Without a proven trade secret there can be no action for misappropriation, even if defendants’ actions were wrongful.4

1. In defining the existence of a trade secret as the threshold issue, we first focus upon the “property rights” in the trade secret rather than on the existence of a confidential relationship. 12 Business Organizations, Milgrim, Trade Secrets § 1.09 at 1-42 (1981). We recognize that the confidential relationship is also a prerequisite to an action for misappropriation, Jostens, Inc. v. National Computer Systems, Inc., 318 N.W.2d 691, 701 (Minn.1982), and that the elements of trade secret status and the confidentiality of the relationship “should not be artificially separated for purposes of analysis since, in a significant sense, they are interdependent.” Id. at 701, quoting 1 Milgrim, Trade Secrets § 7.07(1) at 95. However, without the finding of a trade secret, we cannot grant relief to ECC. Otherwise this court would come dangerously close to expanding the trade secrets act into a catchall for industrial torts.5

2. In order to determine the existence of trade secrets, we must first determine what trade secrets are claimed by ECC and what trade secrets were found by the district court. CMI claims that neither ECC nor the district court were specific enough in defining ECC’s trade secrets. *898CMI also claims that ECC’s definition of its trade secrets changed during the course of the litigation. Therefore, according to CMI the district court should be reversed due to lack of specificity.

Regarding the brushless motor, we agree with CMI that ECC did not specify its trade secrets at trial. Nor did ECC even introduce the dimensions, tolerances, etc. of the brushless motor into evidence. The trial court found trade secrets in the general “design procedures” for the brushless motor. The court then enjoined CMI with respect to duplication of only the dimensions of the brushless motors. This lack of clarity is fatal to ECC’s claim. On the record before us, ECC did not meet its burden of showing that certain features of the brushless motor were protectable trade secrets which might be misappropriated in the future. Furthermore, given ECC’s lack of specificity, it was impossible for the district court to fashion a meaningful injunction which would not overly restrict legitimate competition for the IBM project.

With respect to the moving coil motors, however, ECC claims that the dimensions, tolerances, adhesives,6 and manufacturing processes of the ECC 1125-03-003 motor are trade secrets.7 The thrust of ECC’s claim is that the specific combination of details and processes for the 1125 motor is a trade secret, and the evidence of the specific features of the 1125 motor sold to Storage Technology adequately identifies the information which ECC claims constitutes a trade secret.8 We believe that ECC’s claim was specific enough in identifying its trade secrets to support a misappropriation action with respect to the 1125.

3. In determining whether ECC has proven the existence of a trade secret in the 1125 motor, we look to the common law and the Uniform Trade Secrets Act, Minn. Stat. §§ 325C.01-325C.08 (1982). The Act, which became effective on August 1, 1980, carries forward, explains, and clarifies many of the rules of the common law of trade secrets.9 To the extent, however, that the Act modifies the common law, we are constrained to give effect to the statutory language.

The common law of trade secrets in Minnesota has been defined by two recent pre-Act cases. In Cherne Industrial, Inc. v. Grounds & Associates, Inc., 278 N.W.2d 81 (Minn.1979), we adopted the four-point test of Restatement, Torts § 757: to be a trade secret the information must (1) not be generally known or readily ascertainable, (2) provide a competitive advantage, (3) have been developed at plaintiff’s expense, and (4) be the subject of plaintiff’s intent to keep it confidential. 278 N.W.2d at 90. In Jostens, Inc. v. National Computer Systems, 318 N.W.2d 691 (Minn.1982), we applied the above test to a case quite similar to the one at bar. In Jostens plaintiff used a computer system for designing student class rings. Defendant company hired a former employee of plaintiff to design similar computer systems. The programs were written for defendant by the company which had written plaintiff’s programs. This court used the first and fourth Cherne factors to uphold the trial court’s finding that plaintiff *899had no protectable trade secrets: the system was readily ascertainable because it was not novel, and plaintiff did not show an intent to keep the system secret. 318 N.W.2d at 698-701.

The test used in Cherne and Jostens is now more or less embodied in the Act, Minn.Stat. § 325C.01, subd. 5 (1982):

“Trade secret” means information, including a formula, pattern, compilation, program, device, method, technique, or process, that:

(i) derives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable by proper means by, other persons who can obtain economic value from its disclosure or use, and

(ii) is the subject of efforts that are reasonable under the circumstances to maintain its secrecy.

(Emphasis added). Applying the statutory test, we hold that, like the plaintiff in Jos-tens, ECC has not met its burden of proving the existence of any trade secrets. Therefore, we must reverse the district court’s order even though the district court in this case found trade secrets to exist and made specific findings leading up to that ultimate conclusion. Because we do not lightly reverse the district court, we feel compelled to discuss all three elements of trade secret status under the Act as they relate to the evidence and the findings of the district court.

(a) Not generally known, readily ascertainable. The trial court found the information regarding the ECC 1125-03-003 to be secret. This finding was not clearly erroneous. First, the trial court found on conflicting evidence that CMI could not readily (i.e. quickly) reverse engineer a motor with exactly the same dimensions, tolerances, and materials as the ECC 1125-03-003. This finding was not clearly erroneous. Reverse engineering time is certainly a factor in determining whether information is readily ascertainable. ILG Industries, Inc. v. Scott, 49 Ill.2d 88, 94, 273 N.E.2d 393, 396 (1971); Kubik, Inc. v. Hull, 56 Mich.App. 335, 359-60, 224 N.W.2d 80, 92-93 (1974). The complexity and detail of dimensional data also bears on its ascertain-ability. A.H. Emergy Co. v. Marcan Products Corp., 389 F.2d 11, 16 (2d Cir.1968) (“[I]t is well settled that detailed manufacturing drawings and tolerance data are pri-ma facie trade secrets.”); Henry Hope X-Ray Products, Inc. v. Marron Carrel, 674 F.2d 1336, 1340-41 (9th Cir.1982) (configuration of gear system and tolerances was secret).

Second, the district court found that the exact combination of features of the 1125-03-003 is unique, even though none of the processes or features are unique in the industry and the 1125-03-003 is not the only way to achieve the required performance. Novelty is not a requirement for trade secrets to the same extent as for patentability. E.g., Clark v. Bunker, 453 F.2d 1006,1009 (9th Cir.1972). On the other hand, some novelty is required; mere variations on widely used processes cannot be trade secrets. Thus, in Jostens, a type of computer system was held not to be secret where it merely combined known subsystems (and where defendant had produced a different system).10 In the present case the exact combination of features of the 1125-03-003 could be characterized as a unique solution to the needs of one customer in the industry.

The district court’s findings as to ascertainability and uniqueness are supported by the record. ECC introduced testimony that ECC’s particular combination of production techniques is unique and that *900only general design principles circulate in the industry. Further, expert testimony conflicted on how “readily” ascertainable are the features of the motor by reverse engineering, (although the evidence does contradict the trial court’s finding that the adhesives are not ascertainable). Therefore, the finding of the trial court, that the features of the motor are not generally known or readily ascertainable, is supported by substantial evidence and is not clearly erroneous.11

This is not to suggest that ECC could make out a claim for trade secret status for the entire class of ECC moving coil motors, or even for the ECC 1125 in general. If a new customer devised an application for the ECC 1125 and CMI modified its 440 motor to meet those new specifications, ECC could not object. In that case ECC would be trying to protect, not a specific combination of features, but the design process of trial and error, (including the talent of ECC’s employees), by which those features are adapted to a given use. The law of trade secrets will not protect talent or expertise, only secret information. See, Dynamics Research Corp. v. Analytic Sciences Corp., 9 Mass.App. 254, 274-75, 400 N.E.2d 1274, 1286 (1980) (engineering management system really involved talent of consultants, not information, so not a trade secret), and cases cited therein.

(b) Independent economic value from secrecy. This statutory element carries forward the common law requirement of competitive advantage. The trial court found ECC to have a competitive advantage in the ECC 1125-03-003, not over all competitors, but over “any company which has not, through its own efforts or through license, obtained similar information.” CMI claims that ECC was required to show a competitive advantage over all competitors, including those competitors already successfully producing motors.

The statute requires that a trade secret “[derive] independent economic value ... from not being generally known ... and not being readily ascertainable .... ” Minn.Stat. § 325C.01, subd. 5(i). This does not mean, as CMI contends, that the owner of the trade secret must be the only one in the market. Several developers of the same information, for example, may have trade secret rights in that information. Uniform Trade Secrets Act § 1, Commissioner’s Comment, 14 U.L.Á. 542-43 (1979). If an outsider would obtain a valuable share of the market by gaining certain information, then that information may be a trade secret if it is not known or readily ascertainable.

It is true that, as CMI contends, the trial court’s stated reasons for its finding of a competitive advantage are not valid as to the 1125-03-003. First, the court cited a “dearth of effective competitors for its principal customer” as evidence of advantage. The trial court could only have meant brushless motors customer IBM, since Honeywell is an approved source for the 1125 motor sold to Storage Technology. *901Second, the court cited the time and money ECC reasonably expended in developing its motors. That ECC expended time and money between 1966 and 1975 in the development of the 1125 motor and its predecessors does not support a finding of competitive advantage unless, under the present state of the art, a prospective competitor could not produce a comparable motor without a similar expenditure of time and money.12 The trial court found, however, that such time and money would be required of a prospective competitor today, and that finding was supported by some evidence. The ECC 1125, therefore, did provide ECC with economic value from its secrecy, as the statute requires — value that ECC would lose if any prospective competitor could enter the market (cutting into ECC’s market share) without a substantial development expense.

(c) Reasonable efforts to maintain secrecy. It is this element upon which ECC’s claim founders. The district court found that, even though ECC had no “meaningful security provisions,” ECC showed an intention to keep its data and processes secret. This finding does not bear upon the statutory requirement that ECC use “efforts that are reasonable under the circumstances to maintain ... secrecy.” MinmStat. § 325C.01, subd. 5(ii). The “intention” language used by the district court comes from the common law test for trade secret status. Cherne Industrial, Inc. v. Grounds & Associates, supra, 278 N.W.2d at 90. However, even under the common law, more than an “intention” was required — the plaintiff was required to show that.it had manifested that intention by making some effort to keep the information secret. Id. at 91; Jostens, supra, 318 N.W.2d at 700.

This element of trade secret law does not require maintenance of absolute secrecy; only partial or qualified secrecy has been required under the common law. Radium Remedies Co. v. Weiss, 173 Minn. 342, 347-48, 217 N.W. 339, 341 (1928). What is actually required is conduct which will allow a court acting in equity to enforce plaintiff’s rights. In speaking of the requirement, one commentator has stated:

It would appear, from the standpoint of a broad overview, that the policy goal of the doctrine is to preclude employee liability unless the employer can establish that his treatment of the knowledge in issue has been adequate to indicate a breach of the confidential relationship. It might also be said that the goal is to preclude vindictive employers from placing employees in mental-bondage.

Sloan, Trade Secrets: Real Toads in a Conceptual Garden, 1 W.StU.L.Rev. 113, 145 (1973). To put it another way, the employer must come into court with clean hands; the employer cannot complain of the employee’s use of information if the employer has never treated the information as secret.

It is this aspect of trade secret law which truly sets it apart from the other two means through which employers can protect information — patents, and employment contracts containing a non-competition clause. The latter two remedies depend on only a single act by the employer. Trade secret protection, on the other hand, depends upon a continuing course of conduct by the employer, a course of conduct which creates a confidential relationship. This relationship, in turn, creates a reciprocal duty in the employee to treat the information as confidential insofar as the employer has so treated it (see discussion of misappropriation, section B., infra).

In the present case, even viewing the evidence most favorably to the findings below, we hold that ECC did not meet its burden of proving that it used reasonable efforts to maintain secrecy as to the ECC 1125-03-003. We acknowledge that ECC took minimal precautions in screening its Handbook and publications for confidential information and by requiring some of its *902employees to sign a confidentiality agreement,13 but these were not enough.

First, ECC’s physical security measures did not demonstrate any effort to maintain secrecy. By “security” we mean the protection of information from discovery by outsiders. Security was lax in this case. For example, the main plant had a few guarded entrances, but seven unlocked entrances existed without signs warning of limited access. Employees were at one time required to wear badges, but that system was abandoned by the time of the events giving rise to this case. The same was generally true of the Amery, Wisconsin plant where ECC 1125 and brushless motors were manufactured. One sign was posted at each plant, however, marking the research and development lab at Hopkins and the machine shop at Amery as restricted to “authorized personnel.” Discarded drawings and plans for motors were simply thrown away, not destroyed. Documents such as motor drawings were not kept in a central or locked location, although some design notebooks were kept locked.

The relaxed security by itself, however, does not preclude a finding of reasonable efforts by ECC to maintain secrecy. Other evidence did not indicate that industrial espionage is a major problem in the servo motor industry. Therefore, “security” measures may not have been needed,14 and the trial court could have found trade secrets if ECC had taken other reasonable measures to preserve secrecy.

However, ECC’s “confidentiality” procedures were also fatally lax, and the district court was clearly in error in finding ECC’s efforts to be reasonable. By “confidentiality” in this case we mean the procedures by which the employer signals to its employees and to others that certain information is secret and should not be disclosed.15 Confidentiality was important in this case, for testimony demonstrated that employees in the servo motor business frequently leave their employers in order to produce similar or identical devices for new employers. ECC has hired many employees from other corporations manufacturing similar products.16 If ECC wanted to prevent its employees from doing the same thing, it had an obligation to inform its employees that certain information was secret.17

ECC’s efforts were especially inadequate because of the nonintuitive nature of ECC’s claimed secrets here. The dimensions, etc., of ECC’s motors are not trade secrets in as obvious a way as a “secret formula” might be. ECC should have let its employees *903know in no uncertain terms that those features were secret.

Instead, ECC treated its information as if it were not secret. None of its technical documents were marked “Confidential”, and drawings, dimensions and parts were sent to customers and vendors without special marking. Employee access to documents was not restricted. ECC never issued a policy statement outlining what it considered to be secret. Many informal tours were given to vendors and customers without warnings as to confidential information. Further, two plants each had an “open house” at which the public was invited to observe manufacturing processes.

The district court relied on certain contrary evidence to show ECC’s “intention,” but this evidence does not demonstrate reasonable efforts to maintain confidentiality. There was no showing that a 1977 memo from the president of ECC to its managerial employees, warning them to restrict unannounced laboratory tours in the interests of protecting secrets, had ever been enforced. The confidentiality agreements signed by the employees were too vague to apprise the employees of specific “secrets.” (See note 1, supra).

The exit interviews also did not constitute reasonable efforts to maintain secrecy. The exit interviews, a procedure initiated by ECC only after it became clear that the employees were about to work for Maho-ney, occurred a mere ten days before the commencement of this litigation. These “interviews” were little more than attempts to intimidate or threaten employees, to prevent them from leaving ECC and engaging in legitimate competition using their skill and expertise. Such thinly-veiled threats certainly do not qualify as ongoing efforts to maintain the secrecy of specific information. The law of trade secrets does not condone, and this court certainly will not reward, ECC’s conduct.

In summary, ECC has not met its burden of proof in establishing the existence of any trade secrets. The evidence does not show that ECC was ever consistent in treating the information here as secret.

B. Misappropriation.

Since no trade secrets existed to be misappropriated, we technically need not reach the issue of whether misappropriation occurred. However, as we noted above the concept of trade secret status and the concept of misappropriation should not be artificially separated. In the present case the concepts are so interrelated that we feel compelled to discuss the tort of misappropriation. Misappropriation involves the acquisition, disclosure, or use of a trade secret through improper means. Minn.Stat. § 325C.01, subd. 3. “Improper means” are defined as

[TJheft, bribery, misrepresentation, breach or inducement of breach of a duty to maintain secrecy, or espionage through electronic or other means.

Minn.Stat. § 325C.01, subd. 2. In the employer-employee context of the present case, ECC was required to show some duty on the part of the employee not to disclose the information. ECC claims that the employees’ duty here arose from the employee agreements and from a confidential employer-employee relationship.

However, a common law duty of confidentiality arises out of the employer-employee relationship only as to information which the employer has treated as secret:

[T]he employee is entitled to fair notice of the confidential nature of the relationship and what material is to be kept confidential.

Jostens, supra, 318 N.W.2d at 702 (citing Ellis, Trade Secrets 79 (1953)). Therefore, in the present case, ECC’s failure to make reasonable efforts to maintain secrecy, discussed above, was fatal to its claim of a confidential relationship. The employees were never put on notice of any duty of confidentiality. The employee agreements do not help ECC’s claim for the same reason — ECC never treated specific information as secret. Therefore, the agreements’ vague language prohibiting the employee from taking “secrets” did not create a duty of confidentiality in the employee, and no misappropriation occurred.

*904We reverse the district court’s final order of October 19, 1981 (Appeal No. CO-81-1188).

C. Contempt Order.

Since we find no misappropriation, we reverse the district court’s final order and need not decide whether the final relief granted by the district court was proper. In a separate order, however, the district court held CMI in civil contempt, and that order must be reviewed.

On April 16, 1981, the district court enjoined CMI from selling any D.C. servo motor “for which any component not available on the open market has dimensions closer than 10% to Plaintiff’s 1125 motor.” CMI then sent the armature of a CMI 440 motor to Magnedyne Corp. for an estimate of Magnedyne’s price to produce the armature. Magnedyne quoted a price for the armature. (The normal procedure for obtaining the quotation, however, would have been to send drawings and dimensions of the armature to Magnedyne). Because of the quotation from Magnedyne, CMI contends that all parts of the CMI 440 are “available on the open market” and that CMI did not violate the injunction by selling the 440.

The district court stated that CMI’s interpretation would render the injunction meaningless. We agree, for under CMI’s interpretation any part of any motor would be “available on the open market,” if CMI could find someone who would copy the part, and therefore no motors could ever be the subject of the injunction. CMI’s interpretation was unreasonable and was in fact a shallow pretext used to justify a clear violation of the injunction.

We agree with CMI that the injunction was extremely broad and that compliance would have made continued operations difficult for CMI, and we doubt that the injunction was necessary to protect ECC from irreparable harm. CMI could have sought modification or review of the injunction. Its decision to violate the injunction instead of seeking its modification demonstrated disrespect for an order of a court of this state, and the district court properly held CMI in contempt.

The district court originally awarded ECC contempt damages of $50 per CMI motor and also ordered CMI to pay the premiums on ECC’s cost bond. Later the per-unit damages were rescinded for lack of supporting evidence, but CMI remained liable for the premiums. We believe that the bond premium liability is an appropriate remedy because CMI’s actions rendered the bond meaningless. Therefore, we affirm the district court’s contempt order (Appeal No. C7-81-894), in all respects.

Appeal No. C7-81-894 is affirmed; appeal No. CO-81-1182 is reversed.

EDITH BROWN CLEMENT, Circuit Judge:

This dispute arises from the marketing and sale of the cardiovascular health drug Arterosil. Daniels Health Sciences (“DHS”) engaged Vascular Health Sciences (“VHS”) to market and sell the drug Provasca. After that relationship ended, VHS began to manufacture, market, and sell Arterosil, a product similar in many respects to Provasca. DHS sued VHS for misappropriation of trade secrets, breach of contract, and trademark violations. The district court first granted a temporary restraining order on the grounds that DHS would likely succeed on its breach of contract and misappropriation of trade secrets claims. It later granted a preliminary injunction on the same grounds, also finding a substantial threat of irreparable injury absent an injunction, that the balance of hardships favored the plaintiff, and that the public interest would not be dis-served by a grant of injunctive relief. VHS filed a motion in this court requesting a stay of the injunction, which was granted, and now appeals the grant of the preliminary injunction and its scope. We AFFIRM the preliminary injunction, lift the stay, and remand to the district court with *581instructions to expedite trial and to attempt to narrow its preliminary injunction.

FACTUAL AND PROCEDURAL BACKGROUND

In 2007, a dietary supplement company called Endomatrix filed for bankruptcy. One of its researchers, Dr. Daniels, purchased Endomatrix’s intellectual property out of the bankruptcy estate. This property included the trademark for a dietary supplement named Provasca and the leftover inventory and raw materials for the supplement. Dr. Daniels approached his brother, David, and a marketing executive, Robert Long, in 2010 to help him secure funding for further research and marketing of Provasca. They formed two entities in 2011: DHS to research Provasca and VHS to market the supplement. Dr. Daniels compiled a distilled version of the science and research behind Provasca into a PowerPoint presentation titled “The Path to Provasca” to educate David Daniels and Long on the supplement and help them communicate its potential to investors. Dr. Daniels required that VHS receive his approval before it presented potential investors with information on the supplement, including The Path to Provasca, and also that VHS have them sign a Confidentiality and Non-Disclosure Agreement (“CNDA”) prior to the presentation. He also prohibited VHS from presenting information on the supplement to venture capital groups and others that might seek to use it independently. Moreover, according to DHS’s complaint, - before sharing information with VHS, “Dr. Daniels made it clear that the information he was sharing needed to be kept confidential.” If VHS principals had not agreed to those terms, “Dr. Daniels wouldn’t have shared [the information] with them.”

A year after assembling The Path to Provasca, Dr. Daniels also had VHS sign a CNDA. DHS was less successful, however, in reaching a licensing agreement with VHS, and in February 2012, VHS called off the deal. Despite severing its ties with DHS, VHS continued to attempt to secure investors for a supplement. It also developed a rival product, Arterosil. Provasca and Arterosil both contain the same seaweed extract alleged to help repair and maintain blood vessel walls.' But Provas-ca’s other ingredient is L-Arginine, an amino acid, while Arterosil has a blend of fruit and vegetable extracts in addition to the seaweed extract.

DHS filed suit, asserting that VHS violated the CNDA by using confidential information to develop and market Arterosil. It also alleged misappropriation of trade secrets and trademark violations, among several other claims. It requested and obtained a temporary restraining order to prevent VHS from using DHS’s confidential information to research or sell Artero-sil. It then filed for a preliminary injunction. Following briefing, the district court heard two days of testimony on the merits of the injunction. Three weeks later, the court granted the injunction. The court concluded, based on its factual findings, that DHS was substantially likely to succeed on its breach of contract and trade secret misappropriation claims. It went on to determine that DHS faced a substantial, threat of irreparable injury absent an injunction, which outweighed the harm that might result if the injunction was granted. Finally, it concluded that an injunction would not disserve the public interest. VHS moved to clarify the scope of the injunction, and the court denied the motion.

VHS appeals, arguing that this court should vacate the injunction because DHS did not sustain its burden of proof, and because the injunction covered activities that did not involve the use of DHS’s alleged confidential information. A panel of this court granted VHS’s motion to stay *582the preliminary injunction pending appeal and expedited the appeal.

STANDARD OP REVIEW

Reviewing a district court’s grant of a preliminary injunction, this court asks “whether the .issuance of the injunction, in the light of the applicable standard, constitutes an abuse of discretion.” Concerned Women for Am., Inc. v. Lafayette Cnty., 883 F.2d 32, 34 (5th Cir.1989) (citation and alterations omitted). “[Fjindings of fact that support the district court’s decision are examined for clear error, whereas conclusions of law are reviewed de novo.” Affiliated Prof'l Home Health Care Agency v. Shalala, 164 F.3d 282, 284-85 (5th Cir.1999).

DISCUSSION

VHS asks the court to vacate the district court’s preliminary injunction. It contends that DHS did not offer proof that it met the standard for a preliminary injunction. VHS also argues that the court did not limit the injunction to use of DHS’s alleged confidential information.

I. The district court’s grant of the preliminary injunction

VHS first argues that the district court erred when it found that DHS carried its burden of proof for each of the four requirements for a preliminary injunction: substantial likelihood of success on the merits, substantial threat of irreparable harm absent an injunction, a balance of hardships in DHS’s favor, and no disservice to the public interest. Byrum v. Landreth, 566 F.3d 442, 445 (5th Cir.2009). Because the district court found sufficient facts to support its conclusion that the four requirements were met, however, it did not abuse its discretion by granting the injunction.

A. Likelihood of success on the merits

VHS argues that the district court erred in determining that DHS was likely to succeed on its breach of contract and misappropriation of trade secrets claims. To show a likelihood of success, the plaintiff must present a prima facie case, but need not prove that he is entitled to summary judgment. See Janvey v. Alguire, 647 F.3d 585, 595-96 (5th Cir.2011). With respect to the contract claim, VHS first contends that DHS did not show damages. It is true that the court did not specifically discuss damages in the section of its opinion on the merits of DHS’s contract claim. But it later expressly observed that VHS’s alleged breach and misappropriation would impair Provasca’s chances at FDA approval for medical use and make it difficult for DHS to secure research funding for the drug. These findings are sufficient to support the court’s conclusion that DHS likely will be able to prove damages at trial.

Next, VHS argues that the district court did not identify the confidential information that VHS allegedly used in breach of its agreement with DHS. But the court found that “Dr. Daniels disclosed [in meetings with VHS] new scientific and clinical research, some of which he found in the public domain, but had compiled and distilled.” It also found that “the compilation was not public knowledge” and that Dr. Daniels “later expanded that compilation into a presentation entitled ‘The Path to Provasca.’ ” The court went on to explain:

The evidence shows that after learning the science and reviewing the compilation of research that Dr. Daniels generated and produced for fundraising, [VHS] and its operatives set out to de-legitimatize Provasca and produce their own product, all without notice or [DHS’s] permission. The CNDA prohibits this conduct; hence, the Court is of the view that a factfinder will con-*583elude that [VHS] breached its agreement with [DHS].

These findings specifically identify the confidential information that VHS misused and how it misused that information in breach of the CNDA. Although, under VHS’s interpretation of the CNDA, the agreement does not cover this information, the district court did not err in determining that the ultimate factfinder would likely agree with DHS that the information is covered. In particular, the district court noted that the CNDA broadly includes “all confidential or proprietary written, recorded, electronic or oral information ... (whether such confidentiality or proprietary status is indicated orally or, whether or not the specific words ‘confidential’ or ‘proprietary’ are used).”

VHS also argues that DHS did not offer sufficient evidence for the district court to conclude that DHS likely would succeed on its trade secrets claim. To establish trade secret misappropriation in Texas, a plaintiff must show “(a) the existence of a trade secret; (b) a breach of a confidential relationship or improper discovery of the trade secret; (c) use of the trade secret; and (d) damages.” Taco Cabana Int’l, Inc. v. Two Pesos, Inc., 932 F.2d 1113, 1123 (5th Cir.1991) (citation omitted) aff'd sub nom. Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 112 S.Ct. 2753, 120 L.Ed.2d 615 (1992). VHS contends that the first requirement was not met because neither DHS nor the district court specifically identified the alleged trade secret, and, in any event, DHS had no trade secrets because all of its research on the cardiovascular benefits of seaweed extract was public knowledge. This argument is largely repetitive of VHS’s charge that the confidential information was not identified in the breach of contract claim because no such information existed. But in this case, the court looks to the legal definition of a trade secret rather than the broad contractual definition of “confidential information” in the CNDA.

“A trade secret is any formula, pattern, device or compilation of information used in one’s business, and which gives an opportunity to obtain an advantage over competitors who do not know or use it.” Id. (citing Hyde Corp. v. Huffines, 158 Tex. 566, 314 S.W.2d 763, 776 (1958)) (emphasis added). Under Texas law, this trade secret determination is made by weighing six factors: “(1) the extent to which the information is known outside of the business; (2) the extent to which it is known by employees and others involved in the business; (3) the extent of measures taken to guard the secrecy of the information; (4) the value of the information to the business and to its competitors; (5) the amount of effort or money expended in developing the information; (6) the ease or difficulty with which the information could be properly acquired or duplicated by others.” Tewari De-Ox Sys., Inc. v. Mountain States/Rosen, L.L.C., 637 F.3d 604, 610 (5th Cir.2011) (quoting In re Union Pac. R.R. Co., 294 S.W.3d 589, 592 (Tex.2009)). Although “[information that is public knowledge or that is generally known in an industry cannot be a trade secret,” this court has “specifically rejected the contention that a combination of disclosed technologies cannot itself constitute a trade secret.” Id. at 611, 613 (citations and internal quotation marks omitted).

As with the breach of contract claim, the district court specifically identified Dr. Daniels’s “compilation [that] was not public knowledge” of “new scientific and clinical research, some of which he found in the public domain,” but which he had “compiled and distilled” and later expanded into The Path to Provasca. The district court did not err when it concluded that this compilation of information, consisting par*584tially of disclosed technologies, nevertheless met the standard for a trade secret under Texas’s six-factor test. Although it did not analyze each factor, the court found that (1) the compilation “was not public knowledge”; (2) the compilation consisted of “Dr. Daniels’ concept and his work behind Provasca”; (3) “[b]efore sharing his confidential information with his brother and others, Dr. Daniels insisted that the information that he shared with them be kept confidential”; and (4) it had taken Dr. Daniels “many years of scientific research” to develop the information. These findings are sufficient to support the district court’s determination that the compilation is a trade secret.

Finally, VHS challenges the district court’s conclusions that VHS and DHS had a confidential relationship and that VHS used the alleged trade secret to develop and market Arterosil. VHS claims that it had no confidentiality obligation when DHS disclosed The Path to Provasca because there was no oral agreement on confidentiality in place at that time. However, under Texas law, an agreement is not required to prove the existence of a confidential relationship. “[A]n express agreement [is] not necessary where the actions of the parties, the nature of their arrangement, the ‘whole picture’ of their relationship established the existence of a confidential relationship.” Furr’s Inc. v. United Specialty Adver. Co., 385 S.W.2d 456, 459 (Tex.App.-El Paso 1964, writ ref'd n.r.e.) (citing Huffines, 158 Tex. 566, 314 S.W.2d 763); see also H.E. Butt Grocery Co. v. Moody’s Quality Meats, Inc., 951 S.W.2d 33, 35-36 (Tex.App.-Corpus Christi 1997).

There is sufficient evidence in the pleadings and exhibits to present a prima facie case that VHS should have understood that the information DHS provided was confidential. As stated above, the district court found that “[b]efore sharing his confidential information with his brother and others, Dr. Daniels insisted that the information that he shared with them be kept confidential. At least, this was the agreement between Dr. Daniels, David Daniels and Bob Long.” Whenever VHS sought presentation materials from Dr. Daniels, he specified that presentations could only be made to potential investors and not others who might use the information for their own ends. Further, Dr. Daniels required VHS to obtain his approval before each use of the information, and to have meeting attendees sign a nondisclosure agreement. Dr. Daniels’s sharing of the information was contingent on VHS’s agreement with these conditions. This evidence is sufficient to constitute a prima facie case that DHS and VHS had established a confidential relationship — regardless of the absence of an express agreement.

With regard to VHS’s use of the alleged confidential information, VHS argues that it “independently” developed Arterosil, and, even if it did not, DHS must show that VHS used the entire trade secret compilation, not merely pieces of publically available information within it. Despite these claims of independence, the court found that “[w]ithin a few months” of breaking ties with DHS, VHS was able to develop a “rival product” and “make the same essential claims [about it] that were attributed to Provasca.” Furthermore, in its view, “[t]he evidence shows that after learning the science and reviewing the compilation of research that Dr. Daniels generated and produced for fundraising, [VHS] and its operatives set out to delegi-timatize Provasca and produce their own product, all without notice or [DHS’s] permission.” These findings, although weaker, are still enough to support the court’s conclusion that VHS used the compilation itself to develop Arterosil. Because DHS need only present a prima facie case, rath*585er than meet the standard for summary judgment, to establish a likelihood of success on the merits, the district court’s findings are sufficient. It did not err when it determined that DHS satisfied this element of the preliminary injunction test.

B. Irreparable harm

To satisfy this prong of the preliminary injunction test, DHS must show that it is “likely to suffer irreparable harm,” that is, harm for which there is no adequate remedy at law. Winter v. Natural Res. Def. Council, Inc., 555 U.S. 7, 20, 129 S.Ct. 365, 172 L.Ed.2d 249 (2008); see Janvey, 647 F.3d at 600. VHS’s only argument on this element of the test is that DHS has not shown that irreparable harm is likely rather than merely possible. VHS contends that the claim of reputational harm is “speculative.” It is true that “[speculative injury is not sufficient; there must be more than an unfounded fear on the part of the applicant.” Holland Am. Ins. Co. v. Succession of Roy, 777 F.2d 992, 997 (5th Cir.1985). But Dr. Daniels testified that other scientists in the field were familiar with his work and that a poor knock-off would be associated with him to. his reputational detriment. The district court did not err when it credited this unrebutted testimony and concluded that such harm was likely. VHS also argues that DHS presented no evidence that VHS’s alleged knock-off supplement would threaten funding opportunities for DHS’s Provasca. As with the issue of reputational harm, however, the court heard testimony that an ill-conceived or even unsafe knock-off likely would damage DHS’s funding chances. Crediting that testimony was not erroneous.

C. Balance of hardships

Against these harms, the court weighed the money that VHS has spent to market and sell Arterosil. It observed that these damages were compensable, that the market for the supplement would not go away, and that there was no evidence of substantial contracts between VHS and supplement vendors. In light of these findings, the court did not err when it concluded that these harms did not outweigh the likely irreparable injury to DHS absent an injunction.

D.Disservice to the public interest

Finally, VHS argues that the injunction disserves the public interest by barring public access to a supplement with significant potential health benefits and prohibiting VHS from turning a profit that it could use for further health research. But the district court found that the public had a greater interest in the development of “a historical scientific breakthrough” if Dr. Daniels had the chance to substantiate his early research. The odds of this breakthrough, in the district court’s opinion, would be seriously harmed if VHS’s sale of Arterosil hurt DHS’s funding chances. The district court also found that the public is served when the law is followed. It did not err by placing more weight on these latter public interests than on VHS’s proffered public interests.

The district court did not clearly err in any of its factual determinations, and it correctly applied the legal standard for issuance of a preliminary injunction to those facts. Therefore, it did not abuse its discretion in granting the injunction.

II. The scope of the district court’s preliminary injunction

VHS also argues that the preliminary injunction is overbroad because it prohibits “the use, dissemination, destroying, selling, conveying, or distributing of any information and/or intellectual property that [VHS] and operatives received from [DHS].” (Emphasis added.) It also objects that enjoining it “from marketing, selling, *586advertising, distributing, or conveying any product bearing the word ‘Provasca’ or derivatives of that term; or product based on the science received and reviewed” is too broad. VHS points out that this language would prohibit it from disseminating copies of public third-party journal articles that Dr. Daniels included when he compiled The Path to Provasca. In addition, VHS alleges that the injunction bars it from marketing or selling drugs unrelated to Provasca, such as cholesterol-lowering medications, if they are nevertheless based on “the science received and reviewed” from DHS on general cardiovascular health.

The district court’s order granting the injunction must “state its terms specifically” and “describe in reasonable detail” the conduct restrained or required. Fed.R.CrvP. 65(d). Furthermore, the court “must narrowly tailor an injunction to remedy the specific action which gives rise to the order.” John Doe # 1 v. Venenan, 380 F.3d 807, 818 (5th Cir.2004).

The issue of whether the injunction complies with Rule 65 presents a close question. Even though VHS asked the district court to clarify the injunction, VHS has never indicated its desire to disseminate third-party journal articles or market cholesterol drugs. As the Supreme Court has noted,

If defendants enter upon transactions which raise doubts as to the applicability of the injunction, they may petition the court granting it for a modification or construction of the order. While such relief would be in the sound discretion of the court, we think courts would not be apt to withhold a clarification in the light of a concrete situation that left parties ... in the dark as to their duty toward the court.

Regal Knitwear Co. v. NLRB, 324 U.S. 9, 15, 65 S.Ct. 478, 89 L.Ed. 661 (1945) (emphasis added); cf. Gulf King Shrimp Co. v. Wirtz, 407 F.2d 508, 517 (5th Cir.1969) (“If for some reason Gulf King had doubts about the meaning of any part of the injunction, it could have sought district court clarification.”). VHS has not alleged an intent to enter into such transactions. And it is impossible for courts to craft injunctions that address all hypotheticals.1 Nevertheless, because the injunction is quite broad relative to the “reasonably detailed and sufficiently specific to the underlying action” standard, we instruct the district court on remand to try to narrow the scope of its injunction.

CONCLUSION

The district court granted DHS’s request for a preliminary injunction after making sufficient findings of fact to support each element of the analysis and applying the correct legal standard to those facts. Therefore, we AFFIRM the district court’s grant of a preliminary injunction in full and lift the stay of the injunction. We further remand the case and direct the district court to expedite trial on the permanent injunction and to attempt to narrow the breadth of its preliminary injunction.

OPINION

WILLIAMS, Circuit Judge:

Comprehensive Technologies International, Inc. (CTI), brought this action for copyright infringement against former employees Dean Hawkes, Igor A. Filippides, Randall L. Sterba, Richard T. Hennig, and David R. Bixler (the Defendant employees). CTI also named as defendants Alvan S. Bixler and Software Artisans, Inc. (SA), a corporation formed by Alvan Bixler and several of the Defendant employees shortly after their departure from CTI. CTI contended that “Transend,” a computer program developed by the Defendants, infringed upon the copyrights CTI held in its “Claims Express” and “EDI Link” computer programs. CTI appended numerous state law causes of action, including trade secret misappropriation, breach of confidentiality, and breach of contract.1 CTI also alleged that Hawkes *733breached his covenant not to compete with CTI by performing services for SA, soliciting CTI’s customers, and hiring CTI’s former employees. After a bench trial, the district court entered judgment for the Defendants on all counts.

CTI now appeals and offers four reasons for reversal. First, CTI contends that, in granting judgment for Defendants on its copyright infringement claim, the district court failed to apply the appropriate standard for determining whether Transend is substantially similar to Claims Express and EDI Link. Second, CTI contends that under the Virginia Uniform Trade Secrets Act, Va. Code Ann. § 59.1-336 to -343 (Michie 1992), the district court erred in finding that Defendants did not misappropriate its trade secrets. Third, CTI challenges the district court’s conclusion that Hawkes’s covenant not to compete with CTI is unreasonable and hence unenforceable under Virginia law. Last, CTI charges that the district court committed reversible error by expressing “bias” against its software.

With regard to CTI’s copyright infringement and trade secret misappropriation claims, we affirm the judgment for the Defendants. We agree, however, with CTI’s contention that Hawkes’s covenant not to compete is enforceable. We therefore vacate the judgment for Hawkes on CTI’s claim for breach of contract, and remand for the district court to determine whether Hawkes has breached his covenant not to compete. We find CTI’s charge of bias to be without merit.

I.

Factual Background

CTI, a California corporation with its principal place of business in Chantilly, Virginia, is engaged in defense related services. CTI was founded in 1980 by Celestino Beltran, who at the time of trial served as CTI’s president, chief executive officer, and chairman of the board of directors. By 1988 Beltran was hoping to diversify CTI’s operations into newly emerging technologies. At that time, Beltran’s next door neighbor, Al-van Bixler, worked for the Electronic Data Interchange Association (EDIA). Electronic data interchange, or EDI, is the eomputer-to-computer transmission of business transactions in proprietary or standard formats. After discussing EDI with Alvan Bixler and conducting his own research on the subject, Beltran concluded that EDI technology presented substantial growth potential in the small business market.

With the approval of his board of directors, Beltran established a Software Products Group and designated Dean Hawkes to lead it. Hawkes was given the responsibility to design, develop, test, and market software that would enable clients to process and transmit data through EDI technology. CTI selected Igor Filippides as the Software Products Group’s Acting Vice President for Sales and gave him primary responsibility for marketing the software. Other members of the Software Products Group included Sterba, Hennig, and David Bixler, who together wrote the actual software. To boost the marketability of its products, CTI obtained an agreement from EDIA to assist CTI in the development of its software. Pursuant to that agreement, Alvan Bixler collaborated with CTI as a consultant on EDI technology.

Each of the Defendant employees except Hawkes signed CTI’s standard Confidentiality and Proprietary Information Agreement. Under the Agreement, each employee agreed not to disclose or use, directly or indirectly, during his employment and for three years thereafter any confidential, proprietary, or software-related information belonging to CTI. The Agreement specifically identified the Claims Express and EDI Link projects as confidential. Although Hawkes did not sign a Confidentiality and Proprietary Information Agreement, he did sign an Employment Agreement that contained similar but more restrictive provisions. In addition to promising confidentiality, Hawkes agreed that during the term of his employment he would not compete with CTI, solicit CTI’s *734customers, or employ CTI’s current or former employees.

The Software Products Group undertook to develop two software packages for personal computers. The first, Claims Express, is an electronic medical billing system. Claims Express transmits information that conforms to two specific insurance claims forms, the “HFCA 1500” and the “UB 82.” The program has been successfully marketed. CTI’s second software package, EDI Link, is not specific to the health care industry. It is designed to permit users to create generic forms, enter data on the forms electronically, test that data for errors, and store both the forms and the data on a computer. Although CTI expended substantial effort on EDI Link, at the time of trial the program had not been completed and had never been sold or marketed. Trial testimony indicated that between 35 and 85 percent of the program had been completed.

In February 1991, all of the Defendant employees left CTI. Hawkes executed a formal Termination Agreement with CTI. In that Agreement, Hawkes agreed to rescind his Employment Agreement in return for $50,000 and more than $20,000 worth of equipment. Hawkes also agreed that he would not disclose or use CTI’s confidential information, and that, for a period of one year following his departure, he would not (1) compete with CTI, (2) solicit CTI’s customers, or (3) hire CTI’s employees.

In April 1991, the Defendants incorporated Software Artisans, Inc., located in Fairfax, Virginia. By July 1991, SA had developed and begun to market its own program called Transend. According to its User’s Manual, Transend creates a “paperless office environment” by enabling its users to process business forms on a computer. (J.A. at 1592.) Transend is similar to Claims Express and EDI Link in that it is designed to prepare forms for transmission by EDI. Transend permits the user to input data, check the data for errors, and prepare the data for transmission by EDI.

II.

Copyright Infringement

The district court concluded that Defendants did not infringe upon CTI’s copyrights in Claims Express and EDI Link. See 17 U.S.C.A. § 501(a) (West Supp.1993) (defining copyright infringement). The court found that Transend was not a literal copy of either Claims Express or EDI Link, see Computer Assocs. Int’l, Inc. v. Altai Inc., 982 F.2d 693, 701 (2d Cir.1992) (literal copying can prove infringement), nor was it substantially similar to either program, see Dawson v. Hinshaw Music, Inc., 905 F.2d 731, 782 (4th Cir.1990) (access and substantial similarity can. prove infringement), cert. denied, 498 U.S. 981, 111 S.Ct. 511, 112 L.Ed.2d 523 (1990); Whelan Assocs., Inc. v. Jaslow Dental Lab., Inc., 797 F.2d 1222, 1232 (3d Cir.1986), cert. denied, 479 U.S. 1031, 107 S.Ct. 877, 93 L.Ed.2d 831 (1987). The district court acknowledged that under Whelan CTI could prove infringement by showing substantial similarities between the structure, sequence, and organization of the programs, but found that CTI had not met its burden of proof on that issue. The court accepted CTI’s evidence that the programs shared some characteristics but nonetheless found that the similarities were either derived from common sources available to the average programmer or were dictated by the functions of the programs. The court also found that, to the extent that- CTI may have possessed proprietary algorithms,2 Defendants had not copied any of them.

CTI contends that the district court failed to apply the correct test to determine substantial similarity and therefore erred in concluding that the structure, sequence, and organization of Transend were not substantially similar to that of Claims Express and EDI Link. We review the district court’s factual findings on substantial similarity for clear error, Data East USA, Inc. v. Epyx, Inc., 862 F.2d 204, 206 (9th Cir.1988); Whelan, 797 F.2d at 1233 n. 25, and we review its conclusions of law de novo, Whelan, 797 F.2d at 1233 n. 25.

CTI points out that the district court criticized the Whelan test of substantial similari*735ty as overly simplistic,3 but never stated which alternative test should be used to evaluate substantial similarity. CTI believes that the district court should have applied the Second Circuit’s “abstraction-filtration-comparison” analysis, first announced in the Computer Associates opinion. Computer Assocs., 982 F.2d at 706-12. CTI argues that if the district court had used the Computer Associates test, it would have identified six similarities between Transend and the CTI programs:

(1) the organization of data dictionary in the data base;

(2) the establishment of form data;

(3) the use of the header file name “user-name.rec”;

(4) the unique identifiers in the sort record;

(5) the use of 41 characters for the key length to accommodate names and addresses; and

(6) some of the field names.

(CTI’s Br. at 11.) According to CTI, these similarities prove infringement.

The difficulty with CTI’s position is that it has not identified any evidence in the record indicating that these similarities were proved at trial. We will not sift through the record to piece together support for CTI’s contentions.4 See Fed.R.App.P. 28(a)(4), (e) (appropriate references to the record are required); Friedel v. City of Madison, 832 F.2d 965, 969 (7th Cir.1987) (“It is not the court’s duty on appeal to wade through the record and make arguments for either party.”); AMP, Inc. v. Fleischhacker, 823 F.2d 1199, 1203 (7th Cir.1987) (“We will not search through the record in an endeavor to locate material which the parties have failed to provide us. It is neither the role nor the obligation of an appellate court to cast about in the record for facts and argument upon which the parties may rely.” (Internal quotation marks and brackets omitted).) Because CTI has not identified any evidence demonstrating that the district court clearly erred, we affirm the district court’s finding that Transend was not substantially similar to either Claims Express or EDI Link.5

III.

Trade Secrets

The district court also found that CTI did not prove that the Defendants misappropriated a trade secret. Under Virginia law a “trade secret” is

information, including but not limited to, a formula, pattern, compilation, program, device, method, technique, or process, that:

1. Derives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable by proper means by, other persons who can obtain economic value from its disclosure or use, and

2. Is the subject of efforts that are reasonable under the circumstances to maintain its secrecy.

Va.Code Ann. § 59.1-336; see also Dionne v. Southeast Foam Converting & Packaging, Inc., 240 Va. 297, 397 S.E.2d 110, 113 (1990) (discussing definition of trade secret). For purposes relevant to this case, “misappropriation” means the “use of a trade secret of another without express or implied consent by a person who ... [a]t the time of ... use, knew or had reason to know that his knowledge of the trade secret was ... [ajcquired under circumstances giving rise to a duty to maintain its secrecy or limit its use.” Va. Code Ann. § 59.1-336.

In denying CTI’s claim for trade secret misappropriation, the district court found that CTI did not possess any trade secrets and that, even if CTI did possess trade secrets, the Defendants had not misap*736propriated them.6 The court found no evidence that CTI’s purported trade secrets— the organization of Claims Express and EDI Link, the database access techniques of the two programs, and the unique identifiers of the two programs — derived independent economic value from not being generally known or were not readily ascertainable by proper means. Consequently, the court concluded that CTI’s purported trade secrets failed to satisfy all of the elements necessary to prove a trade secret. The district court also concluded that the Defendants did not “copy” any trade secrets, implying that Defendants did not “use” or otherwise misappropriate them.7 (J.A. at 65.)

CTI argues that in granting judgment for Defendants on its trade secrets claim, the district court misapplied the law. Although we review the district court’s determination that a trade secret does not exist for clear error, Integrated Cash Management Servs., Inc. v. Digital Transactions, Inc., 920 F.2d 171, 174 (2d Cir.1990); Telex Corp. v. International Business Machs. Corp., 510 F.2d 894, 928 (10th Cir.), cert. dismissed, 423 U.S. 802, 96 S.Ct. 8, 46 L.Ed.2d 244 (1975); K-2 Ski Co. v. Head Ski Co., 506 F.2d 471, 477 (9th Cir.1974), we independently review the district court’s conclusions of law and the court’s application of the law to the facts, Rawl v. United States, 778 F.2d 1009, 1014 & n. 9 (4th Cir.1985), cert. denied, 479 U.S. 814, 107 S.Ct. 67, 93 L.Ed.2d 25 (1986).

CTI reads the district court’s opinion as ruling as a matter of law that the organization of its database, its database access techniques, and its unique identifiers could not constitute trade secrets because each of their composite elements was in the public domain. CTI argues vociferously (and correctly) that although a trade secret cannot subsist in information in the public domain, it can subsist in a combination of such information, as long as the combination is itself secret. See Integrated Cash Management, 920 F.2d at 174; Continental Data Sys., Inc. v. Exxon Corp., 638 F.Supp. 432, 443 (E.D.Pa.1986); Q-Co Indus., Inc. v. Hoffman, 625 F.Supp. 608, 617 (S.D.N.Y.1985); Dickerman Assocs., Inc. v. Tiverton Bottled Gas Co., 594 F.Supp. 30, 35-36 (D.Mass. *7371984). According to CTI, each of its alleged trade secrets is just such a combination of publicly available information.

In making this argument, CTI misreads the district court’s opinion. The district court did not rule that unique combinations or arrangements of publicly available information cannot receive protection as trade secrets. Rather, the district court held that CTI failed to present any evidence that its database organization, its access techniques, and its identifiers were not themselves publicly available. The court specifically found that the arrangement and interaction of the functions of Claims Express and EDI Link were “common to all computer programs of this type.” (J.A. at 63.) Information that is generally known cannot qualify as a trade secret. See Va.Code Ann. § 59.1-336 (Michie 1992). Consequently, the district court did not misapply the law; it simply found insufficient evidence to support CTI’s claim. The district court correctly concluded that CTI failed to prove that the organization, database access techniques, and identifiers of CTI’s software constituted trade secrets.

Even if CTI had demonstrated that these items constituted trade secrets, CTI has not convinced us that the district court clearly erred in finding that the Defendants did not misappropriate any of CTI’s alleged trade secrets. CTI points to the short development time and the complete lack of design documentation for Transend as strong circumstantial evidence of misappropriation. Although this evidence does raise some suspicions, Defendants provided a colorable explanation for the absence of design documentation. First, Defendant’s expert, Dr. Ro-tenstreich, testified that it was not atypical for small software companies to neglect to prepare extensive design documentation. Second, Sterba testified that he and the others disliked the amount of paperwork involved in documenting their designs, that they preferred to use a “whiteboard” for their design work, and that they placed much of the information that would ordinarily appear in design documentation in the code itself. In light of this testimony, CTI’s circumstantial evidence is not enough to convince us that the district court clearly erred in finding that the Defendants did not copy (or “use”) any of CTI’s alleged trade secret information. See id. (wrongful use of a trade secret constitutes misappropriation).

We find instructive the Fifth Circuit’s decision in Plains Cotton Cooperative Ass’n v. Goodpasture Computer Service, Inc., 807 F.2d 1256, 1263 (5th Cir.), cert. denied, 484 U.S. 821, 108 S.Ct. 80, 98 L.Ed.2d 42 (1987):

[T]he trade secrets allegedly involved here are particular implementations of software functions.... [T]he misuse of these implementations can occur only through copying the particular software designs on a sufficiently specific level.... [The trade secrets] are matters of design, where the issue of misuse boils down to evidence of copying. If no copying occurred on any level, appellant cannot demonstrate that appellees misused the trade secrets they allegedly possessed.

... If appellees did not in any way “copy” any part of appellant’s protected idea or expression, then appellant cannot demonstrate trade secret misappropriation any more than it can show copyright infringement.

(Emphasis added.) As the district court noted, CTI produced insufficient evidence that Transend copied any unique designs or functions of either Claims Express or EDI Link. Without proof of copying at the functional (or ideational) level, CTI has not proved that Defendants “used” and thereby misappropriated any of its trade secret information. The district court correctly concluded that the Defendants did not misappropriate any trade secret information belonging to CTI.

In sum, we conclude that the district court did not misapply the law and that its findings of fact are not clearly erroneous. We affirm the district court’s entry of judgment for Defendants on CTI’s misappropriation claim.

IV.

Covenant Not to Compete

CTI next argues that the district court should have enforced Dean Hawkes’s *738covenant not to compete. In his Termination Agreement, Hawkes agreed that, for a period of twelve months following his departure from CTI, he would not

engage directly or indirectly in any business within the United States (financially as an investor or lender or as an employee, director, officer, partner, independent contractor, consultant or owner or in any other capacity calling for the rendition of personal services or acts of management, operation or control) which is in competition with the business of CTI. For purposes of this Agreement, the “business of CTI” shall be defined as the design, development, marketing, and sales of CLAIMS EXPRESS) and EDI LINK) type PC-based software with the same functionality and methodology....

(J.A. at 714.) Virginia has established a three-part test for assessing the reasonableness of restrictive employment covenants. Under the test, the court must ask the following questions:

“(1) Is the restraint, from the standpoint of the employer, reasonable in the sense that it is no greater than is necessary to protect the employer in some legitimate business interest?

(2) From the standpoint of the employee, is the restraint reasonable in the sense that it is not unduly harsh and oppressive in curtailing his legitimate efforts to earn a livelihood?

(3) Is the restraint reasonable from the standpoint of a sound public policy?”

Blue Ridge Anesthesia & Critical Care, Inc. v. Gidick, 239 Va. 369, 389 S.E.2d 467, 469 (1990) (citation omitted); Meissel v. Finley, 198 Va. 677, 95 S.E.2d 186, 188 (1956). If a covenant not to compete meets each of these standards of reasonableness, it must be enforced. Roanoke Eng’g Sales Co. v. Rosenbaum, 223 Va. 548, 290 S.E.2d 882, 884 (1982). As a general rule, however, the Virginia courts do not look favorably upon covenants not to compete, Grant v. Carotek, Inc., 737 F.2d 410, 411 (4th Cir.1984), and will strictly construe them against the employer, Clinch Valley Physicians, Inc. v. Garcia, 243 Va. 286, 414 S.E.2d 599, 601 (1992); Grant, 737 F.2d at 411. The employer bears the burden of demonstrating that the restraint is reasonable. Richardson v. Paxton Co., 203 Va. 790, 127 S.E.2d 113, 117 (1962).

The district court refused to enforce the covenant not to compete because it concluded that the covenant was broader than necessary to protect CTI’s legitimate business interests. First, the court held that the scope of the employment restrictions was too broad because the restrictions precluded Hawkes from working for a competitor in any capacity, even as a janitor. The court implied that CTI did not have a legitimate interest in preventing Hawkes from working for a competitor in a menial capacity. Second, the district court concluded that the geographic scope of the agreement was broader than necessary to protect CTI’s interests. The court found that CTI had marketed Claims Express only in Virginia, Nebraska, and perhaps one other state, and therefore CTI did not have a legitimate interest in restricting Hawkes’s employment throughout the United States. (J.A. at 66-67.)

We review the enforceability of the covenant not to compete de novo. Brunswick Corp. v. Jones, 784 F.2d 271, 274 n. 2 (7th Cir.1986).

CTI asserts that under the facts of this case the employment restrictions were reasonably necessary to protect its business interests.

Although the district court believed that the covenant was categorically over-broad because it precluded Hawkes from working for a competitor of CTI in any capacity, the Virginia Supreme Court has enforced similarly broad restrictions. In Roanoke Engineering, the Court enforced a three-year restriction on an employee’s right to “own, manage, operate, control, be employed by, participate in, or be associated in any manner with the ownership, management, operation or control of any business similar to the type of business conducted by” the employer. Roanoke Eng’g, 290 S.E.2d at 882 (emphasis added). In Blue Ridge, the Court upheld a three-year covenant under which the employee could not “open or be employed by or act on behalf of any competitor of Employer which renders the same or *739similar services as Employer”). Blue Ridge, 389 S.E.2d at 468 (emphasis added). The covenant in Hawkes’s agreement properly restricts him from competitive employment that would, in all likelihood, substantially interfere with CTI’s business. See Stoneman v. Wilson, 169 Va. 239, 192 S.E. 816, 819 (1938); cf. Grant, 737 F.2d at 412 (covenant which restrained more than direct competition with the employer was unreasonable); Worrie v. Boze, 191 Va. 916, 62 S.E.2d 876, 881 (1951) (appi’oving use of covenants to prevent injurious competition).

Moreover, as Vice President of CTI’s Software Products Group, Hawkes necessarily came in contact with confidential information concerning both CTI’s products and its customers. Hawkes’s access to such confidential information makes the covenant not to compete more reasonable. As the Virginia Supreme Court has noted,

[t]he fact that the employment is of such a character as to inform the employee of business methods and trade secrets which, if brought to the knowledge of a competitor, would prejudice the interests of the employer, tends to give an element of reasonableness to a contract that the employee will not engage in a similar business for a limited time after the termination of his employment, and is always regarded as a strong reason for upholding the contract.

Stoneman, 192 S.E. at 819 (internal quotations omitted); Meissel, 95 S.E.2d at 191 (possession of trade secrets and confidential information is an “important consideration” in testing the reasonableness of a restrictive covenant); cf. Community Counselling Serv., Inc. v. Reilly, 317 F.2d 239, 244 (4th Cir.1963) (even in absence of covenant not to compete, employee may not appropriate trade secrets and confidential information rightfully belonging to his former employer). Similarly, in Roanoke Engineering, an employee had access to confidential financial records, lists of customers and suppliers, and detailed knowledge of overhead factors, pricing policies, and bidding techniques. Roanoke Eng’g, 290 S.E.2d at 885. The Virginia Supreme Court held that this information enabled the employee to become a “formidable competitor” of his former employer, and concluded that a restriction barring the employee from working for competitors in any capacity was no greater than necessary to protect the employer’s legitimate business interests. Id.

Hawkes poses a similar danger to CTI’s business. As the individual primarily responsible for the design, development, marketing and sale of CTI’s software, Hawkes became intimately familiar with every aspect of CTI’s operation, and necessarily acquired information that he could use to compete with CTI in the marketplace.8 When an employee has access to confidential and trade secret information crucial to the success of the employer’s business, the employer has a strong interest in enforcing a covenant not to compete because other legal remedies often prove inadequate. It will often be difficult, if not impossible, to prove that a competing employee has misappropriated trade secret information belonging to his former employer. Eden Hannon & Co. v. Sumitomo Trust & Banking Co., 914 F.2d 556, 561 (4th Cir.1990) (applying Virginia law). On the facts of this case, we conclude that the scope of the employment restrictions is no broader than necessary to protect CTI’s legitimate business interests.

As a second ground for invalidating the covenant not to compete, the district court concluded that the geographic scope of the employment restrictions — “within the United States” — was greater than necessary to protect CTI’s business. (J.A. at 66.) The district court merely noted that CTI had marketed Claims Express in only three states and therefore did not have a national market for its product.

The district court clearly erred in concluding that CTI did not have a national market for Claims Express. See Fed.R.Civ.P. 52(a) (findings of fact are reviewed for clear error). CTI licensed Claims Express in at least ten states: California, Colorado, Connecticut, Florida, Iowa, Kansas, Maryland, Nebraska, New York, and Oregon. This list alone dem*740onstrates that CTI’s customers were dispersed throughout the country and not concentrated in any particular geographic area. CTI’s operation was neither local nor regional, but national. Other evidence of a national market for Claims Express was similarly compelling. Some of CTI’s customers were value-added resellers who had agreed to market Claims Express to their own customers. These value-added resellers were located in California (BAS Microtech), Colorado (Dakkro Corporation), Connecticut (Robert Austin), Kansas (Computing Services, Inc., and Resource & Development Services), and Nebraska (The Churchhill Group, LTD). CTI also identified for the district court specific customer prospects in Colorado, Connecticut, Georgia, Illinois, Indiana, Kansas, Maryland, Massachusetts, Minnesota, Missouri, Nebraska, New Hampshire, New Jersey, New York, Oregon, Pennsylvania, Tennessee, Texas, Washington, and the District of Columbia.9 CTI presented Claims Express and EDI Link (albeit in preliminary form) at national EDIA trade shows in both 1989 and 1990. Finally, CTI presented evidence that it faced direct competition from companies located in California, Colorado, Georgia, Idaho, Illinois, Indiana, Kansas, Maryland, Michigan, Minnesota, New Jersey, Ohio, Oregon, South Carolina, Texas, Utah, and Virginia, and that it faced potential competition from companies in Arizona, California, Georgia, Maryland, North Dakota, Ohio, Oklahoma, Tennessee, and Texas. Given the breadth of the market for Claims Express, we cannot see how anything less than a nationwide prohibition could conceivably protect CTI’s business interests. Because CTI had a national market for its product, the restrictions on Hawkes’s employment throughout the United States were no greater than necessary to protect it from competition by Hawkes. See Roanoke Eng’g, 290 S.E.2d at 885 (restriction geographically coterminous with territory in which employer did business was reasonable); see also National Homes Corp. v. Lester Indus., Inc., 404 F.2d 225, 227 (4th Cir.1968) (under Virginia law, injunctive relief against former employee should be extended to entire state, even though employer had record of sales in only widely scattered sections of the state); cf. Alston Studios, Inc. v. Lloyd V. Gress & Assocs., 492 F.2d 279, 283 (4th Cir.1974) (invalidating restrictive covenant because it did not limit its applicability to appropriate “areas of possible competition” (emphasis added)). CTI fully satisfied the first test of reasonableness.

Having determined that the covenant not to compete is reasonable from CTI’s point of view, we must next determine whether the covenant is reasonable from Hawkes’s point of view, i.e., whether the curtailment on Hawkes’s ability to earn a living is unduly harsh or oppressive. Although the agreement applies throughout the United States, it restricts Hawkes from engaging in only an extremely narrow category of business. Hawkes may not render personal services to, or perform acts of management, operation, or control for, any business in competition with “the business of CTI,” which the agreement defines as “the design, development, marketing and sales of CLAIMS EXPRESSTM and EDI LINKTM type PC-based software with the same functionality and methodology.” (J.A. at 714.) The agreement therefore permits Hawkes to design, develop, market and sell any software of a type different from Claims Express or EDI Link, any software of the same type having a different functionality or methodology, or any software of the same type having the same functionality and methodology that is not designed to run on personal computers. Hawkes is also free to compete with any other branch of CTI’s business. Because Hawkes retains broad employability under the agreement, the agreement is not unduly harsh or oppressive. See Blue Ridge, 389 S.E.2d at 469.

In light of the foregoing, we conclude that the covenant not to compete is no greater than necessary to protect CTI’s business and is not unduly harsh or oppressive. Hawkes does not suggest, and we do not find, that the covenant is unreasonable from the standpoint of public policy. We therefore hold that the covenant is enforceable.

*741Because the district court found the covenant unenforceable, it did not reach the question whether Hawkes violated the covenant by participating in the development of Tran-send. Consequently, we vacate the judgment of the district court for Dean Hawkes on this issue and remand for a determination whether Hawkes has breached his agreement.

V.

Bias

Finally, CTI contends that the district court committed reversible error by expressing “bias” against CTI’s software in a footnote to its opinion.10 CTI speculates that the district court’s failure to appreciate the value of computers may have predisposed it to grant judgment against CTI. This argument clearly has no merit.

VI.

Conchision

In sum, we affirm the judgment of the district court for the Defendants on CTI’s claims for copyright infringement and trade secret misappropriation. We vacate the judgment of the district court for Dean Hawkes on CTI’s claim for breach of the covenant not to compete and remand for further proceedings.

AFFIRMED IN PART, VACATED IN PART, AND REMANDED.

MURNAGHAN, Circuit Judge,

concurring in part and dissenting in part:

While I fully concur with the majority opinion insofar as it disallows recovery on a copyright or trade secrets basis, I reluctantly come to another conclusion with respect to whether CTI, as employer, could enforce as reasonable and not unduly harsh or oppressive the Hawkes covenant not to compete. The covenant not to compete held valid by the majority is operable “within the United States.” The district court found that a reasonable covenant would restrict competition only in “Virginia, Nebraska and perhaps one other state.” The majority has enhanced CTI’s claim to proof of reasonableness by naming 31 states in which CTI has licenses, clients, or potential clients, but that still leaves 19 others, every one of which, it seems to me, is “within the United States.” While the majority characterizes CTI’s business as “national,” it has provided no justification for calling it all inclusive.

The question of validity or not of the non-compete undertaking is one to be decided by the law of the Commonwealth of Virginia. The decision announced by the majority, that a company with business in only 31 states may enforce a non-compete clause in all 50, is a mathematically dubious one on a Virginia point of law. However, the same decision, if announced by the Virginia Supreme Court, would carry more authority than any decision on the point announced by the Fourth Circuit. See Commissioner v. Estate of Bosch, 387 U.S. 456, 465, 87 S.Ct. 1776, 1782, 18 L.Ed.2d 886 (1967) (applying the rule of Erie R. Co. v. Tompkins1 in a nondiversity case when underlying substantive rule is based on state law because “the State’s highest court is the best authority on its own law.”). I feel and have suggested that the question should be certified to the Supreme Court of Virginia. Unfortunately, my colleagues on the panel feel otherwise. Hence, I must make as *742educated a guess as possible as to what the Virginia law is.2

A restraint on an employee is unreasonable if it is greater than is necessary to protect the employer in its legitimate business interest and unreasonable from the employee’s standpoint as unduly harsh in curtailing his legitimate efforts to earn his livelihood. Richardson v. Paxton Co., 203 Va. 790, 795, 127 S.E.2d 113, 117 (1962). “Whatever [the employer’s] subjective intent, we must give effect to the language of the agreement, strictly construed.” Linville v. Servisoft of Virginia, Inc., 211 Va. 53, 55, 174 S.E.2d 785, 787 (1970).

Because restraints of trade are disfavored in Virginia, we must give effect to the language of the agreement, strictly construed. See Linville v. Servisoft of Virginia, Inc., 211 Va. 53, 55, 174 S.E.2d 785 (1970); Richardson v. Paxton Co., 203 Va. 790, 127 S.E.2d 113 (1962). We construe the agreement, reading it literally and construing it favorably to the employee, as an attempt to impose a post-employment restraint upon Gress [employee] without geographic or other limitation. We must therefore decline Alston’s [employer’s] invitation to read into the agreement limitations which simply are not there.

Alston Studios, Inc. v. Lloyd V. Gress & Associates, 492 F.2d 279, 285 (4th Cir.1974). “Conceivably the non-competition clause could be interpreted more narrowly, but Virginia law requires that it be strictly construed against the employer....” Grant v. Carotek, Inc., 737 F.2d 410, 412 (4th Cir.1984).

The Supreme Court of Virginia has never approved a non-compete clause that restricts employment “within the United States.” To the contrary, Virginia courts have repeatedly held that non-compete clauses should be limited to a geographical area no greater than is necessary to protect the employer’s legitimate business interests. Where Virginia courts have enforced non-compete contracts, the contracts have restricted competition only within “quite narrow and well defined geographic limitations.” Alston Studios, 492 F.2d at 283 n. 5. See, e.g., Blue Ridge Anesthesia & Critical Care, Inc. v. Gidick, 239 Va. 369, 371, 389 S.E.2d 467, 469 (1990) (covenant that prohibited employment only in territories serviced by the former employee, not in the company’s entire market area, enforced); Roanoke Eng’g Sales Co., Inc. v. Rosenbaum, 223 Va. 548, 552, 290 S.E.2d 882, 884 (1982) (three year restriction that broadly limited an employee’s right to be employed by any similar business enforced; court specifically noted that the restriction was reasonable because it was geographically co-terminous with the territory in which the employer did business, which involved only two states); Meissel v. Finley, 198 Va. 577, 581, 95 S.E.2d 186, 190 (1956) (covenant with restrictions that applied only within a radius of 50 miles of Norfolk enforced); Worrie v. Boze, 191 Va. 916, 922-26, 62 S.E.2d 876, 879-81 (1951) (covenant that restricted competition within 25 miles of dance studio enforced); Stoneman v. Wilson, 169 Va. 239, 245, 192 S.E. 816, 818 (1938) (agreement of employee not to go into the hardware business for five years within a limited geographical radius enforced); Power Distribution, Inc. v. Emergency Power Eng’g, 569 F.Supp. 54 (E.D.Va.1983) (non-compete contract that restrained former employee from employment with anyone in competition with employer found too broad because area in which plaintiff competed was not fixed and thus the limitation could extend “to every location where plaintiff might potentially compete, which included at least the entire United States”).

Perhaps the Virginia Supreme Court would agree that the noncompete clause ap*743plicable to the employee, Hawkes, was reasonable, but in doing so it would have severely to limit, curtail or even contradict what it has said before. I do not accept that we are free to treat a controlling state rule of law applied by the highest court in the Commonwealth of Virginia so cavalierly.

Accordingly, I would hold the non-compete clause overbroad and hence invalid. I would uphold the district court throughout. So to that extent I dissent.