Clyde A. Farris developed the first successful, fully automatic system for purifying water for industrial purposes. He sold this system through Water Services, Inc. and Farris Chemical Company under the name 'TREAT-A-MATIC'.1

This diversity case involves the validity of the defendant Philip S. Glad's covenant not to compete against his former employer, Farris Chemical Company. Contrary to the district court's holding, we hold that under Georgia law the covenant was a reasonable restraint. We hold also that the defendants, Glad and his present employer, Tesco Chemicals, Inc., misappropriated Farris's trade secrets: Glad breached his confidential relationship with Farris by disclosing to Tesco the design and composition of the TREAT-A-MATIC system and the identity of the suppliers of the components Farris used in producing the TREAT-A-MATIC. Glad will not be allowed to bite the hand that fed him his expertise.

Since 1953 Farris has engaged in the business of supplying chemicals and equipment for the treatment of water circulated through boilers and large air conditioning systems such as are found in office buildings. The purpose of the treatment is to maintain this water in as near a pure form as possible to minimize injury to the circulating pipes, storage tanks, and other parts of the heating or cooling system caused by corrosion, scale, and algae, wasting as small an amount of water as possible.

The water purification industry is highly competitive. There are between 100 and 150 companies in the business in the southeastern area of the country, with 20 to 25 in Atlanta alone. In this industry, as in many others, the increase in labor costs has accentuated the demand for automation.

Farris Chemical has its headquarters in Knoxville, Tennessee, but employs nine or ten field engineers for the southeastern area, one of whom is responsible for sales and services in the State of Georgia. Farris Chemical has had a substantial growth and now has some 25 employees in management, sales, and the mixing or producing of chemicals sold to the trade. The district court found that the company's success is attributable to the ingenuity, industry, and drive of Clyde Farris, president and founder of the company.

The evaporation of water in air conditioning leaves damaging solids in the system. Until 1960, the 'bleeding' of these damaging solids, the testing of acidity and alkalinity by pH content, and the mixture of control chemicals had to be done manually with inferior measuring devices. Various companies experimented with automatic devices to perform these functions, but the first models were expensive and required considerable manpower. In 1960, after four years of trial-and-error experimentation with pumps, timers, and electronic devices, and the expenditure of $18,000 for materials, Farris perfected the TREAT-A-MATIC system.

The district court described the system as made of two basic units: 'one has a 30-minute timing device at 15 second intervals with a low volume pump for the injection of chemicals direct from shipping containers on the 'rust and scale side'; the other has a 24-hour timer with a skipper wheel up to 12 days and high volume pump for the injection of chemicals direct from shipping containers on the 'algae side'. The system is activated by a conductivity device which measures the pH factor for automatic bleeding. All of the devices, together with the electrical control box (which must be specially wired to turn off when the air conditioning system is not in use) are attractively mounted on a single panel board as the TREAT-A-MATIC system.'

Farris did not patent the TREAT-A-MATIC: All of the components of the system were available on the open market. The district court found:

'Its originality lies in the arrangement of the devices in combination and the knowledge of the supply source of the component parts. The functions performed were being performed by similar devices in the trade prior to TREAT-A-MATIC, but not in combination and not in as easy and incumbersome a manner. To capture the chemical sales where the highest profit lies, the TREAT-A-MATIC has been a successful tool. Being fully automated, it appeals to the industry. In addition, by furnishing it on a sale, lease-purchase, or contract basis (wherein the system plus chemicals are furnished for a stated time), Farris has been able to obtain the more lucrative chemical business from a wide range of customers. As stated at trial, Farris undoubtedly 'built a better mouse-trap' and beat the competition to the market place with it.'

Farris protected the composition of the TREAT-A-MATIC system by concealing the identity of the components and their suppliers.2 The company maintained rigid internal security, and required each employee to sign a noncompetition covenant as a condition of employment.

Tesco Chemicals is a direct competitior of Farris. At least as early as April of 1962, Tesco Chemicals, Inc. realized the absolute necessity for automated equipment in order to survive. Tesco began in the business of swimming pool treatment. It hired Mr. Frank Parker to establish and develop an Industrial Division. Parker immediately began to work on automating, but in late 1965 he was writing to component suppliers for any information they could give on equipment 'packages', despite having had a TREAT-A-MATIC system in the Tesco laboratory for detailed inspection and disassembly in early 1965.

In 1963, Farris hired Philip Glad as a field engineer and sales representative. At the time, Farris stressed the highly competitive nature of the industry and the need for tight secrecy on all information concerning equipment, chemicals, and prices. Glad signed a contract containing a two-year non-competition provision covering the northern third of Georgia and a few towns on the Tennessee side of the line. Despite Glad's formal education in electrical engineering, the great bulk of his present expertise was obtained from his experience and training with Farris (for 3 months under close supervision, then seminars three or four times a year). Farris gave Glad the names of the component suppliers, the wiring method, the design detail, and other confidential information. Indeed, he became privy to all construction and sales information concerning the TREAT-A-MATIC system.

After two years with Farris, Glad telephoned Parker and arranged to meet with him to discuss the possibility of Tesco's employing him.3 By that time Parker had experimented with various components in an attempt to create an automated system similar to TREAT-A-MATIC but had met with no success. He had lost several sales for lack of such a system. Parker had seen TREAT-A-MATIC installations at the sites of customers and, as noted earlier, on at least one occasion had disassembled a TREAT-A-MATIC unit. Tesco hired Glad. The district court found that the 'hiring of Glad was with the specific intent of acquiring for Tesco the skill and knowledge necessary to create a fully automated system similar to the TREAT-A-MATIC'. Glad began immediately to work on reproducing for Tesco the TESCOMATIC system, based on his knowledge of the TREAT-A-MATIC system. The district court commented:

The similarity between the two systems is striking. Without going into detail, suffice it to say that they are as near identical as possible for Parker and Glad to construct. The components are the same, with the exception of the conductivity device (the source of which apparently was unknown to Glad) and minor changes in the electrical terminal strip and service switch. The grouping and panel mounting are practically the same.'

On March 1, 1967, Water Services and Farris brought this suit against Tesco and Glad, asking for injunctive relief and damages. The district court, sitting without a jury, entered judgment in favor of the defendants. We reverse and remand.

Philip Glad, like all other Farris employees, signed the following covenant:

It is agreed and understood that for a period of two (2) years after the termination of this contract, regardless of the reason of cause of such termination or cancellation, the Employee will not engage, either directly or indirectly, on his own account or as agent, servant or employee of any person, firm or corporation, in any business which could reasonably be considered to be in competition with Water Services, Inc., in any territory in which he has performed services under this contract.

The threshold question is whether this covenant is reasonable and therefore enforcible under Georgia law. The parties agree that Georgia law controls the construction of the covenant.4 See The Day Companies v. Patat, 5 Cir. 1968, 403 F.2d 792; Budget Rent-A-Car Corp. v. Fein, 5 Cir. 1965, 342 F.2d 509.

As the district court observed, 'The problem of noncompeting coverants in Georgia is one which has caused great difficulty for the courts and practitioners over a long period of time'. Under Georgia law, covenants not to compete are upheld if they are limited as to time and territory and definite as to the nature, kind, and character of the activities prohibited. Bennett v. Georgia Indus. Catering Co., 1966, 222 Ga. 127, 149 S.E.2d 81; Shirk v. Loftis Bros., 1918, 148 Ga. 500, 97 S.E. 66; Rakestraw v. Lanier, 1898, 104 Ga. 188, 30 S.E. 735. Georgia, like most states, has a public policy prohibiting contracts in 'general restraint of trade'. See Georgia Code 20-504. But, 'While public policy forbids any agreement which unreasonably restrains a person from exercising his trade or business, it is equally true that public policy also requires that the freedom of persons to enter into contracts shall not be lightly interfered with'. Rakestraw v. Lanier, 1898, 104 Ga. 188, 30 S.E. 735.

It is undisputed that Tesco hired Glad within the prohibited two-year period and that Tesco competes with Water Services. The district court held that the covenant here was reasonable as to 'time' and 'territory' but that it was invalid and unenforceable on the ground that the provision was 'unreasonably indefinite' as to 'the nature, kind and character of the activities prohibited by the contract'. It 'contemplates', the court said, 'any activity for any competitor whether it relates to the functions and duties of the original employment or not, whether as a truck-driver, bookkeeper, custodian or as an executive or salesman'.

Our review of the Georgia law, including consideration of Baxley v. Black, 1968, 224 Ga. 456, 162 S.Ed.2d 389, decided after the district court rendered its decision in the instant case, compels us to conclude that today Georgia courts would uphold the covenant as a reasonable restraint.

In Rakestraw v. Lanier, 1898, 104 Ga. 188, 30 S.E. 735, still a leading case, the State Supreme Court defined the test for determining the reasonableness of the restraints on activities:

If, considered with reference to the situation, business, and objects of the parties, and in light of all the surrounding circumstances * * * the restraint contracted for appears to have been for a just and honest purpose, for the protection of the legitimate interests of the party in whose favor it is imposed, reasonable as between them, and not specially injurious to the public, the restraint will be held valid.

The contract in Rakestraw involved the sale of a medical partnership. Georgia courts, however, have applied the Rakestraw test to determine the reasonableness of contracts ancillary to employment contracts. See Baxley v. Black, 1968, 224 Ga. 456, 162 S.Ed.2d 389. The Restatement of Contracts applies a similar test to both types of ancillary restraints. Restatement of Contracts 513-15 (1932); cf. Blake, Employee Agreements Not To Compete, 73 Harv.L.Rev. 625, 648-49 (1960).

In determining whether the restraint is reasonable two factors are important. First, is the employer trying to protect confidential information relating to the business, such as a trade secret, method of operation, sources of suppliers, and names of customers? Such confidential business information may or may not rise to the level of a trade secret. Second, is the restraint reasonably related to the protection of the confidential information? See Restatement of Contracts 515(a), 516; Blake, Employee Agreements Not To Compete, 73 Harv.L.Rev. 625 (1960). When these factors have been present, Georgia courts have approved the covenant not to compete even when the covenant defined the 'nature' of the activities only in the broad language of serving a competing employer 'in any capacity'. In Shirk v. Loftis Bros., 1918, 148 Ga. 500, 97 S.E. 66, 67, the Supreme Court of Georgia upheld a covenant restricting an employee from serving 'in any relation or capacity whatsoever, in the line of business (of the employer)'. In Shirk, the employee had become privy to confidential information and methods from the employer concerning repairs of jewelry, watches, and glasses-- information that might be considered as in domain of any person repairing jewelry, watches, and glasses. In Nelson v. Woods, 1949, 205 Ga. 295, 53 S.E.2d 227, the Court upheld a covenant 'not to directly or indirectly enter into any competitive 'do-nut' business'. The purpose of this covenant was to prevent the employee from divulging the recipe used in baking 'Krispy Kreme Do Nuts'. As in the instant case, the originality lay in the combination of readily available components. Similarly in Franco v. Fulton Bakery, Inc., 1940, 190 Ga. 298, 9 S.E.2d 240, the Court approved a covenant of an employee not to 'be employed directly or indirectly * * * as an individual, partner, * * * employee, or in any relation or capacity' even though there was no danger or revelation of a secret process of baking. See also Aladdin, Inc. v. Kranoff, 1958, 214 Ga. 519, 105 S.E.2d 730; Turner v. Robinson, 1959, 214 Ga. 729, 107 S.E.2d 648; Ogle v. Wright, 1939, 187 Ga. 749, 2 S.E.2d 72.

Tesco and Glad rely principally on Mason, etc. v. Jablin, 1964, 220 Ga. 344, 138 S.E.2d 660; Stein Steel & Supply Co. v. Tucker, 1964, 219 Ga. 844, 136 S.E.2d 355, and Dixie Bearings, Inc. v. Walker, 1963, 219 Ga. 353, 133 S.E.2d 338. These cases, however, are distinguishable or not controlling precedent under the Georgia 'full-bench rule'. This is a codified rule of stare decisis: a unanimous decision of the Supreme Court of Georgia may be overruled only by a unanimous decision.5 In Dixie Bearings, a covenant prohibited the employee from engaging in or becoming associated with any competitive business. The Court held that the agreement was unreasonable because of 'the absolute prohibition of the employee's working in any capacity for a competitor in positions unrelated to trade secrets or customers'-- for example, as a truck driver or night watchman. Chief Justice Duckworth specially concurred in Dixie Bearings, and therefore the decision is not binding under the 'full-bench rule'. Compare the language of the district court's conclusions in the case. In Stein Steel, the facts were similar to those in Dixie Bearings. With three justices dissenting, the court held that the covenant was 'larger than were necessary for the protection of the promisee'. The dissenting justices argued that 'the nature of employment is not the controlling factor, since (the employee) might as well make use of sales information, trade secrets, etc., of his former employer in one position as in another'. The disenters noted that Dixie Bearings was 'not binding', because of the special concurrence in that case. In Mason, the court held invalid a covenant forbidding the employee for five years from engaging in the business of manufacturing or selling candy as well as engaging in any fund raising activities similar to the fund raising activities of the employer. The court found that the restraint was unreasonably broad, in part because fund-raising activities were not reasonably related to protecting the former employer's candy business. In Taylor Freezer Sales Company v. Sweden Freezer Eastern Corp., 1968, 224 Ga. 160, 160 S.E.2d 356, the plaintiff alleged 'no facts * * * to show that the defendant became familiar with any trade secrets or confidential information, other than a general knowledge of the plaintiff's business.'

The Georgia Supreme Court's most recent full-bench decision on the issue of covenants not to compete is Baxley v. Black, 1968, 224 Ga. 456, 162 S.E.2d 389. Justice Mobley, author of the opinion, wrote the dissenting opinion in Stein Steel. The court rejected the reasoning of Dixie Bearings and Stein Steel, both cited and discussed in the briefs of the parties. The Court returned to the broad test of reasonableness in Rakestraw. The covenant in Baxley prohibited the employee, who had been hired as a cook from 'directly or indirectly, engag(ing) in the preparation or sale of barbecued meats at any location within a redius of ten (10) miles of any of the locations of the employer'. The defendant argued that the covenant should not bar the employee from later employment as a manager, because there is no trade secret involved in the making of a barbecue which is not known to the public generally and the position of manager is totally unrelated to preparing barbecued meats as a cook. The court rejected this argument:

There is evidence that the appellees employed the appellant as a barbecue cook, that they trained him in their specialized method of cooking, which the appellant admits he had never seen before, that they used special cuts of meat, a secret sauce, special type of operation with indoor pit, etc. The evidence shows further that the appellant has accepted employment as manager of a nearby, newly established, competing barbecue place, where he does some of the cooking, directs how the meats are served, and the operation of the place; and shows further that this new barbecue house was built similar in design to the Old Hickory Houses of the appellees, and that in the operation it has duplicated many features of the Old Hickory House, making the operation very similar in practically all particulars. It is obvious that the appellant is violating the covenant in the contract.

A 'business which could reasonably be considered to be in competition with Water Services' (the language of the covenant) is clearly a business involving the preparation and sale of a system for supplying and maintaining purified water for industrial purposes. Glad certainly understood it. And by any objective standard, the description was as definite as the language in Baxley--'preparation and sale of barbecued meats'. The covenants in both cases prohibited the employee 'either directly or indirectly', on his account or for another, competing with the employee. In both cases the physical features and the methods of operation of the new employer's 'system' were substantially similar to those of the former employer. In both cases, the confidential information was developed as a result of the former employer's initiative and investment and was subject to misappropriation regardless of the employee's activities in his new position-- in Baxley as cook or manager, in the instant case as salesman or engineer.

In short, Glad possessed confidential information revealed to him by Farris concerning the composition of the TREAT-A-MATIC system. This system was not patented, but was nonetheless the industry's only fully automated, integrated system for controlling and supplying purified industrial water. Even though the system had been marketed for a number of years, competitors had been unable to duplicate it. As a result of this competitive advantage, Farris and Water Services had enjoyed substantial financial success. It was reasonable, therefore, for Farris and Water Services to attempt to preserve their limited competitive advantage by keeping the composition of the TREAT-A-MATIC system secret through rigid internal security and elaborate concealment measures. Employee loyalty was especially important since knowledge of the suppliers of the components was a key to the system. In view of the nature of the interest sought to be protected and the competitive environment of the industry, the covenant not to compete meets the Rakestraw test. As stated in Baxley. v. Black, the restraint 'reasonably necessary to protect (a legitimate) interest of the (appellant) in whose favor it is imposed', was for a just and honest purpose, and would not be specially injurious to the public. 149 S.E.2d at 84.

In Part I of this opinion we concluded that the covenant is enforceable, under Georgia law, without discussing whether the interest Farris sought to protect rises to the level of a trade secret. In Part II of this opinion we hold that the design and composition of the TREAT-A-MATIC system (including the identity of suppliers of the components) add up to a protectible trade secret. The fact that Farris sought to protect a trade secret provides a compelling case for finding that the restrictive covenant was reasonable.

The district court took pains to point out that the equities were with Farris:

In connection with this case, the moralities are completely with the plaintiff. Whatever he had, he transmitted to Glad as a trusted employee. Glad subsequently, albeit without any inducement by Tesco, conveyed the information to Tesco regarding the TREAT-A-MATIC. Tesco immediately copied the plaintiff's product through the use of this information. However, the right to free copy is apparently not dependent upon honourable conduct. See Hampton v. Blair Mfg. Company, 8 Cir. 1967, 374 F.2d 969. Here, all of this was done in spite of a solemn contract attempting to prevent the same. However, the sine qua non of such an action is the existence of the trade secret itself. McGraw Edison Co. v. Central Transformer Corp., 8 Cir. 1962, 308 F.2d 70. 12 Milgrim, Business Organizations, 7.07(1). And this prerequisite is necessary regardless of the contract, and of any breach of confidence or not. 2 Callman, Unfair Competition and Trade Marks, 53.3. In this respect, and in this respect only, the plaintiff's claims fail.

No Georgia case goes so far as to say that the existence of a trade secret is the sine qua non of an action to enforce a restrictive covenant ancillary to an employment contract. Although covenants not to compete are proper to protect trade secrets they may also be valid simply to prevent a former employee's using his expertise against his former employer. For example, see Baxley v. Black; Franco v. Fulton Bakery, Inc., 1940, 190 Ga. 298, 9 S.E.2d 240; Nelson v. Woods, 1949, 205 Ga. 295, 53 S.E.2d 227; cf. W. Walley, Inc. v. Saks & Co., 1943, 266 A.D. 193, 41 N.Y.S.2d 739. Conversely, even without an express restrictive covenant, one of the implied terms of a contract of employment is that the employee will not disclose a trade secret learned during his employment, to a competitor of his former employer. But since it may be difficult to determine, as a matter of law, what is a trade secret, the covenant not to compete is a pragmatic solution to the problem of protecting confidential information. See Irvington Varnish & Insulator Co. v. Van Norde, 1946, 138 N.J.Eq. 99, 46 A.2d 201; Kadis v. Britt, 1944, 224 N.C. 154, 29 S.E.2d 543, 152 A.L.R. 405; 2 Callman, Unfair Competition Trademarks and Monopolies 64.4(b), at 641 (1968). Technological refinements in industrial espionage increase the employer's need for covenants not to compete.

An action for breach of a covenant not to compete and an action for misappropriation of a trade secret by breach of a confidential relationship are independent actions. They are related here, however, because an employer has a legitimate interest in protecting a trade secret by means of a restrictive covenant. And the existence of a trade secret bears on the reasonableness of the restraints imposed by the covenant.

The Restatement of the Law of Torts (1939), 757, Comment b defines a trade secret as follows:

'A trade secret may consist of any formula, pattern, device or compilation of information which is used in one's business, and which gives him an opportunity to obtain an advantage over competitors who do not know or use it. It may be a formula for a chemical compound, a process of manufacturing, treating or preserving materials, a pattern for a machine or other device, or a list of customers. * * * A trade secret is a process or device for continuous use in the operation of the business. Generally it relates to the formula for the production of an article. It may, however, relate to the sale of goods or to other operations in the business, such as a code for determining discounts, rebates or other concessions in a price list or catalogue, or a list of specialized customers, or a method of bookkeeping or other office management.'

See also 2 Callman, Unfair Competition and Trademarks, 52.1.

The policy reasons for affording protection to these commercial intangibles are to prevent exploitation by reprehensible business methods and to encourage innovation. See Developments in the Law-- Competitive Torts, 77 Harv.L.Rev. 888, 947-948 (1964). If a trade secret is protected, the competitive advantage realized by the owner of the secret will enable him to recoup his development costs, hopefully before his competitors can 'reverse-engineer' the product and duplicate it.

The district court attached great weight to two 1964 Supreme Court cases, Sears, Roebuck & Co. v. Stiffel Co., 376 U.S. 225, 81 S. Ct. 784, 11 L. Ed. 2d 661 and Compco Corp. v. Day-Brite Lighting, Inc., 376 U.S. 234, 84 S. Ct. 779, 11 L. Ed. 2d 669. These decisions create a more 'liberal right to copy' and raise questions as to the scope of protection a state may afford to federally unpatentable devices. Under the supremacy clause, a state cannot patent a device that is patentable under the federal patent statutes; federal patent laws have preempted the field. But Sears and Compco dealt with the right of a competitor to free copy a product on the open market by independent study. The are not applicable to a situation in which an employer hires an employee of a competitor to take advantage of the employee's expertise in reproducing a complicated system first developed by the former employer. Sears and Compco did not involve trade secrets. Three cases have held them not to have displaced the law of trade secrets. Two of the three cases flatly disregarded Sears and Compco as inapposite to any issue in a trade secret case. Servo Corp. v. General Electric Co., 4 Cir. 1964, 337 F.2d 716, cert. denied, 1966, 383 U.S. 934, 86 S. Ct. 1061, 15 L. Ed. 2d 851, reh. den. 1966, 384 U.S. 914, 86 S. Ct. 1333, 16 L. Ed. 366; Schulenberg v. Signatrol, Inc., 33 Ill. 2d 379, 212 N.E.2d 865, 868, (1965), cert. denied, 383 U.S. 959, 86 S. Ct. 1225, 16 L. Ed. 2d 302 (1966). Hampton v. Blair Mfg. Co., 8 Cir. 1967, 374 F.2d 969, applied Sears, not to deny trade secret status, but to limit the duration of the injunction.

A patent may be distinguished from a trade secret in that a patent is totally exclusionary for the period of time for which it is granted; there are necessarily substantial anticompetitive effects. An ordinary trade secret, however, is not totally exclusionary; it is protected only as long as competitors fail to duplicate it by legitimate, independent research. See F. Machlup, An Economic Review of the Patent System 33 (Study No. 15, Senate Subcomm. on Patents, Trademarks, and Copyrights of the Comm. on the Judiciary, 85th Cong., 2d Sess., Comm. Print. 1958). The anticompetitive effects are relatively slight since the competitive advantage exists only for what is usually a short period, either because the restrictive covenant, if there be one, is for a small number of years (here, only two years) or because a competitor, by reverse-engineering, may legitimately duplicate the device. This limited protection with its relatively slight anticompetitive effects is helpful to insure growth characteristic of innovative genius. See Doerfer, The Limits on Trade Secret Law Imposed by Federal Patent and Antitrust Supremacy, 80 Harv.L.Rev. 1432, 1447-56 (1967). We think that substantial policy considerations militate against construing Sears and Compco as preempting a state's right to protect a trade secret. See Winston Research Corp. v. Minnesota Min. & Mfg. Co., 9 Cir. 1965, 350 F.2d 134; Servo Corp. v. General Elec. Co., 4 Cir., 1964, 337 F.2d 716; Blazon, Inc. v. DeLuxe Game Corp., S.D.N.Y. 1965, 268 F. Supp. 416; Van Products Co. v. General Welding & Fabricating Co., 1965, 419 Pa. 248, 213 A.2d 769.

Georgia law follows the majority view and the Restatement rationale for protecting trade secrets on the ground of a confidential relationship. See Alexis, Inc. v. Werbel, 1953, 209 Ga. 665, 668, 75 S.E.2d 168, 171; Stewart v. Hook, 1903, 118 Ga. 445, 45 S.E. 369, 63 L.R.A. 255. These decisions seem to establish a broad area of protection of trade secrets and suggest that a Georgia court would follow the generally accepted requirements for protection of a trade secret. Basically, these requirements are that the parties view the process or device as a secret and that the secret be revealed in confidence. See Developments in the Law-- Competitive Torts, 77 Harv.L.Rev. 888, 949 (1964).

Tesco urges this Court to impose a requirement of novelty as a prerequisite to protection as a trade secret. The Restatement of Torts, 759, Comment b states:

'A trade secret may be a device or process which is patentable; but it need not be that. It may be a device or process which is clearly anticipated in the prior art or one which is merely a mechanical improvement that a good mechanic can make. Novelty and invention are not requisite for a trade secret as they are for patentability. These requirements are essential to patentability because a patent protects against unlicensed use of the patented device or process even by one who discovers it properly through independent research. The patent monopoly is a reward to the inventor. But such is not the case with a trade secret. Its protection is not based on a policy of rewarding or otherwise encouraging the development of secret processes or devices. The protection is merely against breach of faith and reprehensible means of learning another's secret. For this limited protection it is not appropriate to require also the kind of novelty and invention which is a requisite of patentability.'

Some courts have adopted a requirement of novelty.6 We make an Erie guess that a Georgia court would follow the majority view7 and not require novelty.

The district court found that the components were not developed by Farris, were available on the open market, and were ascertainable. It is clear that the theory of the Treat-A-Matic was not new to the industry. The court also found that the design was apparent from inspection, and that the system had been sold and advertised without restriction. These findings are supported by the evidence. But these findings are not conclusive. The trade secret here was the application of known techniques and the assembly of available components to create the first successful system in the industry. In the trial judge's words, 'no components had been brought together into one lightweight automated system until the perfection of the Treat-A-Matic'. As the Second Circuit stated in Imperial Chemicals, Ltd. v. National Distillers & Chemical Corp., 2 Cir. 1965, 342 F.2d 737, 742, 'a trade secret can exist in a combination of characteristics and components, each of which, by itself, is in the public domain, but the unified process, design and operation of which in unique combination, affords a competitive advantage and is a protectible secret'. This characterization describes the TREAT-A-MATIC. The composition was unique and conferred a substantial competitive advantage on Farris. Indeed, the company enjoyed unparalleled financial success during the four years that competitors were unsuccessfully trying to duplicate the TREAT-A-MATIC.

The parties considered the knowledge of the TREAT-A-MATIC a trade secret revealed in confidence; Farris used elaborate concealment measures to conceal the identity of the components and exacted a covenant not to compete from his employees. That no other competitor had been able to duplicate the TREAT-A-MATIC during the four years it was on the open market is strong evidence that Farris had possession of secrets his competitors could not easily penetrate. The information used in the construction of the TESCOMATIC was not obtained by free copy, or independent analysis, of the TREAT-A-MATIC. The successful reproduction of the system did not take place until Tesco hired Glad for the specific purpose of learning the arrangement and components of the TREAT-A-MATIC. Cf. Head Ski Co. v. Kam Ski Co., D.Md.1958, 158 F. Supp. 919.

We hold, therefore, that the covenant not to compete is valid and enforcible and that the TREAT-A-MATIC is a protectible trade secret. Accordingly, we reverse and remand to the district court for proceedings consistent with this opinion.