LEVI STRAUSS & CO., a Delaware corporation, Plaintiff-Appellant, v. BLUE BELL, INC., a Delaware corporation, Defendant-Appellee, 778 F.2d 1352

MERRILL, Circuit Judge: Levi Strauss & Co. has brought this action against Blue Bell, Inc., seeking damages and injunctive relief for federal trademark infringement, 15 U.S.C. § 1114 (1982), and false designation of origin, 15 U.S.C. § 1125(a) (1982), and, under California state law, for trademark infringement, unfair competition and trademark dilution.

The claimed trademark is for a pocket tab, which Strauss describes as a folded cloth ribbon sewn in the vertical seam of a “garment patch pocket.” Blue Bell is alleged to have violated Strauss’s trademark rights by using a horizontal tab, bearing the word “Wrangler” or “Maverick,” on the pockets of its shirts. I The pocket tab trademark was adopted and put to use by Strauss on September 1, 1936, and on May 10, 1938, registration was granted by the Patent and Trademark Office.

The mark was described as “a small marker or tab of textile material or the like, colored red____” The goods in connection with which it was used were stated to be “overalls of the patch pocket type.” In subsequent years other tabs in other colors, some bearing the “Levi’s” name, have been registered for use on jackets and “garments, particularly trousers.” In 1969 Strauss began using a pocket tab on shirts. An application for registration of a shirt tab bearing the Levi’s name has been filed.

Blue Bell has filed opposition to that application, and proceedings have been suspended pending the outcome of this litigation. A shirt tab was registered by Strauss under California state law on June 12, 1974. Commencing around 1975, Blue Bell began marketing a shirt bearing a “Wrangler” or “Maverick” label imprinted on a tab consisting of a folded cloth ribbon affixed to the seam of a shirt pocket. It is this use of a shirt pocket tab by Blue Bell that is the subject of this action.

In order to prevail on its federal claims, Strauss must establish both (1) that it has a protected interest (or trademark right) in the pocket tab allegedly infringed by Blue Bell’s shirt usages and (2) that Blue Bell’s shirt usages are likely to cause consumer confusion and thus infringe upon that interest. See, e.g., Levi Strauss & Co. v. Blue Bell, Inc., 632 F.2d 817, 820-22 (9th Cir.1980) (Blue Bell I).

Strauss cannot simply rely on the federal registration of certain tabs, most notably of those on pants, to establish a protected interest in a pocket tab on garments generally, because registration constitutes prima facie evidence of a protected interest with respect to the goods specified in the registration only. 15 U.S.C. § 1115. Further, Strauss has not alleged that the tab mark is “inherently distinctive.” See 1 J. McCarthy, Trademarks and Unfair Competition, §§ 11:1-2 at 433-36 (2d ed. 1984) (hereinafter J. McCarthy).

In order to establish the first element of its cause of action, namely a protected interest, Strauss has, therefore, relied upon the doctrine of secondary meaning. Secondary meaning is another way of expressing the distinctiveness of a trademark. Blue Bell I, 632 F.2d at 820. In this circuit, “ ‘[secondary meaning has been defined as association, nothing more.’ ” Id. at 820, quoting Carter-Wallace, Inc. v. Proctor & Gamble Co., 434 F.2d 794, 802 (9th Cir.1970).

The basic element of secondary meaning is, thus, the mental association by a substantial segment of consumers and potential consumers “between the alleged mark and a single source of the product.” 1 J. McCarthy, §§ 15:2 at 659, and 15:11(

B) at 686. See Blue Bell I, 632 F.2d at 820. After a bench trial, the district court found that the Strauss tab as applied to the shirt market did not possess secondary meaning and, moreover, that even if the tab as used on shirts was assumed to possess secondary meaning, Blue Bell’s use of its tab on shirts was not likely to cause customer confusion. The district court consequently ruled in favor of Blue Bell with respect to both Strauss’s federal trademark and state law claims.

We affirm the district court as to Strauss’s federal trademark and state unfair competition claims, but we remand for further consideration of Strauss’s state trademark infringement and dilution claims. II The parties agree that a de novo standard of review is appropriate for the district court’s rulings on matters of law. They also agree that the question of secondary meaning is one of fact which is reviewed under the “clearly erroneous” standard. Blue Bell I, 632 F.2d at 821.

They differ, however, as to what standard is appropriate for our review of the trial court’s decision that there was no likelihood of confusion. Strauss urges a de novo review, while Blue Bell argues for the clearly erroneous standard. We hold that henceforth the clearly erroneous standard should be applied in reviewing a trial court’s determination concerning likelihood of confusion. In United States v.

McConney, 728 F.2d 1195 (9th Cir.) (en banc), cert. denied, — U.S. -, 105 S.Ct. 101, 83 L.Ed.2d 46 (1984), this court stated that where a mixed question of law and fact is predominantly one of fact, we review the trial court’s determination under the clearly erroneous standard. 728 F.2d at 1203-04. The issue of likelihood of confusion is a mixed question which appears to be predominantly factual in nature.

Likelihood of confusion is the type of mixed question which fits within the categories suggested in McConney as suited to clearly erroneous review — cases not implicating constitutional rights and those “in which the applicable legal standard provides for a strictly factual test, such as state of mind.” Id. at 1203. De novo review of this issue would demand a significant diversion of appellate court resources to a task which more properly belongs to the district court judge. See id. at 1201 & n. 7. See also Miss Universe, Inc. v. Flesher, 605 F.2d 1130, 1133 (9th Cir.1979).

Moreover, the limited precedential value of likelihood of confusion decisions, each of which stands upon its own facts, reduces the need for de novo review. McConney, 728 F.2d at 1201. See Matter of McLinn, 739 F.2d 1395, 1401 (9th Cir.1984) (en banc). Whether confusion is likely is a factual determination woven into the law. We routinely treat the issue as one of fact in a variety of procedural contexts. For the sake of accuracy, uniformity, and consistency with the predominant view in other circuits, we will hereafter review findings of likelihood of confusion under the clearly erroneous standard.

Ill In Blue Bell I, an earlier lawsuit brought by Strauss over Blue Bell’s use of a tab on the right rear pocket of its pants, this court determined that Strauss’s pocket tab had acquired secondary meaning in the pants market. See 632 F.2d at 821. Under the doctrine of collateral estoppel, the Strauss pant pocket tab continues to possess secondary meaning. See, e.g., Parklane Hosiery Co. v. Shore, 439 U.S. 322, 326 n. 5, 99 S.Ct. 645, 649 n. 5, 58 L.Ed.2d 552 (1979). Indeed, the district court explicitly recognized that the pant pocket tab possessed strong secondary meaning.

Strauss argues, however, that Blue Bell I established that Strauss has (through secondary meaning) a trademark right in a pocket tab on all garments. We cannot agree. Secondary meaning in the pant tab does not necessarily inhere in the shirt tab since the pant tab was registered as a location specific mark. See 15 U.S.C. § 1057(b); In re Levi Strauss & Co., 165 U.S.P.Q. 348, 350 (TTAB 1970). More important, we do not regard the Blue Bell I litigation as an appropriate basis for collateral estoppel in any area other than the pants market.

In the first place, it is clear that this court in Blue Bell I had no intention of deciding the questions of secondary meaning and likelihood of confusion in any area other than the pants market. Our ultimate conclusion was: Strauss has established that a substantial number of prospective buyers associate its pocket tab trademark with only one seller of pants, and that Wrangler’s use of a projecting label with its ends captured in an edge of rear patch pocket of its pants is likely to cause confusion, to cause mistake, or to deceive. 632 F.2d at 822 (emphasis added).

Footnote 7 of Blue Bell I makes the point clear beyond dispute: “The parties have argued toward general

definitions of Strauss’s trademark rights under various hypothetical circumstances. Such

definitions must await presentation of an actual case or controversy involving established facts.” Id. Under the doctrine of collateral estoppel, judgment in a prior suit between parties precludes relitigation by the parties of issues actually litigated and necessary to the outcome of the first action. Parklane, 439 U.S. at 326 n. 5, 99 S.Ct. at 649 n. 5; Clark v. Watchie, 513 F.2d 994, 998 (9th Cir.), cert. denied, 423 U.S. 841, 96 S.Ct. 72, 46 L.Ed.2d 60 (1975). See IB Moore’s Federal Practice ¶ 0.441[2], at 722 (1984).

Similarity between issues does not suffice; collateral estoppel is applied only when the issues are identical. Parklane, 439 U.S. at 326, 99 S.Ct. at 649; Clark v. Watchie, 513 F.2d at 998. If different facts are in issue in a second case from those that were litigated in the first case, then the parties are not collaterally estopped from litigation in the second case. If the litigated issues are the same — the same facts at issue — estoppel will apply and an offer of different proof in a later case will not provide escape. Here, more than the proof is different; the facts are different.

We deal with different products (jeans v. shirts) and different markets; and the assumptions, perceptions and recognitions with which customers approach a purchase are different. See Litton Industries v. Litronix, Inc., 577 F.2d 709, 711 (C.C.P.A.1978) (admission that the previous proceedings involved an application by the same party to register a mark “for different goods is fatal to the board’s reliance on collateral estoppel.”). We cannot say, therefore, that Blue Bell I established secondary meaning in the Strauss tab as applied to all garments, let alone as applied to shirts.

In the alternative, Strauss argues that even if collateral estoppel is inapplicable, “the evidence offered at trial established as a matter of law that the Levi Strauss & Co. pocket tab as used on shirts had acquired secondary meaning sufficient to entitle Levi Strauss & Co. to exclusive usage” (emphasis supplied). Strauss’s argument is two-fold. First, Strauss claims that the district court erred as a matter of law by not considering the secondary meaning in the pant tab as established by collateral estoppel from Blue Bell I.

It insists that this proof was necessarily relevant to a determination of secondary meaning in the shirt tab. The strength of the tab as a trademark for pants is not irrelevant to the issue of secondary meaning in the shirt market where evidence exists of public awareness and transference of its trademark function to shirts. See Levi Strauss & Co. v. Genesco, 742 F.2d 1401, 1405 (Fed.Cir.1984). Strauss did present such evidence. The district court did not, however, ignore the strong secondary meaning of the pant tab.

The court noted: “The strong secondary meaning of the pants’ pocket TAB, and the potential for customer confusion arising out of a competing manufacturer’s use of a pocket tab device on its pants, has recently been confirmed by the Ninth Circuit Court of Appeals [Blue Bell The court then gave a very good reason why the strength of the mark in the context of the pants market should not control in the context of the shirt market. It stated, “[tjhere is ample and sufficient evidence in the trial to indicate that consumers treat pants and shirts differently.

The court finds that consumer perceptions of shirts and pants are different.” That statement is more than sufficient to establish that in the judgment of the district court secondary meaning in the pants market does not carry over into the shirt market. The second prong of Strauss’s argument that its shirt tab had acquired secondary meaning is an attack on the validity of the affirmative evidence of no secondary meaning relied upon by the district court.

As a panel of this court has recently explained: Factors considered in determining whether a secondary meaning has been achieved include: (1) whether actual purchasers of the product bearing the claimed trademark associate the trademark with the producer, (2) the degree and manner of advertising under the claimed trademark, (3) the length and manner of use of the claimed trademark and, (4) whether use of the claimed trademark has been exclusive. See 1 Gilson, Trademark Protection & Practice, § 209[1], Transgo, Inc. v. AJAC Transmission Parts Corp., 768 F.2d 1001, 1015 (9th Cir.1985). Accord Zatarains, Inc. v.

Oak Grove Smokehouse, Inc., 698 F.2d 786, 795 (5th Cir.1983). An expert survey of purchasers can provide the most persuasive evidence on secondary meaning. 1 J. McCarthy, § 15:13(

D) at 690. See Zatarains, 698 F.2d at 795. The district court’s finding of no secondary meaning in the tab as applied to the shirt market was based primarily upon careful, expert surveys of purchasers, including Strauss’s own survey. The survey which the district court believed supported a finding of no secondary meaning was conducted for Strauss and was probative evidence that a substantial segment of the purchasing public did not associate the tab as used on shirts with a single source, i.e., Strauss.

The court also recognized the relevance of such circumstantial factors “as the length of time the mark has been used, the extent of advertising, and the volume of sales” and “whether Levi Strauss was the exclusive user of such a projecting label on patch pockets of shirts during that period.” Strauss was required to prove that the tab as used on shirts had acquired secondary meaning by 1976, when Blue Bell began using a protruding label on shirts. See HMH Publishing Co. v. Brincat, 504 F.2d 713, 715 (9th Cir.1974).

The court found, though, that Strauss had only begun using the tab on shirts in 1969 and that it introduced no evidence of sales of shirts with the tab until 1973. Strauss’s evidence on the issue of the tab’s first use on shirts is mainly uncorroborated testimony which the district court was entitled to discredit.

The court also found that “there is little evidence of Levi’s advertising the shirt tab” during the 1969-76 period and that “other companies utilized a shirt pocket tab (even though Levi Strauss, through threat of legal action, was able to have them cease the use of a tab).” Evidence of third party usage did exist and is relevant to disprove the existence of secondary meaning. See Carter-Wallace, 434 F.2d at 802-03. Finally, the district court might well have concluded that Strauss’s advertising did not primarily promote the shirt tab as against the shirt as a whole.

Our task is not to determine whether a finding might have been made alternative to that of the district court; it is to determine whether the finding of no secondary meaning was clearly erroneous. Mindful of the evidence in the record supporting the court’s finding, we conclude that it was not. In Inwood Laboratories v. Ives Laboratories, 456 U.S. 844, 102 S.Ct. 2182, 72 L.Ed.2d 606 (1982), the Supreme Court stated: By rejecting the District Court’s findings simply because it would have given more weight to evidence of mislabeling than did the trial court, the Court of Appeals clearly erred.

Determining the weight and credibility of the evidence is the special province of the trier of fact. Because the trial court’s findings concerning the significance of the instances of mislabeling were not clearly erroneous, they should not have been disturbed____ An appellate court cannot substitute its

interpretation of the evidence for that of the trial court simply because the reviewing court “might give the facts another construction, resolve the ambiguities differently, and find a more sinister cast to actions which the District Court apparently deemed innocent.” Id. at 856-58, 102 S.Ct. at 2189-90, quoting United States v. National Ass’n. of Real Estate Boards, 339 U.S. 485, 495, 70 S.Ct. 711, 717, 94 L.Ed. 1007 (1950). We similarly decline to reweigh the evidence in this case. In

summary, Strauss pleaded this case not as an infringement of its tab as applied to the pants market or as applied to the shirt market, but as an infringement of its alleged trademark rights in a pocket tab on garments generally. The district court correctly recognized that secondary meaning inhered in the tab as applied to the pants market, but it did not err in finding no secondary meaning in the tab as applied to the shirt market. Because the tab is a location specific mark, the court could not properly have made findings concerning trademark rights in a pocket tab on garments generally.

IV The finding of no secondary meaning in the tab as applied to the shirt market would normally foreclose any possibility of likelihood of confusion. See Sleeper Lounge Co. v. Bell Manufacturing Co., 253 F.2d 720, 723 (9th Cir.1958). In order to establish customer confusion Strauss must establish that use of the Blue Bell mark on shirts is likely to cause a reasonably knowledgeable and prudent purchaser to believe that Strauss was the manufacturer of Blue Bell shirts. See Toho Co. v. Sears, Roebuck & Co., 645 F.2d 788, 790 (9th Cir.1981).

A buyer cannot, however, be “confused unless he is looking for a label he recognizes and picks up another in his confusion. Ergo, a buyer who does not recognize plaintiffs ‘mark’ and does not distinguish it from any other, cannot be confused.” 1 J. McCarthy, § 15:3 at p. 668. Strauss argues that this case is unusual in that Strauss has established secondary meaning in the tab as applied to the pants market and that likelihood of confusion may, therefore, also be established between the Strauss pant tab and the Blue Bell shirt tab.

The independent finding that Strauss’s shirt tab does not possess secondary meaning dooms this argument to failure. Ultimately, even if the tab does possess secondary meaning in the pants market, Strauss must establish that consumers are likely to believe that Blue Bell shirts emanate from Strauss. If, however, consumers do not distinguish Strauss’s mark on shirts, then it is axiomatic that they will not attribute Blue Bell shirts to Strauss, even if they distinguish Strauss’s mark on pants.

To allow Strauss to establish likelihood of confusion in the shirt market on the basis of the secondary meaning of its tab in the pants market would, in short, be tantamount to allowing a second opportunity to prove that consumers distinguish Strauss’s mark on shirts. We have held that they do not. Strauss’s appeal to the “related goods” doctrine, see AMF, Inc. v. Sleekcraft Boats, 599 F.2d 341, 348 (9th Cir.1979), is to no avail.

The related goods analysis is designed to protect a mark from infringement by a similar mark on noncompetitive goods; a buyer may still be likely to assume a common source or sponsorship. 2 J. McCarthy, § 24:6 at 182. In other words, the related goods test serves largely to define the potential scope of secondary meaning. If the doctrine serves any purpose other than to define the potential reach of secondary meaning, it is only as a manifestation of the ultimate test of likelihood of confusion. See id. at 183.

Here, however, even if pants and shirts are related goods, an independent analysis has established that secondary meaning in the Strauss tab does not extend to the shirt market, and there is consequently no need to undertake a separate “related goods” analysis. V The district court found, in the alternative, that even if Strauss’s pocket tab had acquired secondary meaning, and thus a protected trademark right, as applied to the shirt market, Strauss could not prevail in this action because it had failed to demonstrate infringement of that right through likelihood of confusion.

As an initial matter, we note that the likelihood of confusion issue is only reached in this case upon the assumption that the Strauss tab as applied to the shirt market possesses secondary meaning. We reiterate our conclusion, see supra note 8, that the related goods test incorporates the same factors as are applied in cases involving competitive goods, with the addition of such factors as the proximity of the goods and the likelihood of expansion of product lines.

Proceeding upon the assumption that a directly competing good manufactured by Strauss possesses secondary meaning renders factors such as “likelihood of expansion of product lines” superfluous, and we, therefore, see no need to employ any supplementary related goods analysis. Strauss argues that the district court ignored the multifactor test which has been set forth by this court. This simply is not so.

The court stated: Factors that are relevant in determining whether a likelihood of confusion exists are strength or weakness of the marks; similarity in appearance, sound, and meaning; class of goods in question; marketing channels used; evidence of actual confusion; type of goods and degree of care exercised by purchaser; and intent of defendant [citations omitted]. No allegation has been made that the district court wrongly prohibited Strauss from introducing evidence on any of these factors.

It is true that certain aspects of the multifactor test describe the circumstances to which a trier of fact would refer in making an educated gúess as to what was going on in the minds of consumers in the absence of direct proof, such as survey evidence and testimony, as to how consumers were responding to the marks. Survey evidence and testimony may, however, outweigh whatever circumstantial evidence has been introduced. Cf Anti-Monopoly, Inc. v.

General Mills Fun Group, Inc., 684 F.2d 1316, 1323 (9th Cir.1982), cert. denied, 459 U.S. 1227, 103 S.Ct. 1234, 75 L.Ed.2d 468 (1983) (consumer surveys are ample and legitimate evidence in trademark cases); 2 J. McCarthy, §§ 32:46-55 (discussion of the use of survey evidence in trademark cases). We recognize that survey evidence and retailer testimony are primarily relevant to the existence of actual confusion, which is but one facet of the multifactor test.

It was, however, the prerogative of the district court to evaluate and weigh the evidence introduced by Strauss and Blue Bell, and to find that Blue Bell’s direct evidence outweighed Strauss’s circumstantial evidence. Our job as an appellate court is not to reweigh the evidence. See Inwood Laboratories, 456 U.S. at 856-58, 102 S.Ct. at 2189-90. Retailers called by Blue Bell testified that, based upon their experience, they did not believe that Blue Bell’s shirt tab would confuse customers.

Further, both Strauss and Blue Bell employed independent survey organizations to conduct surveys of purchasers of children’s shirts. The district court discussed those surveys in detail, including the manner in which survey organizations selected the random samples, the questions that the survey interviewers asked, and the results of those questions. The possibility that the consumer views expressed were easily influenced, uncertain or uninformed is, therefore, quite low.

Further, the possibility that the surveys were artificial or unreflective of the views of persons who simply notice Blue Bell shirts during casual street encounters was diminished by the use of shirt tabs by both parties over several years and their substantial presence in the shirt market. The court, in making credibility determinations and weighing the evidence, found Blue Bell’s survey evidence more credible than that of Strauss. The court concluded, however, that both parties’ survey evidence indicated that there was no reasonable likelihood of customer confusion.

We conclude, therefore, that the ultimate finding of the district court of no likelihood of confusion was not clearly erroneous. VI Counts III and IV of Strauss’s complaint alleged infringement of state trademark rights, dilution of its trademarks under state law, and unfair competition under

section 3369 of the California Civil Code. A We have recently suggested that the standard of review for infringement claims based upon state law is more deferential than that used for Lanham Act decisions. Monte Carlo Shirt, Inc. v. Daewoo International (America) Corp., 707 F.2d 1054, 1056-57 (9th Cir.1983). The state law-federal law distinction in review was eliminated, however, in Matter of McLinn, 739 F.2d at 1397, where our court held en banc that in both cases review should be de novo.

While McLinn did not discuss review of issues which mix fact with state law, and McConney apparently dealt with issues which mix fact with federal law, 728 F.2d at 1199-1204, McLinn relied extensively upon the McConney opinion to reach the holding that “appellate review of conclusions of state law should be under the same ... standard as conclusions of federal law.” McLinn, 739 F.2d at 1403. Thus, we conclude that the standard of review is the same for both federal infringement claims and those pendent state claims presented here.

We will review questions of state law de novo and the factual conclusions of likelihood of confusion under the clearly erroneous standard. B In considering Strauss’s state trademark infringement claim, the district court, after finding that Strauss’s shirt tab trademark was validly registered, erroneously failed to give Strauss the benefit of a presumption that it thereby met its burden of proof in establishing the validity of this trademark under state law.

State registration of a trademark constitutes “prima facie evidence of the ownership of such mark ... in the State of California.” Cal.Bus. & Prof.Code § 14241 (West 1984). See, e.g., North Carolina Dairy Foundation, Inc. v. Foremost-McKesson, Inc., 92 Cal.App.3d 98, 106 n. 5, 154 Cal.Rptr. 794, 798 n. 5 (1979). Moreover, the district court erred when it rejected Strauss’s state law trademark infringement claim largely on the basis of survey evidence which was found reflective of consumer attitudes throughout the continental United States.

Certain state law trademark infringement claims may be joined with federal claims for trial, and a plaintiff can prevail on either, neither, or both. Golden Door, Inc. v. Odisho, 646 F.2d 347, 352 (9th Cir.1980); International Order of Job’s Daughters v. Lindeburg & Co., 633 F.2d 912, 915-16 (9th Cir.1980), cert. denied, 452 U.S. 941, 101 S.Ct. 3086, 69 L.Ed.2d 956 (1981). If confusion in California (the location of Strauss’s headquarters) can be shown, remedies under California law may be available.

C In considering Strauss’s dilution claim, the district court erroneously concluded that absent a showing of likelihood of confusion, Strauss could not establish a right to relief. The dilution statute expressly states that the possibility of dilution of a trademark “shall be a ground for injunctive relief notwithstanding ... the absence of confusion as to the source of goods or services.” Cal.Bus. & Prof.Code § 14330 (West 1984). See Monte Carlo Shirt, 707 F.2d at 1058 n. 4; Toho, 645 F.2d at 793; Walt Disney Productions v.

Air Pirates, 581 F.2d 751, 760 & n. 19 (9th Cir.1978), cert. denied, 439 U.S. 1132, 99 S.Ct. 1054, 59 L.Ed.2d 94 (1979). Thus, the district court erred as a matter of law in dismissing the dilution claim upon this basis alone. D The district court did not err in dismissing Strauss’s state unfair competition claim, since both secondary meaning and a likelihood of public confusion are required in order to prevail. See, e.g., Allied Artists Pictures Corp. v. Friedman, 68 Cal.App.3d 127, 134-36, 137 Cal.Rptr. 94, 98-99 (1977); 7 B. Witkin,

Summary of California Law, Equity §§ 74, 75 (8th ed. 1974). E We, therefore, vacate the portion of the judgment dismissing Strauss’s dilution and state trademark infringement claims and remand those portions of the action to the district court. Because we affirm the dismissal of the federal portions of Strauss’s action, however, we leave to the discretion of the district court the question of whether exercise of pendent jurisdiction over this purely state claim is appropriate or whether it should be dismissed without prejudice. United Mine Workers v.

Gibbs, 383 U.S. 715, 727-29, 86 S.Ct. 1130, 1139-40, 16 L.Ed.2d 218 (1966). See Wham-O-Mfg. Co. v. Paradise Manufacturing Co., 327 F.2d 748, 753-54 (9th Cir.1964). We note that diversity jurisdiction does not exist between the parties because both Strauss and Blue Bell are incorporated in Delaware, see 28 U.S.C. § 1332(c), and the sole basis of the district court’s jurisdiction over the dilution claim remains pendent jurisdiction under 28 U.S.C. § 1338(b). JUDGMENT AFFIRMED IN PART, VACATED AND REMANDED IN PART. . 15 U.S.C. § 1114(l)(

a) creates civil liability in anyone who makes use in commerce of “any reproduction, counterfeit, copy, or colorable imitation of a registered mark” in connection with the sale of goods where such use “is likely to cause confusion, or to cause mistake, or to deceive.”

Section 1125(

a) creates civil liability in one who markets goods with a false designation of origin including "words or other symbols tending falsely to describe or represent the same." . 15 U.S.C. § 1114, which prohibits trademark infringement, applies only to marks with federal registration, and, therefore, pertains only to any infringement which can be proved of registered marks.

Section 1125(a), which prohibits false designation of origin, is not, however, limited to registered marks. . Most of the evidence at trial pertained to the market for children's shirts. Even so, the district court recognized that Strauss and Blue Bell compete in the markets for men’s, women's, and children’s shirts, and our holding here covers all shirts manufactured by Blue Bell and Strauss which bear projecting tabs on patch pockets. . The parties focus upon their disagreement concerning whether the trial court’s decision was based upon undisputed facts.

Until today, the degree of deference accorded in this circuit to trial court likelihood of confusion determinations has hinged upon the nature of the evidence below. We have reviewed such determinations de novo when based upon undisputed facts, but have applied a different, two-level test where disputed facts have been involved, applying the clearly erroneous standard to the underlying "foundational facts” and giving de novo review to the “legal conclusion” on likelihood of confusion. AMF, Inc. v. Sleekcraft Boats, 599 F.2d 341, 346-47 (9th Cir.1979). See Alpha Industries, Inc. v.

Alpha Steel Tube & Shapes, Inc., 616 F.2d 440, 443-44 (9th Cir.1980); J.B. Williams Co. v. Le Conte Cosmetics, Inc., 523 F.2d 187, 191 (9th Cir.1975), cert. denied, 424 U.S. 913, 96 S.Ct. 1110, 47 L.Ed.2d 317 (1976). The distinction between these alternative tests has, however, often been omitted or obscured. See Lindy Pen Co. v. Bic Pen Corp., 725 F.2d 1240, 1243-44 (9th Cir.1984) (omitting disputed facts/undisputed facts distinction), cert. denied, — U.S. -, 105 S.Ct. 955, 83 L.Ed.2d 962 (1985); Golden Door, Inc. v. Odisho, 646 F.2d 347, 350 (9th Cir.1980); Blue Bell I, 632 F.2d at 822 n. 6.

At least one commentator concluded that the distinction had been abolished and that the two-level test had become universally applicable. J. Gilson, Trademark Protection and Practice, § 8.15 at 8-194 (1984) (hereinafter Gil-son). In now dispensing with our previous distinctions based upon the disputed or undisputed nature of the facts, we are in accord with analogous principles from non-trademark fields. See Commissioner of Internal Revenue v. Duberstein, 363 U.S. 278, 291, 80 S.Ct. 1190, 1200, 4 L.Ed.2d 1218 (1959) (Rule 52(

a) applies to factual inferences from undisputed facts); United States v. Chesher, 678 F.2d 1353, 1358 n. 3 (9th Cir.1982) (clearly erroneous standard applies to trial judge’s decision based on stipulated facts); Lundgren v. Freeman, 307 F.2d 104, 113-15 (9th Cir.1962) (Rule 52(

a) applies to findings based upon documentary evidence as well as to those involving witness credibility); Note of Advisory Committee on Rules, Fed.R.Civ.P. 52 (deference mandated by Rule 52(

a) is not limited to the trial judge’s resolution of disputed facts). . See 2 J. McCarthy, § 32:37 at 745. For instance, trial courts disfavor deciding trademark cases in

summary judgments because the ultimate issue is so inherently factual. See, e.g., United States Jaycees v. San Francisco Junior Chamber of Commerce, 354 F.Supp. 61, 68-69 (N.D.Cal.1972), aff’d, 513 F.2d 1226 (9th Cir.1975) (per curiam). Additionally, the question of likelihood of confusion is routinely submitted for jury determination as a question of fact. See Stuart v. Collins, 489 F.Supp. 827, 830 (S.D.N.Y.1980); Five Platters, Inc. v. Purdie, 419 F.Supp. 372, 381-82 (D.Md.1976). .

A review of trademark cases in other circuits demonstrates that the clearly erroneous standard predominates for review of a district court’s determination on likelihood of confusion. See generally Gilson, § 8:14 at 8-189; 2 J. McCarthy, § 23:22 at 111 n. 15, 113 n. 20; Comment, Appellate Review of Lanham Act Violations: Is Likelihood of Confusion a Question of Law or Fact?, 38 Sw.L.J. 743, 752 & n. 77, 752-762 (1984). In dissenting from the denial of certiorari for a case in which the Sixth Circuit adopted the two-level test.

Justice White partially documented the disparity in the standards employed by the various circuits. Elby’s Big Boy of Steubenville, Inc. v. Frisch’s Restaurants, 459 U.S. 916, 103 S.Ct. 231, 74 L.Ed.2d 182 (1982) (White, J., dissenting). Justice White noted that the First, Fifth, and Eighth Circuits follow the clearly erroneous standard. Id. (citing Keebler Co. v. Rovira Biscuit Corp., 624 F.2d 366, 377 (1st Cir.1980); Sun Banks of Florida, Inc. v. Sun Federal Savings & Loan Ass’n, 651 F.2d 311, 314-15 (5th Cir.1981); Squirtco v. Seven-Up Co., 628 F.2d 1086, 1091 (8th Cir.1980)).

In addition, the Fourth, Tenth, and Eleventh Circuits appear to follow the clearly erroneous standard. See Pizzeria Uno Corp. v. Temple, 747 F.2d 1522, 1526 (4th Cir.1984); Frito-Lay, Inc. v. Morton Foods, Inc., 316 F.2d 298, 300 (10th Cir.1963); Jellibeans, Inc. v. Skating Clubs of Georgia, Inc., 716 F.2d 833, 839-40 (11th Cir.1983). The Second Circuit and the Seventh Circuit treat likelihood of confusion as a question of fact, utilizing de novo review only for comparisons of the similarity of the marks. Playboy Enterprises v.

Chuckleberry Publishing, Inc., 687 F.2d 563, 566 (2d Cir.1982); Henri’s Food Products Co. v. Kraft, Inc., 717 F.2d 352, 354 (7th Cir.1983); James Burrough Ltd. v. Sign of the Beefeater, Inc., 540 F.2d 266, 273 (7th Cir.1976). . The issue in this case is not whether Strauss abandoned a trademark, but whether trademark rights had attached at all. .

The related goods test is not a separate test for establishing likelihood of confusion but incorporates the same factors which are applied in cases involving competitive goods with the addition of such factors as the "proximity" of the goods and the likelihood of expansion of the product lines. See, e.g., Lindy Pen, 725 F.2d at 1243; Golden Door, 646 F.2d at 349-50; AMF, 599 F.2d at 348-49; New West Corp. v. NYM Co., 595 F.2d 1194, 1201 (9th Cir.1979); K-S-H Plastics, Inc. v. Carolite, Inc., 408 F.2d 54, 57 (9th Cir.), cert. denied, 396 U.S. 825, 90 S.Ct. 69, 24 L.Ed.2d 76 (1969); Paul Sachs Originals Co. v.

Sachs, 325 F.2d 212, 214 (9th Cir.1963). . We also note that, for the reasons discussed supra, the issue of likelihood of confusion in the shirt market is no more subject to any collateral estoppel effect from the pants litigation than was the issue of secondary meaning. . The absence of evidence of actual confusion need not give rise to an inference of no likelihood of confusion. Golden Door, 646 F.2d at 351. Strauss was, however, by no means foreclosed from putting on evidence of actual confusion in proving likelihood of confusion. In fact, Strauss did introduce such evidence.

Moreover, Blue Bell was not barred from putting on evidence of no actual confusion merely because Strauss was not required to prove actual confusion. . Our holding is limited to the facts on this record and we express no opinion as to whether some other use of a pocket tab by Blue Bell or another manufacturer might lead to a likelihood of confusion with products manufactured by Strauss.