CUDAHY, Circuit Judge.

The plaintiff, Ronald H. Selle, brought a suit against three brothers, Maurice, Robin and Barry Gibb, known collectively as the popular signing group, the Bee Gees, alleging that the Bee Gees, in their hit tune, "How Deep Is Your Love," had infringed the copyright of his song, "Let It End." The jury returned a verdict in plaintiff's favor on the issue of liability in a bifurcated trial. The district court, Judge George N. Leighton, granted the defendants' motion for judgment notwithstanding the verdict and, in the alternative, for a new trial. Selle v. Gibb, 567 F. Supp. 1173 (N.D.Ill. 1983). We affirm the grant of the motion for judgment notwithstanding the verdict.

I

Selle composed his song, "Let It End," in one day in the fall of 1975 and obtained a copyright for it on November 17, 1975. He played his song with his small band two or three times in the Chicago area and sent a tape and lead sheet of the music to eleven music recording and publishing companies. Eight of the companies returned the materials to Selle; three did not respond. This was the extent of the public dissemination of Selle's song. Selle first became aware of the Bee Gees' song, "How Deep Is Your Love," in May 1978 and thought that he recognized the music as his own, although the lyrics were different. He also saw the movie, "Saturday Night Fever," the sound track of which features the song "How Deep Is Your Love," and again recognized the music. He subsequently sued the three Gibb brothers; Paramount Pictures Corporation, which made and distributed the movie; and Phonodisc, Inc., now known as Polygram Distribution, Inc., which made and distributed the cassette tape of "How Deep Is Your Love."

The Bee Gees are internationally known performers and creators of popular music. They have composed more than 160 songs; their sheet music, records and tapes have been distributed worldwide, some of the albums selling more than 30 million copies. The Bee Gees, however, do not themselves read or write music. In composing a song, their practice was to tape a tune, which members of their staff would later transcribe and reduce to a form suitable for copyrighting, sale and performance by both the Bee Gees and others.

In addition to their own testimony at trial, the Bee Gees presented testimony by their manager, Dick Ashby, and two musicians, Albhy Galuten and Blue Weaver, who were on the Bee Gees' staff at the time "How Deep Is Your Love" was composed. These witnesses described in detail how, in January 1977, the Bee Gees and several members of their staff went to a recording studio in the Chateau d'Herouville about 25 miles northwest of Paris. There the group composed at least six new songs and mixed a live album. Barry Gibb's testimony included a detailed explanation of a work tape which was introduced into evidence and played in court. This tape preserves the actual process of creation during which the brothers, and particularly Barry, created the tune of the accused song while Weaver, a keyboard player, played the tune which was hummed or sung by the brothers. Although the tape does not seem to preserve the very beginning of the process of creation, it does depict the process by which ideas, notes, lyrics and bits of the tune were gradually put together.

Following completion of this work tape, a demo tape was made. The work tape, demo tape and a vocal-piano version taken from the demo tape are all in the key of E flat. Lead sheet music, dated March 6, 1977, is in the key of E. On March 7, 1977, a lead sheet of "How Deep Is Your Love" was filed for issuance of a United States copyright, and in November 1977, a piano-vocal arrangement was filed in the Copyright Office.

The only expert witness to testify at trial was Arrand Parsons, a professor of music at Northwestern University who has had extensive professional experience primarily in classical music. He has been a program annotator for the Chicago Symphony Orchestra and the New Orleans Symphony Orchestra and has authored works about musical theory. Prior to this case, however, he had never made a comparative analysis of two popular songs. Dr. Parsons testified on the basis of several charts comparing the musical notes of each song and a comparative recording prepared under his direction.

According to Dr. Parsons' testimony, the first eight bars of each song (Theme A) have twenty-four of thirty-four notes in plaintiff's composition and twenty-four of forty notes in defendants' composition which are identical in pitch and symmetrical position. Of thirty-five rhythmic impulses in plaintiff's composition and forty in defendants', thirty are identical. In the last four bars of both songs (Theme B), fourteen notes in each are identical in pitch, and eleven of the fourteen rhythmic impulses are identical. Both Theme A and Theme B appear in the same position in each song but with different intervening material.

Dr. Parsons testified that, in his opinion, "the two songs had such striking similarities that they could not have been written independent of one another." Tr. 202. He also testified that he did not know of two songs by different composers "that contain as many striking similarities" as do the two songs at issue here. However, on several occasions, he declined to say that the similarities could only have resulted from copying.

Following presentation of the case, the jury returned a verdict for the plaintiff on the issue of liability, the only question presented to the jury. Judge Leighton, however, granted the defendants' motion for judgment notwithstanding the verdict and, in the alternative, for a new trial. He relied primarily on the plaintiff's inability to demonstrate that the defendants had access to the plaintiff's song, without which a claim of copyright infringement could not prevail regardless how similar the two compositions are. Further, the plaintiff failed to contradict or refute the testimony of the defendants and their witnesses describing the independent creation process of "How Deep Is Your Love." Finally, Judge Leighton concluded that "the inferences on which plaintiff relies is not a logical, permissible deduction from proof of `striking similarity' or substantial similarity; it is `at war with the undisputed facts,' and it is inconsistent with the proof of nonaccess to plaintiff's song by the Bee Gees at the time in question." 567 F. Supp. at 1183 (citations omitted).

II

Both we and the district court must be reluctant to remove an issue from the purview of the jury on either a directed verdict or a judgment notwithstanding the verdict. Nonetheless, we have a duty to determine whether there is sufficient evidence to support the position of the nonmoving party, in this case, the plaintiff. The standards applicable to a motion for judgment notwithstanding the verdict and to a directed verdict are, of course, the same. All the evidence, taken as a whole, must be viewed in the light most favorable to the nonmoving party. This evidence must provide a sufficient basis from which the jury could have reasonably reached a verdict without speculation or drawing unreasonable inferences which conflict with the undisputed facts. Brady v. Southern Railway, 320 U.S. 476, 480, 64 S.Ct. 232, 234, 88 L.Ed. 239 (1943); United States v. An Article of Device, 731 F.2d 1253, 1257 (7th Cir. 1984); Chillicothe Sand Gravel Co. v. Martin Marietta Corp., 615 F.2d 427, 430 (7th Cir. 1980); Hohmann v. Packard Instrument Co., 471 F.2d 815, 819 (7th Cir. 1973).

It is, of course, not relevant that, in this case, the trial court denied defendants' motion for a directed verdict and submitted the issue to the jury. It is generally more efficient to proceed in this fashion, so that, in the event the reviewing court reverses, the entire case will not have to be retried. Mattivi v. South African Marine Corp., 618 F.2d 163, 166 (2d Cir. 1980). Since we affirm the district court's grant of a judgment notwithstanding the verdict, it is not necessary to consider either the grant of the motion, in the alternative, for a new trial or the defendants' cross-appeal on the district court's denial of summary judgment. We note, however, that the cross-appeal with respect to the summary judgment motion is inappropriate and redundant since the issues it raised were incorporated in the motion for judgment notwithstanding the verdict. The cross-appeal may be little more than a device to win an opportunity to file the last brief or to argue the evidence of witnesses not presented at trial who furnished summary judgment affidavits — and is a procedure not to be encouraged.

III

Selle's primary contention on this appeal is that the district court misunderstood the theory of proof of copyright infringement on which he based his claim. Under this theory, copyright infringement can be demonstrated when, even in the absence of any direct evidence of access, the two pieces in question are so strikingly similar that access can be inferred from such similarity alone. Selle argues that the testimony of his expert witness, Dr. Parsons, was sufficient evidence of such striking similarity that it was permissible for the jury, even in the absence of any other evidence concerning access, to infer that the Bee Gees had access to plaintiff's song and indeed copied it.

In establishing a claim of copyright infringement of a musical composition, the plaintiff must prove (1) ownership of the copyright in the complaining work; (2) originality of the work; (3) copying of the work by the defendant, and (4) a substantial degree of similarity between the two works. See Sherman, Musical Copyright Infringement: The Requirement of Substantial Similarity. Copyright Law Symposium, Number 92, American Society of Composers, Authors and Publishers 81-82. Columbia University Press (1977) [hereinafter "Sherman, Musical Copyright Infringement"]. The only element which is at issue in this appeal is proof of copying; the first two elements are essentially conceded, while the fourth (substantial similarity) is, at least in these circumstances, closely related to the third element under plaintiff's theory of the case.

Proof of copying is crucial to any claim of copyright infringement because no matter how similar the two works may be (even to the point of identity), if the defendant did not copy the accused work, there is no infringement. Arnstein v. Edward B. Marks Music Corp., 82 F.2d 275 (2d Cir.), motion to set aside decree denied, 86 F.2d 715 (2d Cir. 1936). However, because direct evidence of copying is rarely available, the plaintiff can rely upon circumstantial evidence to prove this essential element, and the most important component of this sort of circumstantial evidence is proof of access. See generally 3 Nimmer, Copyright § 13.02 at 13-9 (1983) [hereinafter "Nimmer, Copyright"]. The plaintiff may be able to introduce direct evidence of access when, for example, the work was sent directly to the defendant (whether a musician or a publishing company) or a close associate of the defendant. On the other hand, the plaintiff may be able to establish a reasonable possibility of access when, for example, the complaining work has been widely disseminated to the public. See, e.g., Abkco Music, Inc. v. Harrisongs Music, Ltd., 722 F.2d 988, 998 (2d Cir. 1983) (finding of access based on wide dissemination); Sherman, Musical Copyright Infringement, at 82.

If, however, the plaintiff does not have direct evidence of access, then an inference of access may still be established circumstantially by proof of similarity which is so striking that the possibilities of independent creation, coincidence and prior common source are, as a practical matter, precluded. If the plaintiff presents evidence of striking similarity sufficient to raise an inference of access, then copying is presumably proved simultaneously, although the fourth element (substantial similarity) still requires proof that the defendant copied a substantial amount of the complaining work. The theory which Selle attempts to apply to this case is based on proof of copying by circumstantial proof of access established by striking similarity between the two works.

One difficulty with plaintiff's theory is that no matter how great the similarity between the two works, it is not their similarity per se which establishes access; rather, their similarity tends to prove access in light of the nature of the works, the particular musical genre involved and other circumstantial evidence of access. In other words, striking similarity is just one piece of circumstantial evidence tending to show access and must not be considered in isolation; it must be considered together with other types of circumstantial evidence relating to access.

As a threshold matter, therefore, it would appear that there must be at least some other evidence which would establish a reasonable possibility that the complaining work was available to the alleged infringer. As noted, two works may be identical in every detail, but, if the alleged infringer created the accused work independently or both works were copied from a common source in the public domain, then there is no infringement. Therefore, if the plaintiff admits to having kept his or her creation under lock and key, it would seem logically impossible to infer access through striking similarity. Thus, although it has frequently been written that striking similarity alone can establish access, the decided cases suggest that this circumstance would be most unusual. The plaintiff must always present sufficient evidence to support a reasonable possibility of access because the jury cannot draw an inference of access based upon speculation and conjecture alone.

For example, in Twentieth Century-Fox Film Corp. v. Dieckhaus, 153 F.2d 893 (8th Cir.), cert. denied, 329 U.S. 716, 67 S.Ct. 46, 91 L.Ed. 621 (1946), the court reversed a finding of infringement based solely on the similarities between plaintiff's book and defendant's film. The court stated that the plaintiff herself presented no evidence that the defendant had had access to her book, and the only people to whom the plaintiff had given a copy of her book testified that they had not given it to the defendant. While the court also concluded that the similarities between the book and the film were not that significant, the result turned on the fact that "[t]he oral and documentary evidence in the record . . . establishes the fact that the defendant had no access to plaintiff's book unless the law of plagiarism permits the court to draw an inference contrary to such proof from its finding of similarities on comparison of the book with the picture." Id. at 897. Thus, although proof of striking similarity may permit an inference of access, the plaintiff must still meet some minimum threshold of proof which demonstrates that the inference of access is reasonable.

The greatest difficulty perhaps arises when the plaintiff cannot demonstrate any direct link between the complaining work and the defendant but the work has been so widely disseminated that it is not unreasonable to infer that the defendant might have had access to it. In Cholvin v. B. F. Music Co., 253 F.2d 102 (7th Cir. 1958), the plaintiffs' work had been distributed in 2000 professional copies of sheet music and four recordings, of which 200,000 records were sold, and it had been performed on several nationwide broadcasts. The court held that, in light of this circumstantial evidence, it was reasonable to infer, in combination with similarities between the two pieces, that there had been an infringement. In Abkco Music, Inc. v. Harrisongs Music, Ltd., 722 F.2d 988, 997-99 (2d Cir. 1983), the court found that there had been a copyright infringement based on a theory of subconscious copying. The complaining work, "He's So Fine," had been the most popular song in the United States for five weeks and among the thirty top hits in England for seven weeks during the year in which George Harrison composed "My Sweet Lord," the infringing song. This evidence, in addition to Harrison's own admission that the two songs were "strikingly similar," supported the finding of infringement. On the other hand, in Jewel Music Publishing Co. v. Leo Feist, Inc., 62 F. Supp. 596, 598 (S.D.N.Y. 1945), almost 10,000 copies of the complaining song had been distributed or sold and the music had also been broadcast on national performances. The court still concluded that the showing of access was insufficient, in combination with the other evidence, to support a reasonable inference of access.

The possibility of access in the present case is not as remote as that in Dieckhaus because neither side elicited testimony from the individuals (primarily employees of the publishing companies) to whom the plaintiff had distributed copies of his song. Such evidence might have conclusively disproved access. On the other hand, Selle's song certainly did not achieve the extent of public dissemination existing in Cholvin, Jewel Music Publishing Co., or Harrisongs Music, and there was also no evidence that any of the defendants or their associates were in Chicago on the two or three occasions when the plaintiff played his song publicly. It is not necessary for us, given the facts of this case, to determine the number of copies which must be publicly distributed to raise a reasonable inference of access. Nevertheless, in this case, the availability of Selle's song, as shown by the evidence, was virtually de minimis.

In granting the defendants' motion for judgment notwithstanding the verdict, Judge Leighton relied primarily on the plaintiff's failure to adduce any evidence of access and stated that an inference of access may not be based on mere conjecture, speculation or a bare possibility of access. 567 F. Supp. at 1181. Thus, in Testa v. Janssen, 492 F. Supp. 198, 202-03 (W.D.Pa. 1980), the court stated that "[t]o support a finding of access, plaintiffs' evidence must extend beyond mere speculation or conjecture. And, while circumstantial evidence is sufficient to establish access, a defendant's opportunity to view the copyrighted work must exist by a reasonable possibility — not a bare possibility" (citation omitted). See also Ferguson v. National Broadcasting Co., 584 F.2d 111, 113 (5th Cir. 1978); Scott v. Paramount Pictures Corp., 449 F. Supp. 518, 520 (D.D.C. 1978), aff'd mem., 607 F.2d 494 (D.C. Cir. 1979), cert. denied, 449 U.S. 849, 101 S.Ct. 137, 66 L.Ed.2d 60 (1980).

Judge Leighton thus based his decision on what he characterized as the plaintiff's inability to raise more than speculation that the Bee Gees had access to his song. The extensive testimony of the defendants and their witnesses describing the creation process went essentially uncontradicted, and there was no attempt even to impeach their credibility. Judge Leighton further relied on the principle that the testimony of credible witnesses concerning a matter within their knowledge cannot be rejected without some impeachment, contradiction or inconsistency with other evidence on the particular point at issue. Dieckhaus, supra, 153 F.2d at 899-900. See also Chesapeake and Ohio Railroad Co. v. Martin, 283 U.S. 209, 216, 51 S.Ct. 453, 456, 75 L.Ed. 983 (1931). Judge Leighton's conclusions that there was no more than a bare possibility that the defendants could have had access to Selle's song and that this was an insufficient basis from which the jury could have reasonably inferred the existence of access seem correct. The plaintiff has failed to meet even the minimum threshold of proof of the possibility of access and, as Judge Leighton has stated, an inference of access would thus seem to be "at war with the undisputed facts." 567 F. Supp. at 1183.

IV

The grant of the motion for judgment notwithstanding the verdict might, if we were so minded, be affirmed on the basis of the preceding analysis of the plaintiff's inability to establish a reasonable inference of access. This decision is also supported by a more traditional analysis of proof of access based only on the proof of "striking similarity" between the two compositions. The plaintiff relies almost exclusively on the testimony of his expert witness, Dr. Parsons, that the two pieces were, in fact, "strikingly similar." Yet formulating a meaningful definition of "striking similarity" is no simple task, and the term is often used in a conclusory or circular fashion.

Sherman defines "striking similarity" as a term of art signifying "that degree of similarity as will permit an inference of copying even in the absence of proof of access. . . ." Sherman, Musical Copyright Infringement, at 84 n. 15. Nimmer states that, absent proof of access, "the similarities must be so striking as to preclude the possibility that the defendant independently arrived at the same result." Nimmer, Copyright, at 13-14.

"Striking similarity" is not merely a function of the number of identical notes that appear in both compositions. Cf. Wilkie v. Santly Brothers, Inc., 13 F. Supp. 136 (S.D.N.Y. 1935), aff'd, 91 F.2d 978 (2d Cir.), cert. denied, 302 U.S. 735, 58 S.Ct. 120, 82 L.Ed. 568 (1937), aff'd on reargument, 94 F.2d 1023 (2d Cir. 1938) (comparison of note structure demonstrates striking similarity), and Jewel Music Publishing Co. v. Leo Feist, Inc., 62 F. Supp. 596 (S.D.N.Y. 1945) (in light of plaintiff's inability to establish access, degree of similarity despite identity or near identity of several bars was not striking). An important factor in analyzing the degree of similarity of two compositions is the uniqueness of the sections which are asserted to be similar.

If the complaining work contains an unexpected departure from the normal metric structure or if the complaining work includes what appears to be an error and the accused work repeats the unexpected element or the error, then it is more likely that there is some connection between the pieces. See, e.g., Nordstrom v. Radio Corporation of America, 251 F. Supp. 41, 42 (D.Colo. 1965). If the similar sections are particularly intricate, then again it would seem more likely that the compositions are related. Finally, some dissimilarities may be particularly suspicious. See, e.g., Meier Co. v. Albany Novelty Manufacturing Co., 236 F.2d 144, 146 (2d Cir. 1956) (inversion and substitution of certain words in a catalogue in a "crude effort to give the appearance of dissimilarity" are themselves evidence of copying); Blume v. Spear, 30 F. 629, 631 (S.D.N.Y. 1887) (variations in infringing song were placed so as to indicate deliberate copying); Sherman, Musical Copyright Infringement, at 84-88. While some of these concepts are borrowed from literary copyright analysis, they would seem equally applicable to an analysis of music.

The judicially formulated definition of "striking similarity" states that "plaintiffs must demonstrate that `such similarities are of a kind that can only be explained by copying, rather than by coincidence, independent creation, or prior common source.'" Testa v. Janssen, 492 F. Supp. 198, 203 (W.D.Pa. 1980) (quoting Stratchborneo v. Arc Music Corp., 357 F. Supp. 1393, 1403 (S.D.N.Y. 1973)). See also Scott v. WKJG, Inc., 376 F.2d 467, 469 (7th Cir. 1967) (the similarities must be "so striking and of such nature as to preclude the possibility of coincidence, accident or independent creation."); Arnstein v. Porter, 154 F.2d 464, 468 (2d Cir. 1946) (same); Scott v. Paramount Pictures Corp., 449 F. Supp. 518, 520 (D.D.C. 1978) (same). Sherman adds:

To prove that certain similarities are "striking," plaintiff must show that they are the sort of similarities that cannot satisfactorily be accounted for by a theory of coincidence, independent creation, prior common source, or any theory other than that of copying. Striking similarity is an extremely technical issue — one with which, understandably, experts are best equipped to deal.

Sherman, Musical Copyright Infringement, at 96.

Finally, the similarities should appear in a sufficiently unique or complex context as to make it unlikely that both pieces were copied from a prior common source, Sheldon v. Metro-Goldwyn Pictures Corp., 81 F.2d 49, 54 (2d Cir.), cert. denied, 298 U.S. 669, 56 S.Ct. 835, 80 L.Ed. 1392 (1936), or that the defendant was able to compose the accused work as a matter of independent creation, Nichols v. Universal Pictures Corp., 45 F.2d 119, 122 (2d Cir. 1930), cert. denied, 282 U.S. 902, 51 S.Ct. 216, 75 L.Ed. 795 (1931). See also Darrell v. Joe Morris Music Co., 113 F.2d 80 (2d Cir. 1940) ("simple, trite themes . . . are likely to recur spontaneously . . . and [only few] . . . suit the infantile demands of the popular ear"); Arnstein v. Edward B. Marks Music Corp., 82 F.2d 275, 277 (2d Cir. 1936). Cf. Abkco Music, Inc. v. Harrisongs Music, Ltd., 722 F.2d 988, 998 (2d Cir. 1983) (finding of a "highly unique pattern" makes copying more likely). With these principles in mind, we turn now to an analysis of the evidence of "striking similarity" presented by the plaintiff.

As noted, the plaintiff relies almost entirely on the testimony of his expert witness, Dr. Arrand Parsons. The defendants did not introduce any expert testimony, apparently because they did not think Parsons' testimony needed to be refuted. Defendants are perhaps to some degree correct in asserting that Parsons, although eminently qualified in the field of classical music theory, was not equally qualified to analyze popular music tunes. More significantly, however, although Parsons used the magic formula, "striking similarity," he only ruled out the possibility of independent creation; he did not state that the similarities could only be the result of copying. In order for proof of "striking similarity" to establish a reasonable inference of access, especially in a case such as this one in which the direct proof of access is so minimal, the plaintiff must show that the similarity is of a type which will preclude any explanation other than that of copying.

In addition, to bolster the expert's conclusion that independent creation was not possible, there should be some testimony or other evidence of the relative complexity or uniqueness of the two compositions. Dr. Parsons' testimony did not refer to this aspect of the compositions and, in a field such as that of popular music in which all songs are relatively short and tend to build on or repeat a basic theme, such testimony would seem to be particularly necessary. We agree with the Sixth Circuit which explained that "we do not think the affidavit of [the expert witness], stating in conclusory terms that `it is extremely unlikely that one set [of architectural plans] could have been prepared without access to the other set,' can fill the gap which is created by the absence of any direct evidence of access." Scholz Homes, Inc. v. Maddox, 379 F.2d 84, 86 (6th Cir. 1967).

To illustrate this deficiency more concretely, we refer to a cassette tape, Plaintiff's Exhibit 27, and the accompanying chart, Plaintiff's Exhibit 26. These exhibits were prepared by the defendants but introduced into evidence by the plaintiff. The tape has recorded on it segments of both themes from both the Selle and the Gibb songs interspersed with segments of other compositions as diverse as "Footsteps," "From Me To You" (a Lennon-McCartney piece), Beethoven's 5th Symphony, "Funny Talk," "Play Down," and "I'd Like To Leave If I May" (the last two being earlier compositions by Barry Gibb). There are at least superficial similarities among these segments, when played on the same musical instrument, and the plaintiff failed to elicit any testimony from his expert witness about this exhibit which compared the Selle and the Gibb songs to other pieces of contemporary, popular music. These circumstances indicate that the plaintiff failed to sustain his burden of proof on the issue of "striking similarity" in its legal sense — that is, similarity which reasonably precludes the possibility of any explanation other than that of copying.

The plaintiff's expert witness does not seem to have addressed any issues relating to the possibility of prior common source in both widely disseminated popular songs and the defendants' own compositions. At oral argument, plaintiff's attorney stated that the burden of proving common source should be on the defendant; however, the burden of proving "striking similarity," which, by definition, includes taking steps to minimize the possibility of common source, is on the plaintiff. In essence, the plaintiff failed to prove to the requisite degree that the similarities identified by the expert witness — although perhaps "striking" in a non-legal sense — were of a type which would eliminate any explanation of coincidence, independent creation or common source, including, in this case, the possibility of common source in earlier compositions created by the Bee Gees themselves or by others. In sum, the evidence of striking similarity is not sufficiently compelling to make the case when the proof of access must otherwise depend largely upon speculation and conjecture.

Therefore, because the plaintiff failed both to establish a basis from which the jury could reasonably infer that the Bee Gees had access to his song and to meet his burden of proving "striking similarity" between the two compositions, the grant by the district court of the defendants' motion for judgment notwithstanding the verdict is affirmed. Because of our doubts concerning the defendants' cross-appeal on the denial of the summary judgment, we order that, under Fed.R.App.P. 38, each party shall bear its own costs.