Applied Innovations, Inc. (defendant), appeals from a final judgment entered in the District Court1 for the District of Minnesota, after a bench trial, in favor of the Regents of the University of Minnesota (the university) and National Computer Systems, Inc. (NCS) (together referred to as plaintiffs), finding that AI had infringed the university’s copyrights in a psychological test, the Minnesota Multiphasic Personality Inventory (MMPI). Regents of University of Minnesota v. Applied Innovations, Inc., No. 3-86-CIV-683 (D.Minn. Oct. 9, 1987) (reported at 685 F.Supp. 698), amended (Jan. 4, 1988) (order).
For reversal defendant argues the district court erred in (1) holding that the university had standing to sue for copyright infringement, (2) holding that the university owned a valid copyright in works funded in part with government grants, (3) holding that the work was copy- ' rightable and that defendant had infringed the university’s copyrights, and (4) calculating the amount of the damages. On cross-appeal, plaintiffs argue the district court /erred in (1) holding that the social introversion and correction (K) scales and the correlation or conversion tables are copyrightable only as compilations, (2) refusing to require defendant to recall and destroy software from its customers, and (3) refusing to award reasonable attorney’s fees.
*628For the reasons discussed below, we affirm the judgment of the district court. The appeal and cross-appeal are denied. Defendant’s motion to strike the first 19 pages of plaintiffs’ reply brief is granted.
PROCEDURAL BACKGROUND
Much of the following statement of facts is taken from the district court memorandum opinion (as amended by order dated January 4, 1988).
NCS is a Minnesota corporation and the university’s exclusive commercial licensee in the United States for marketing the MMPI and associated products and services; its principal place of business is located in Minnesota. Defendant is a New Jersey corporation; its principal place of business is located in Rhode Island.
The MMPI is a psychometric test used by medical and psychological professionals to make objective assessments of major personality characteristics that affect professional and social adjustment, such as truthfulness, hypochondria, introversion, depression, and sexual orientation. Plaintiffs describe the MMPI as a revolutionary development in the field of psychological testing. The MMPI consists of four types of materials: (1) 550 “test statements,” or short declarative sentences, to which subjects respond by answering “true,” “false,” or “cannot say”; (2) scale membership or definitions (there are 10 clinical scales and 4 validity scales); (3) normative statements (t-scores); and (4) correlation or conversion tables. The test statements can be administered at random in card format or in one of three “Form” booklets. The correlation or conversion tables sort the test statements in each format of the MMPI by identification number. The responses can be “hand-scored,” but hand-scoring is time-consuming. Most tests are scored by computer. NCS markets the MMPI test and related products and services, including several forms of computerized scoring services. NCS bills its customers on a per-test or per-use basis.
In early 1984 defendant had developed personal computer software to administer, score and interpret the MMPI test. One version of defendant’s software, MMPI Scoring Program I, included 38 MMPI test statements known as the “Grayson critical items” and scored and interpreted the test. A second version of the software, MMPI Scoring Program II, did not include any test statements and only scored and interpreted the test. Defendant sold its MMPI software and on an unlimited-use per program basis. Plaintiffs promptly informed defendant that its MMPI scoring software infringed plaintiffs’ MMPI copyrights. In 1986 plaintiffs filed a five-count complaint against defendant, alleging copyright infringement, trademark infringement and unfair competition, and deceptive trade practices. Defendant denied any copyright infringement and counterclaimed for a declaratory judgment that plaintiffs’ claimed copyrights were invalid or had not been infringed, misuse of copyright, intentional interference with business relations, and antitrust violations.
In the meantime, in April 1985, NCS had begun marketing its own personal computer software for scoring the MMPI, Micro-test. However, Microtest was more expensive than defendant’s MMPI software.
The district court dismissed defendant’s counterclaims against the university except to the extent that they sought declaratory or prospective injunctive relief. The district court also dismissed the misuse of copyright counterclaim against NCS. The parties agreed to bifurcate for later trial the intentional interference and antitrust counterclaims. After an eight-day bench trial, the district court generally found in favor of plaintiffs on their copyright claims, but in favor of defendant on the trademark infringement, unfair competition, and deceptive practices counts. The parties stipulated that, in developing its MMPI scoring software, defendant had copied directly or indirectly 38 test statements (the Grayson critical items), scale membership and item direction scoring for 13 basic scales, t-scores for those scales, the correction (K) factors or scale, the Minnesota adult norms for five corrected (K) scales, and the correlation or conversion tables (from the Form R booklet to the Group Form booklet).
*629The district court found that defendant’s MMPI scoring software copied everything of commercial significance with regard to the scoring and interpreting of MMPI scores and held that defendant’s software infringed plaintiffs’ MMPI copyrights, except for the social introversion and correction (K) scales and the correlation or conversion tables. The district court permanently enjoined defendant from reproducing or distributing software, documentation or copies of the MMPI test statements and scale definitions, correction (K) factors and normative or t-scores for 12 MMPI scales and from inducing or authorizing others to reproduce any portion of those parts of the MMPI, and required defendant to deliver to plaintiffs all of its copies of the infringing software and documentation. The district court awarded NCS $226,598 for lost profits and the university $162,161 for lost royalty payments, plus costs and disbursements. The district court denied plaintiffs’ request for attorney’s fees. This appeal and cross-appeal followed.
THE MMPI
Two University of Minnesota professors, Starke R. Hathaway and J. Charnley McKinley, developed the MMPI over a 10-year period during the late 1930s and early 1940s. Other university researchers, in particular Paul E. Meehl and Lewis E. Drake, also worked on the MMPI. Hathaway and McKinley began their work with psychometric tests used to evaluate mental patients and then simplified and revised them for use on “normal” individuals. Their basic hypothesis was that individuals who share a particular psychological symptom or personality trait or characteristic were likely to respond to certain groups of test statements in the same way and that each response to a particular test statement was indicative of a particular psychological symptom or personality trait or characteristic.
Their basic research method, greatly simplified, involved administering the test statements to clinical patients and “adult normals,” comparing the responses given by the clinical patients with those of the “adult normals,” converting the “raw scores” into normative scores or “t-scores,” reviewing the t-scores and assigning varying “weights” to particular test statements and responses in light of their clinical experience and expertise, and then testing the accuracy of the resulting scores by using validation tests. For example, the correction (K) scale adjusted the score to account for a subject’s conscious and unconscious efforts to make his or her score “look better.”
Sometime before August 1942, Hathaway and McKinley had distributed test statements to a limited group of individuals for research and comment purposes, including the United States Navy, Dr. Burton P. Grimes of the state hospital in Fort Ben-ning, Georgia, Mr. H.D. Rempel of the federal reformatory at El Reno, Oklahoma, and Col. G.W. Guthrie of the Minnesota State Home for Girls.
The hypochondriasis scale was the first to be developed. Hathaway and McKinley published two articles about the MMPI in a 1940 issue of the Journal of Psychology. Part I described the MMPI in general; Part II described the hypochondriasis scale and included 103 test statements and scoring direction and t-score conversion data. There was a general copyright notice to the Journal Press, the publisher of the Journal of Psychology, but no copyright notice to Hathaway and McKinley or to the university-
In May 1942 Hathaway and McKinley published the Minnesota Multiphasic Personality Schedule {MMP Schedule). The copyright notice was in the name of the university. The MMP Schedule was a comprehensive work and contained the 550 test statements, scale membership, scoring direction, and t-score conversion data for the hypochondriasis, depression, hysteria, psychopathic deviate, sexual interest, question, truthfulness, and validity scales.
In July 1942 Hathaway and McKinley published a third article about the MMPI in the Journal of Psychology. This article contained the scale membership, scoring direction and t-score conversion data for the depression scale.
*630In October 1942 the MMPI Manual was published under a copyright notice to the university; this work contained, among other things, the scale membership, scoring direction and t-score conversion data for the psychasthenia scale. The psychasthe-nia scale materials were also published in November 1942 in an article in the Journal of Applied Psychology.
In 1943 a revised edition of the MMPI was published under a copyright notice to the university. This work contained all the scale materials previously published as well as testing materials and scoring data for the hypomania scale.
In 1944 Hathaway and McKinley published an article in the Journal of Applied Psychology containing the testing materials and data for the hysteria, hypomania and psychopathic deviate scales. In 1946 two additional MMPI articles were published in the Journal of Applied Psychology, one by Drake about the social introversion scale in card format and the other by Hathaway and Meehl about the correction (K) scale in card format.
COPYRIGHT OWNERSHIP AND STANDING TO SUE
At trial the university introduced certificates of registration for copyrights and certificates of renewal registration in various versions of the MMPI works. The university also introduced recorded assignments of copyrights to the university from Hathaway and McKinley, Nettie M. Evans, Louise E. Swedien, Meehl, W. Grant Dahlmstrom, George Schlager, Mildred Drake, the Helen Dwight Reid Educational Foundation (for the Journal Press) and assignments of copyrights from the Psychological Corp. On the basis of this evidence, the district court concluded that the university was the sole proprietor of the MMPI copyrights.
As a preliminary matter we note that the Copyright Act of Mar. 4, 1909 (formerly codified at 17 U.S.C. §§ 1-216 (1972)) (hereinafter 1909 Act), applies in this case because all the MMPI works in question were published before January 1, 1978. The 1909 Act was revised in its entirety and superseded by the Copyright Act of Oct. 19, 1976 (effective date Jan. 1, 1978) (now codified at 17 U.S.C. §§ 101-810 (1988) (hereinafter 1976 Act)).
Defendant first argues the district court erred in holding that the university had standing to sue for infringement of the MMPI works first published as articles in periodicals. According to defendant, the university can only claim a copyright interest as the assignee of either the authors of the articles or the publishers of the periodicals. Defendant argues that, in fact, the university received nothing via these assignments because the authors and the publishers, with the exception of the Journal Press, failed to renew their copyrights and therefore this material is now in the public domain. Defendant argues the district court erroneously extended the constructive trust holding in Goodis v. United Artists Television, Inc., 425 F.2d 397 (2d Cir.1970) (Goodis), from the author of a work first published in a periodical to a mere assignee of the publisher of the periodical. Defendant further argues that Goodis should not be extended in any event to the copyright renewal term.
Plaintiffs admit that the MMPI articles published in the 1940 issue of the Journal of Psychology contained a significant, although comparatively small, portion of the total MMPI work. Plaintiffs argue that Goodis does apply here and that publication of the articles under a copyright notice in the name of the publisher of the periodical, the Journal Press, was sufficient to impose a constructive trust on the publisher in order to secure the authors’ copyrights in the articles. Plaintiffs note that the authors assigned their MMPI copyrights to the university in 1942. Plaintiffs further note that the Journal Press renewed its Journal of Psychology copyright in 1968 and that its assignee, the Helen Dwight Reid Educational Foundation, subsequently assigned its interest to the university in 1987. Thus, plaintiffs argue that the university does have standing to sue for copyright infringement as the assignee of both the authors and the publisher.
The university, as the indirect assignee of a previously registered copyright, had *631the burden of persuasion and production to establish the chain of title. Once the university presented evidence of its chain of title, the burden of production shifted to defendant to establish invalidity. See 3 M. Nimmer, Nimmer on Copyright § 12.11[C], at 12.81-82 (1988) (hereinafter Nimmer). We think the university presented sufficient evidence of its chain of title to the copyright in these articles as the assignee, through the Helen Dwight Reid Educational Foundation, of the Journal Press.
The MMPI articles were first published in the Journal of Psychology bearing a copyright notice in the name of the publisher only. There was no separate notice in the names of the authors. Nor was there any evidence that the authors had registered the unpublished manuscripts or that they had reserved all rights other than the right to reproduce the articles in periodical form. Thus, under the 1909 Act, the Journal Press was the proprietor of the statutory copyright in the entire Journal of Psychology issue, including the individual articles. There was no evidence that the Journal Press had assigned the copyright in any component part of that issue of the Journal of Psychology to the authors. Under these circumstances, the authors had no copyright in the articles, and publication of the articles with a general copyright notice in the name of the publisher prevented the articles from being thrust into the public domain. See generally 3 Nimmer § 10.01[C]. Compare § 201(c) of the 1976 Act, 17 U.S.C. § 201(c) (absent an express transfer from the author, the publisher of a collective work such as a periodical acquires only the privilege of reproducing and distributing the contribution as part of that collective work).
In our view, the university cannot rely upon the authors ’ copyrights in the articles published in the 1940 Journal of Psychology to establish its chain of title because, as discussed above, the authors had no statutory copyright interest in the articles. Instead, the university established its chain of title to the 1940 Journal of Psychology articles as the assignee of the publisher’s copyright. The university presented certificates of registration of copyright and renewal of copyright for most of the other MMPI works. Certain MMPI articles, for example, those published in the Journal of Applied Psychology, are in the public domain. However, those articles are derivative works. Much of the work discussed in those articles originally appeared in works that had been published earlier with a proper copyright notice in the name of the university, and the underlying works remain protected by copyright. See 1 Nim-mer § 3.04, at 3-16.
THE DOCTRINE OF INDIVISIBILITY AND THE GOODIS CASE
Under the analysis set forth above, it is not necessary to apply the Goodis holding. The Goodis court was forced to adopt a constructive trust theory in order to avoid the “doctrine of indivisibility of copyright.” The doctrine of indivisibility arose because the 1909 Act referred to a single “copyright” and to a single “copyright proprietor,” and “it was inferred that the bundle of rights which accrued to a copyright owner were ‘indivisible,’ that is, incapable of assignment in parts.” 3 Nimmer § 10.01[A] at 10-4. Under this view of the nature of the right of copyright, it was “impossible to ‘assign’ anything less than the totality of rights commanded by copyright. A transfer of anything less than such a totality was said to be a ‘license’ rather than an assignment.” Id. at 10-4 to -5 (footnotes omitted). Because “only the copyright proprietor (which would include an assignee but not a licensee) had standing to bring an infringement action,” the doctrine of indivisibility protected alleged infringers from successive law suits. Id. at 10-5 (footnotes omitted).
Unfortunately, as noted by Professor Nimmer, “[w]hatever justification for indivisibility remained in terms of avoidance of multiplicity of actions was far outweighed by the impeding effect it had upon commerce in copyrighted works_ and produced technical pitfalls for both buyers and sellers.” Id. at 10-5 to -6.
Perhaps the most serious consequence of the doctrine of indivisibility occurred by reason of the rule that statutory copy*632right in a theretofore unregistered work was obtained by publication bearing a copyright notice in the name of the copyright owner. The problem thereby raised was felt most acutely and can best be illustrated in the field of magazine rights. If an author of an unpublished (and unregistered) manuscript granted to a magazine publisher the right to reproduce the work in magazine form, but reserved all other rights (e.g., motion picture and book publication rights) then ... the publisher would be merely a licensee, and not the proprietor of the work. Therefore, if, as was and is usually the case, the magazine carried a notice only in the name of the magazine publisher without a separate notice in the name of the contributing author, applying the doctrine of indivisibility the result was that the author’s work was published without a valid notice in the name of the work’s “proprietor” and it was consequently injected into the public domain.
Id. § 10.01[C][2], at 10-12 (footnotes omitted).
The court in Goodis was confronted with just such a fatal technical pitfall. In that case the author of a novel had arranged to publish the novel and had also sold the motion picture rights. Before the novel was published, however, the author made arrangements for the novel to appear in serial form in a magazine. The book publisher postponed distribution of the novel and the novel was first published in the magazine. Each issue of the magazine contained a copyright notice in the name of the magazine, and there was no notice in the author’s name. The defendant, a television network which had produced a successful television series based on the novel, argued that the novel had fallen into the public domain because the magazine was a mere licensee. The court refused to apply the doctrine of indivisibility and held that first publication in a magazine under a general copyright notice in the name of the magazine was sufficient to secure the author’s copyright in his or her contribution, even though the author had reserved all rights other than magazine rights. 425 F.2d at 400-01; see, e.g., Saturday Evening Post Co. v. Rumbleseat Press, Inc., 816 F.2d 1191, 1201 (7th Cir.1987).
In the present case the doctrine of indivisibility did not affect the publisher’s copyright. Because the authors had not granted some rights in their articles to one publisher and some rights in the same works to another, the publisher was not a mere licensee and was therefore not in the same position as the magazine in Goodis. Moreover, “[t]he most frequently cited policy for applying the indivisibility rule is to avoid multiple infringement actions, each brought by the holder of a particular right in a literary work without joining as co-plaintiff the author or proprietor of the copyrighted work.” Goodis, 425 F.2d at 400, citing New Fiction Publishing Co. v. Star Co., 220 F. 994 (S.D.N.Y.1915). Because the university had obtained assignments from all of its potential “rival” copyright claimants, there is no possibility of multiple infringement actions.
In sum, we do not need to apply the Goodis holding, and we hold the university has standing to sue for infringement of copyright, either as the proprietor or as an assignee.
GOVERNMENT FUNDING FOR MMPI RESEARCH
As noted above, the MMPI was developed primarily during the late 1930s and early 1940s. The university paid the salaries of the university professors and graduate students, a total of approximately $117,000. In 1937 Hathaway and McKinley applied for and received government funding for their MMPI research through the Works Progress Administration (WPA). In accepting WPA funding, Hathaway and McKinley agreed, among other things, to publish the results of their research in professional journals, to acknowledge their receipt of government support, to make their basic research data available to the public, and to comply with applicable WPA regulations. From March 1938 to February 1943 they received a total of $7,462.46 from WPA as Subproject Nos. 262 and 379. These WPA funds were used to pay clerical workers for transcribing, tabulating and *633summarizing the data collected from administration of the MMPI.
WPA supervisors monitored the MMPI subprojects and prepared periodic reports. Most of the MMPI works published during the period of WPA funding included ac-knowledgements of WPA support. The district court found that none of the periodic WPA reports referred to copyright. A WPA report dated January 4, 1939, noted that all the original MMPI research materials were to be filed in Millard Hall at the university and to be accessible to the public. Another WPA report dated August 29, 1942, referred to the MMP Schedule by title and contained extensive references. In the spring of 1942 Hathaway and McKinley sent pre-publication copies of their third Journal of Psychology article to the WPA.
There was evidence that there were at least three versions of WPA regulations addressing the copyrightability of WPA-funded research materials: Operating Procedure No. W-ll (Mar. 13, 1937) (hereinafter WPA regulation), No. W-ll (rev. July 1, 1938), and No. G-5 (Jan. 10, 1940). The parties stipulated that, if applicable, the relevant WPA regulation would be the 1938 version. The copyright section of the 1938 version provided (emphasis added):
Data secured or reports published from studies financed in whole or in part by funds allotted to the Works Progress Administration are intended for public use. Copyrighting of any such research, statistical and survey materials by an individual or organization, public or private, shall not be sought “except where such materials are included in copyrighted scientific periodicals, books, or other publications not devoted primarily to the results of Works Progress Administration projects. Whenever research project reports are included, together with other scientific materials, in such copy-righted publications it shall be understood that holders of the copyrights shall not be entitled to restrict public use of the data collected and/or analyzed as activities of survey and research projects operated under the auspices of the Works Progress Administration.”
None of the WPA Operating Procedures was published in the Federal Register. There was no direct evidence that Hathaway and McKinley or other university officials knew about the WPA regulation. The district court held that the issue of government funding, and its potential effect on the copyrightability of the MMPI works, was an affirmative defense and accordingly placed the burden of persuasion and production on defendant. The district court held that the WPA regulation was not legally binding because it had never been published in the Federal Register and university officials had no actual knowledge of it. Alternatively, the district court held that the MMPI works were copyrightable because they had been “included in copyrighted scientific periodicals, books, or other publications not devoted primarily to the results of Works Progress Administration projects.”
Defendant argues the district court erred in holding that the university could own valid copyrights in works funded in part by government funds because the university had agreed to follow WPA rules and regulations as a condition of WPA funding. Defendant argues the WPA regulations, specifically the 1938 revision, barred any copyright in the 1942 MMP Schedule and other MMPI works. Defendant also argues that the finding that the university did not have actual knowledge of the WPA regulation is clearly erroneous. We disagree.
The WPA regulation was promulgated by WPA as a statement of “general applicability and legal effect.” Because it was never published in the Federal Register as required by the Federal Register Act, 44 U.S.C. § 1505 (formerly 44 U.S.C. §§ 5, 7), the WPA regulation was not legally binding in the absence of actual knowledge. See, e.g., Timber Access Industries Co. v. United States, 213 Ct.Cl. 648, 553 F.2d 1250, 1255 (1977) (Forest Service Manual); Andrews v. Knowlton, 509 F.2d 898, 905 (2d Cir.) (Cadet Honor Code), cert. denied, 423 U.S. 873, 96 S.Ct. 142, 46 L.Ed.2d 105 *634(1975); United States v. Aarons, 310 F.2d 341, 345-48 (2d Cir.1962) (Coast Guard order); In re Pacific Far East Line, Inc., 314 F.Supp. 1339, 1348 (N.D.Cal.1970) (navy port regulations), aff'd, 472 F.2d 1382 (9th Cir.1973).
As the party seeking to assert the legal effectiveness of the WPA regulation, defendant had the burden of persuasion and production on the issue of actual knowledge. Resolution of this issue necessarily required the reconstruction of events that occurred more than forty-five years ago. There was no direct evidence of actual knowledge and the circumstantial evidence on this issue was not conclusive. There was evidence that, as required by the WPA regulation, the authors used the form of acknowledgement of government support specified by the WPA regulation and made their basic research data accessible to the public. There was also evidence, however, that the WPA supervisors had reviewed a copy of the 1942 MMP Schedule, which contained a copyright notice to the university, and had not raised any objections about its copyrightability. Nor was there any evidence that the WPA regulation restricting copyright had been incorporated into the WPA funding agreement. Cf. S & H Computer Systems, Inc. v. SAS Institute, Inc., 568 F.Supp. 416, 418-19 (M.D.Tenn.1983) (Department of Agriculture contract funding development of computer software contained provisions restricting copyright and granting “all benefits of all patentable results of all research” to the public). Defendant failed to carry its burden of proof on the issue of actual knowledge.
Given the above analysis, we need not review the district court’s alternate holding that the MMPI works were copyrightable because they were “publications not devoted primarily to the results of Works Progress Administration projects” within the meaning of the WPA regulation.
Nor do we reach defendant’s argument that the governmental exception, § 8 of the 1909 Act authorized WPA to promulgate regulations about the copyrightability of works created under government grants. WPA’s authority to promulgate regulations such as Operating Procedure No. W-ll was not disputed. In any event, we note that the Copyright Office did accept the MMPI works for registration and did not reject them because the works had been funded in part by the government. Cf. Schnapper v. Foley, 215 U.S.App.D.C. 59, 667 F.2d 102, 108-12 (1981) (governmental exception in 1909 and 1976 Acts does not bar registration of works commissioned by the government or assignment of copyrights in such works to the government; Copyright Office consistently accepted for registration federally commissioned works under 1909 Act), cert. denied, 455 U.S. 948, 102 S.Ct. 1448, 71 L.Ed.2d 661 (1982).
In sum, we hold the district court did not err in holding that the university could copyright the MMPI works even though the works were funded in part by the government.
COPYRIGHT PROTECTION
The district court held that the MMPI test statements could be copyrighted. The test statements are for the most part short, simple, declarative sentences, such as “I am a good mixer.” and “No one seems to understand me.” The district court held that the authors had used sufficient creativity and originality in drafting the test statements or in revising the questions and statements used in earlier psychometric tests. The district court further held that the testing data (scale membership, scoring directions, t-scores) could be copyrighted as the expressions of discovered scientific facts or processes. Plaintiffs do not claim copyright interests in their research per se; in fact, plaintiffs acknowledge that other researchers have used their MMPI research to develop other psychometric tests. The district court found that although the authors used statistical rules and algebraic formulas in developing their testing data, they selected particular formulas and then, on the basis of their clinical experience and expertise, adjusted the results they had obtained by applying those formulas.
Defendant argues the district court erred in holding that the MMPI test statements and testing data could be copyrighted. De*635fendant argues the test statements cannot be copyrighted because they are “short phrases” within the meaning of 37 C.F.R. § 202.1(a) and because they are derivative works that do not contain any variation recognizable as that of the authors and therefore lack the requisite “originality.” Defendant also argues the testing data (scale membership, scoring directions, t-scores) cannot be copyrighted because they constitute noncopyrightable facts or processes. In particular, defendant argues that the t-scores represent an attempt to copyright an algebraic formula.
Computer programs may be protected by copyright. E.g., Whelan Associates, Inc. v. Jaslow Dental Laboratory, Inc., 797 F.2d 1222, 1233 (3d Cir.1986) (Whelan) (overall structure, sequence and organization of computer software), cert. denied, 479 U.S. 1031, 107 S.Ct. 877, 93 L.Ed.2d 831 (1987); Apple Computer, Inc. v. Franklin Computer Corp., 714 F.2d 1240, 1246-47 (3d Cir.1983) (computer object and source code), cert. dismissed, 464 U.S. 1033, 104 S.Ct. 690, 79 L.Ed.2d 158 (1984). Defendant’s personal computer-based software, MMPI Scoring Program I and II, duplicated first NCS’s in-house computerized MMPI scoring and interpreting services and then NCS’s own personal computer-based MMPI software, Microtest. Because we have already determined that plaintiffs owned the MMPI copyrights and it is not disputed that defendant’s software copied at least some of the test statements (only MMPI Scoring Program I) and all the testing data needed to score and interpret the MMPI (both MMPI Scoring Program I and II), we are concerned only with whether the test statements and testing data are per se uncopyrightable.
Defendant’s first argument is that the test statements are not copyrightable because they lack originality.
The standard for “originality” is minimal. It is not necessary that the work be novel or unique, but only that the work have its origin with the author— that it be independently created. Little more is involved in this requirement than a “prohibition of actual copying.”
To be the original work of an author, a work must be the product of some “creative intellectual or aesthetic labor.” However, “a very slight degree of such labor[,] ... almost any ingenuity in selection, combination or expression, no matter how crude, humble or obvious, will be sufficient” to make the work copyrightable.
West Publishing Co. v. Mead Data Central, Inc., 799 F.2d 1219, 1223 (8th Cir.1986) (West) (citations omitted) (1976 Act; page numbering due to publisher’s arrangement of opinions in reporters), cert. denied, 479 U.S. 1070, 107 S.Ct. 962, 93 L.Ed.2d 1010 (1987).
The test statements are short, simple, declarative sentences, but they are not merely fragmentary words and phrases within the meaning of 37 C.F.R. § 202.1(a). They are not names or titles or slogans.
We think the test statements satisfy the minimal standard for original works of authorship within the meaning of the copyright laws, at least within the context of the administration of the MMPI. Clearly, the test statements that Hathaway and McKinley, and their university colleagues, independently created meet the originality standard. Rubin v. Boston Magazine Co., 645 F.2d 80, 83 (1st Cir.1981) (particular questions about love and romance held copyrightable as original forms of expression); cf. Educational Testing Service v. Katzman, 793 F.2d 533, 539 (3d Cir.1986) (questions in scholastic aptitude and achievement tests); Association of American Medical Colleges v. Mikaelian, 571 F.Supp. 144, 150 (E.D.Pa.1983) (questions in medical school admission test), aff'd without opinion, 734 F.2d 3 (3d Cir.1984); National Conference of Bar Examiners v. Multistate Legal Studies, Inc., 495 F.Supp34, 36 (N.D.Ill.1980) (questions in bar exam), aff'd in part and rev’d in part, 692 F.2d 478 (7th Cir.1982), cert. denied, 464 U.S. 814, 104 S.Ct. 69, 78 L.Ed.2d 83 (1983).
We also think the test statements that are revisions of the questions in pre-exist-ing psychometric tests represent “distinguishable” variations of the prior works. 1 Nimmer §§ 2.01[B], 3.01, 3.03. The revi*636sions are recognizable as the work of the authors and thus are sufficiently original to warrant copyright protection as derivative works. See Toro Co. v. R & R Products Co., 787 F.2d 1208, 1213 (8th Cir.1986) (citations omitted); see generally 1 Nimmer § 3.04.
Defendant’s second argument is that the test statements and testing data cannot be copyrighted because they are facts or methods or processes for discovering facts. Plaintiffs do not claim copyright protection for the MMPI research in itself, that is, the fact that there is a correlation between certain responses to certain test statements and particular psychological traits or characteristics. What they do claim is protected by copyright are the specific testing data developed by Hathaway and McKinley and their university colleagues to measure and evaluate this correlation. In other words, plaintiffs argue the MMPI testing data are copyrightable as expressions of facts or processes.
Copyright protection does not extend to ideas or facts in published works. E.g., Harper & Row Publishers, Inc. v. Nation Enterprises, 471 U.S. 539, 547, 105 S.Ct. 2218, 2223, 85 L.Ed.2d 588 (1985) (Harper & Row); Worth v. Selchow & Richter Co., 827 F.2d 569, 572 (9th Cir.1987) (Worth) (works on trivia), cert. denied, — U.S. -, 108 S.Ct. 1271, 99 L.Ed.2d 482 (1988); Frybarger v. International Business Machines Corp., 812 F.2d 525, 529 (9th Cir.1987) (Frybarger) (video games). “The discovery of a fact, regardless of the quantum of labor and expense, is simply not the work of an author.” 1 Nimmer § 2.11[E], at 2-169 to -170. “The copyright is limited to those aspects of the work — termed ‘expression’ — that display the stamp of the author’s originality.” Harper & Row, 471 U.S. at 547, 105 S.Ct. at 2224. This is particularly true of factual works. “Because authors who wish to express ideas in factual works are usually confined to a ‘narrow range of expression ..., similarity of expression may have to amount to verbatim reproduction or very close paraphrasing before a factual work will be deemed infringed.’ ” Worth, 827 F.2d at 572, citing Landsberg v. Scrabble Crossword Game Players, Inc., 736 F.2d 485, 488 (9th Cir.), cert. denied, 469 U.S. 1037, 105 S.Ct. 513, 83 L.Ed.2d 403 (1984); see also Wainwright Securities, Inc. v. Wall Street Transcript Corp., 558 F.2d 91 (2d Cir.1977) (abstracts of financial reports), cert. denied, 434 U.S. 1014, 98 S.Ct. 730, 54 L.Ed. 2d 759 (1978); Toro Co. v. R & R Products Co., 787 F.2d at 1211-12 (parts numbers).
This is a close question. We think the MMPI testing data are copyrightable as expressions of facts or processes. Our conclusion is expressly based upon the district court’s findings of fact about the methods the authors used to develop the MMPI testing ' data. The district court found that although the authors began with certain discovered facts, statistical models and mathematical principles, which cannot be copyrighted, they then made certain adjustments on the basis of their expertise and clinical experience. In other words, the MMPI testing data, at least for purposes of analysis under the copyright law, do not represent pure statements of fact or psychological theory; they are instead original expressions of those facts or processes as applied and as such are copyrightable. Rubin v. Boston Magazine Co., 645 F.2d at 83; see generally 1 Nimmer § 2.03[E], 2.11.
On cross-appeal plaintiffs argue the district court erred in holding that the social introversion and correction (K) scales and the correlation or conversion tables are copyrighted only as compilations. We disagree. “If the underlying work is in the public domain, a copyright in the derivative work will not render the underlying work protectible.” 1 Nimmer § 3.04, at 3-16 to -16.1 (footnote omitted). Plaintiffs conceded at trial that the social introversion and correction (K) scales in the card format were in the public domain, apparently because these materials were first published in the Journal of Applied Psychology under a copyright notice in the name of the publisher and the publisher of that periodical failed to renew the copyright. Under these circumstances, the district court cor*637rectly extended copyright protection only to the revised formats of these materials as they appeared in the 1960 MMPI Handbook or in other copyrighted publications.
CALCULATION OF DAMAGES
The district court awarded plaintiffs damages for the loss of profits they would have earned but for defendant’s copyright infringement. On the basis of financial information provided by NCS, the district court determined that NCS’s average revenue for the Micro test program was $1104 per customer. The district court then determined that NCS lost 222 customers to defendant’s MMPI scoring software (total of 584 customers, less 24 duplicate and 4 foreign customers, reduced by 20% for cost differential, further reduced by 50% as allowance for incompatible computer hardware). NCS’s yearly lost revenue was $245,088. Again using financial information provided by NCS, the district court determined that NCS’s profit margin was 20% and that NCS’s yearly lost profits were $49,018. The district court then determined that defendant’s customers would use their software for an average of, 5 years because, according to expert testimony, the average “life” and amortization period for computer hardware was 5 years. Thus, NCS’s total amount of lost profits was $245,090. The district court reduced this amount to present value, using the discount rate specified by state law, and awarded NCS a total of $226,598. The district court also awarded the university a total of $162,161, reduced to present value, for lost royalty payments.
Defendant argues there was insufficient evidence to support the 20% profit margin and the 5 year period of use. We disagree. The evidence in the record supported the district court’s calculation of damages on the basis of a 20% profit margin and 5 years as the period of time defendant’s customers would use their computer hardware and their MMPI scoring software.
Defendant also argues the district court improperly awarded NCS damages for lost profits in 1984 and early 1985 when Microtest was not on the market. We disagree. The district court recognized that Microtest was not available until April 1985 and did not award NCS any lost profits for defendant’s 1984 and early 1985 sales. Because defendant’s’ pre-April 1985 sales adversely affected NCS’s Microtest sales in the future, the district court properly awarded NCS the profits it would have earned after April 1985 and in future years, but for defendant’s 1984 and early 1985 sales.
Defendant also argues the district court should have calculated NCS’s lost profits from the month each sale was made and not on a year-to-year basis. Calculation of damages from the month each sale was made rather than on a year-to-year basis would not have made a difference. For example, the district court awarded NCS lost profits for defendant’s 1985 sales for years 1985-1989. The 5-year period for all of defendant’s 1985 sales ended as of December 31, 1989. Otherwise, the 5-year period for July 1985 sales would have ended in July 1990, August 1985 sales in August 1990 and so forth. Calculation on a year-to-year basis was simpler and did not distort the award.
Nor do we think the district court erred in refusing to reduce the award of royalties to the university to reflect its profit margin; there was no evidence in the record that the university incurred any costs in collecting their royalties from NCS.
RECALL
On cross-appeal plaintiffs argue the district court abused its discretion in refusing to require defendant to recall and destroy its MMPI scoring software from its customers. In the present case the district court structured the damages award to cover a 5-year period. As noted above, this 5-year period reflects the average period of time defendant’s customers would use their software. The district court did not abuse its discretion in awarding damages over a period of several years in lieu of recall. Moreover, it is doubtful that recall would be an available remedy against defendant’s customers who are innocent third-parties in this copyright in*638fringement litigation. See generally 3 Nimmer § 14.08, at 14-63 & n. 6.
ATTORNEY’S FEES
On cross-appeal plaintiffs argue the district court abused its discretion in refusing to award them reasonable attorney’s fees. The decision whether to award attorney’s fees in copyright cases is committed to the sound discretion of the district court. See generally id. § 14.10[D]. In some circuits reasonable attorney’s fees are routinely awarded to the prevailing party. See, e.g., Micromanipulator Co. v. Bough, 779 F.2d 255, 259 (5th Cir.1985). Some circuits distinguish between prevailing plaintiffs and prevailing defendants and award attorney’s fees to prevailing defendants only if the plaintiff’s claims are not colorable. See, e.g., Original Appalachian Artworks, Inc. v. McCall Pattern Co., 825 F.2d 355, 356 (11th Cir.1987); Diamond v. Am-Law Publishing Corp., 745 F.2d 142, 148 (2d Cir.1984). Other circuits require a showing of bad faith or frivolity. See, e.g., Cooling Systems & Flexibles, Inc. v. Stuart Radiator, Inc., 777 F.2d 485, 493 (9th Cir.1985). It does not appear that this circuit has adopted a definitive standard for awarding attorney’s fees in copyright cases. See United Telephone Co. v. Johnson Publishing Co., 855 F.2d 604, 612 (8th Cir.1988) (reference to abuse of discretion standard in general); Hartman v. Hallmark Cards, Inc., 833 F.2d 117, 122-23 (8th Cir.1987) (affirming denial of attorney’s fees to prevailing defendant where plaintiff’s claim was not baseless).
We are reluctant to adopt a particular standard. For purposes of this appeal, it is sufficient to decide that attorney’s fees should not be awarded to a prevailing plaintiff as a matter of course. See, e.g., Lieb v. Topstone Industries, Inc., 788 F.2d 151, 156 (3d Cir.1986). In the present case the district court decided not to award attorney’s fees to plaintiffs because the litigation involved numerous complex or novel questions which defendant had litigated vigorously and in good faith. The district court’s assessment of this case is amply supported by the record and the issues raised on appeal and cross-appeal. We cannot say that the district court abused its discretion in refusing to award reasonable attorney’s fees to plaintiffs.
Accordingly, the judgment of the district court is affirmed. The appeal and cross-appeal are denied. Defendant’s motion to strike portions of plaintiffs’ reply brief is granted. Each party shall bear its own costs on appeal.
. The Honorable Donald D. Alsop, Chief Judge, United States District Court for the District of Minnesota.