The principle proved easier to state than apply. See King Research, Inc. v. Shulton, Inc., 454 F.2d 66, 67 & n. 5 (2d Cir.1972) (collecting and comparing inconsistent cases). Finally, in 1961, Judge Friendly set forth an extensive list of criteria useful in weighing a prior owner's claim that use on a noncompeting product amounts to an infringement:

Where the products are different, the prior owner's chance of success is a function of many variables: the strength of his mark, the degree of similarity between the two marks, the proximity of the products, the likelihood that the prior owner will bridge the gap, actual confusion, and the reciprocal of defendant's good faith in adopting its own mark, the quality of defendant's product, and the sophistication of the buyers.

Polaroid Corp. v. Polarad Elecs. Corp., 287 F.2d 492, 495 (2d Cir.), cert. denied, 368 U.S. 820, 82 S. Ct. 36, 7 L. Ed. 2d 25 (1961). The wisdom of this approach is its recognition that each trademark infringement case presents its own unique set of facts. No single Polaroid factor is preeminent, nor can the presence or absence of one determine, without analysis of the others, the outcome of an infringement suit. Rather, the trial court is required to sift the evidence relevant to each of the criteria, in the particular circumstances before it. The crucial issue is whether there exists a likelihood that an appreciable number of ordinary prudent purchasers will be misled, or simply confused, as to the source of the goods in question, Mushroom Makers, Inc. v. R.G. Barry Corp., 580 F.2d 44, 47 (2d Cir.1978), cert. denied, 439 U.S. 1116, 99 S. Ct. 1022, 59 L. Ed. 2d 75 (1979). With reference to this ultimate question, and from a balancing of the determinations reached on all of the Polaroid factors, a conclusion is reached whether the parties have a right to the relief sought. Kiki Undies Corp. v. Promenade Hosiery Mills, Inc., 411 F.2d 1097, 1099 (2d Cir.1969), cert. denied, 396 U.S. 1094, 90 S. Ct. 707, 24 L. Ed. 2d 698 (1970).

In this appeal from a judgment and order of the United States District Court for the Eastern District of New York, Edward R. Neaher, Judge, Lever Brothers Company contends it was error to deny its request for injunctive relief, for alleged infringement of its mark AUTUMN. Specifically, Lever asserts that Judge Neaher's factual findings were clearly erroneous within the meaning of Fed.R.Civ.P. 52(a)1 and the district court improperly applied the Polaroid factors. For the reasons hereinafter stated, we reject both these claims, and affirm the judgment of the district court, 537 F. Supp. 248.

Lever Brothers is a Maine corporation which has its headquarters in New York. It manufactures and sells a wide variety of consumer products, including detergents, soaps, toothpaste, and food products. It also markets margarine under the trademark AUTUMN, the subject of the litigation before us. AUTUMN margarine, as well as Lever's other products, are sold directly to large retail store chains and through wholesalers to small retail outlets.

AUTUMN represents Lever's third entry into the margarine market. Its IMPERIAL margarine is promoted largely by a description of its taste, the more recent addition PROMISE for its low cholesterol content. Beginning in late 1974, Lever commenced developing a margarine which would combine the taste qualities of the former with the health advantages of the latter. Once the product had been developed, Lever considered various names, eventually settling on AUTUMN. Lever initiated sales of its product in February, 1975, and shortly thereafter introduced it on a test basis in the Los Angeles metropolitan area market. Buoyed by initial sales figures, Lever expanded into other geographic areas in California, and also into Oregon and Washington. AUTUMN has since been sold in eight other states, although the majority of sales have been in the west coast states. The AUTUMN mark was accepted by the Patent and Trademark Office and was registered for use on margarine on October 21, 1975.

In 1977, responding to the success of a competitor's "grain-type" bread in the Charlotte, North Carolina market, American Bakeries began the development of its own bread. American's southeast division, Merita,2 produced a similar, as yet unnamed grain-type bread. A trademark search was conducted, at American's request, by the Patent and Trademark Office, on each potential mark. Eventually, American chose the name AUTUMN GRAIN. It ordered the production of bread wrappers bearing that name before it had received a final report on the mark. AUTUMN GRAIN bread was not offered for sale to the general public, however, before the trademark search had been completed and American received the opinion of counsel that the mark was available for use on bread. The search, completed on April 14, 1977, of course revealed the existence of Lever's registered AUTUMN mark for use on margarine, as well as numerous other uses of the word "autumn," either singly or in combination with other words. The Patent and Trademark Office apparently agreed with American's attorney, that AUTUMN GRAIN was available and registrable for use on bread, since it passed the trademark application on August 23, 1977.3 American continued the sale of AUTUMN GRAIN bread.

American sells its bread primarily in the southeastern portion of the country. As indicated, Lever's AUTUMN margarine sales occur principally in the three west coast states. There has, however, been at least some sales overlap. Joseph McGinley, Executive Vice-President of American, testified at trial that AUTUMN GRAIN was sufficiently successful in the initial test market, North Carolina, that American began selling it in other areas, including parts of California. Moreover, Lever, in its brief, does not appear to dispute the existence of at least some sales area overlap, but merely states that "the two products in issue [were] not sold in any particularly large quantities in the same geographic regions ...." Brief on Behalf of Plaintiff-Appellant at 14-15 (emphasis added). Lever conceded the same point at oral argument.4

Both Lever and American expended large sums advertising their respective products. Their efforts were highly successful. In a period of approximately five years, from the time AUTUMN GRAIN bread was first introduced in April, 1977, American sold more than fifty million loaves, at a wholesale price of about sixty-five cents each, for a total value of approximately $43,000,000. In Lever's case, "next day recall tests" disclosed that viewers recalled the AUTUMN margarine commercials at the rate of 30%, well above the average 20% recall experienced for most food products. Thirteen million dollars worth of AUTUMN margarine has been sold since 1975.

In response to American's use of the mark AUTUMN GRAIN, Lever instituted this suit for injunctive relief, alleging trademark infringement, 15 U.S.C. Sec. 1114, and false designation of origin, 15 U.S.C. Sec. 1125.5 After a bench trial, Judge Neaher concluded that Lever had not borne its burden of demonstrating a likelihood of confusion as to the source of the two products, 537 F. Supp. 248, 256 (E.D.N.Y.1982), and refused to grant an injunction.6 This appeal followed.

Lever Brothers concedes, as it must, that the district court utilized the proper test in assessing the likelihood of source confusion. It contends, however, that Judge Neaher employed some of the Polaroid factors incorrectly, and applied clearly erroneous factual findings to others.

(1) Strength of the Mark

"Strength" refers to the "distinctiveness" of the mark, that is, its "tendency to identify the goods sold under the mark as emanating from a particular ... source." McGregor-Doniger Inc. v. Drizzle Inc., 599 F.2d 1126, 1131 (2d Cir.1979). In an order roughly corresponding to the degree of protection which will be afforded, the terms applied to a mark's strength are "arbitrary," "suggestive," "descriptive," and "generic." In this case, Judge Neaher determined that the term AUTUMN was an arbitrary mark, and deserved the greatest degree of protection against possible infringement, in the event countervailing factors were not present. See 537 F. Supp. at 253. Because we agree with the district court that the record revealed the existence of such a factor, we conclude that it properly analyzed the issue of "strength."

An arbitrary mark is one in which an otherwise common word is used in an unfamiliar way. Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4, 11 & n. 12 (2d Cir.1976).7 It is not enough, however, to decide upon this classification to conclude, as Lever would have us do, that the mark AUTUMN necessarily deserves the widest possible protection. Rather, although denominating a mark "arbitrary" can be useful in focusing the inquiry, the strength of a mark "depends ultimately on its distinctiveness, or its 'origin-indicating' quality, in the eyes of the purchasing public." McGregor-Doniger Inc. v. Drizzle Inc., supra, 599 F.2d at 1131. Just as an invented, even bizarre term can lose a measure of trademark protection if it has "become merely descriptive of the product," see, e.g., DuPont Cellophane Co. v. Waxed Products Co., 85 F.2d 75, 80 (2d Cir.), cert. denied, 299 U.S. 601, 57 S. Ct. 194, 81 L. Ed. 443 (1936) ("cellophane"), so too can the distinctiveness of an arbitrary mark be diluted by third party use, cf. McGregor-Doniger Inc. v. Drizzle Inc., supra, 599 F.2d at 1131-33.

Judge Neaher concluded that one key factor diminished the strength of the argument that "the AUTUMN mark on margarine possesses such distinctiveness as to bar its use on a non-competitive food product," 537 F. Supp. at 253. He noted that AUTUMN had been in common use since 1932. Lever disputes this factual finding, maintaining Judge Neaher based it on American's trademark search report, and this document only purported to reveal the status of various registrations, not use. Lever simply ignores other relevant evidence. Before trial, the parties stipulated to a set of facts not in dispute, including the current use of five trademarks containing the word "autumn," all on noncompeting food products.8 It is true that Judge Neaher apparently did not have before him information suggesting these other uses "were well promoted or that they were recognized by customers," Scarves By Vera, Inc. v. Todo Imports Ltd., 544 F.2d 1167, 1173 (2d Cir.1976), but neither was it true, as in Scarves By Vera, that "[t]he record [did] not contain any evidence to support the claim that plaintiff's trademark was weakened by uses of similar marks by third parties," id. at 1174. Contrary to Lever's assertion, Judge Neaher did not refuse to afford the mark AUTUMN the protection it deserved. Rather, the district court drew the plausible inference that the mark's strength had been at least somewhat diluted by third party use during a period of fifty years. My brother Meskill has noted:

Trademark strength is "an amorphous concept with little shape or substance when divorced from the mark's commercial context[.]" ... We see no advantage to be derived from barring judicial consideration of the realities that give content to the concept of trademark strength.

McGregor-Doniger Inc. v. Drizzle Inc., supra, 599 F.2d at 1133.

(2) Degree of Similarity

Even close similarity between two marks is not dispositive of the issue of likelihood of confusion. Id.. Rather, the similarity must be assessed in terms of its effect upon prospective purchasers, and among the factors which could reasonably be expected to have an impact on, and be remembered by, potential consumers, the overall packaging context is especially important.9

Lever argues that Judge Neaher erred by placing too much emphasis on dissimilarities in the packaging of bread and margarine, since "[o]ne cannot package bread and margarine in the same type of package." Brief on Behalf of Plaintiff-Appellant at 11. The proposition is hardly arguable, but conveniently omits reference to Judge Neaher's observation that likelihood of confusion is diminished "by the prominent identification on the bread packages of American's regional bakery divisions, [in this case] Merita, ... which very likely are familiar to consumers in those regions." 537 F. Supp. at 253. We have examined the packaging used on the two products, see Miss Universe, Inc. v. Patricelli, supra, 408 F.2d at 509, and we agree with Judge Neaher that the emphasis placed by American on its regional divisions substantially negates the likelihood of source confusion. See McGregor-Doniger Inc. v. Drizzle Inc., supra, 599 F.2d at 1133-34.

(3) Proximity of the Products

Lever argues that a strong degree of "proximity" between the two products exists because the public thinks of them together. Also, it urges, they are sold in the same stores and are likely to be purchased quickly and without the assistance of sales personnel. Judge Neaher noted the trade practices cited by Lever, and agreed that, standing alone, they reflected a danger of source confusion due to proximity. He concluded, however, that "the absence of any evidence of confusion on the part of consumers derogates from the significance of proximity." 537 F. Supp. at 254. Although actual confusion need not be shown by Lever, substantial sales of both products over several years, without a single example of actual confusion, becomes significant. Mushroom Makers, Inc. v. R.G. Barry Corp., supra, 580 F.2d at 48.

Lever responds by noting the products were not generally sold in the same geographic areas, and argues that it is therefore unrealistic to look for actual confusion. We have already noted, supra, that Lever does not contend no sales overlap occurred, nor has it presented evidence to rebut the inference, reasonably drawn by the trial judge, that at least some actual confusion could have been expected to result from sales of the two products in the same regions. Essentially, Lever would have us place the emphasis differently, by holding that minor sales overlap cannot be expected to result in actual confusion, and accordingly the proximity issue should have been decided in Lever's favor and without reference to the absence of confusion. Whether we find this alternative reading of the evidence attractive is not the point. We decline to substitute Lever's proposed analysis for the one employed by Judge Neaher.

Moreover, Lever's claim, that it is reversible error to find no likelihood of confusion based on "locational separation" of products within a store, misstates the law. What we have said is that, in a suit involving direct competition for the same market--which is not this case--the absence of side-by-side sales will not necessarily mean no likelihood of confusion can be shown. Paco Rabanne Parfums, S.A. v. Norco Enters., Inc., 680 F.2d 891, 893 (1982). Here, as Lever itself notes, no side-by-side sales could possibly occur, since margarine must be placed in refrigerated compartments, and bread is not.

Judge Neaher properly analyzed the proximity issue. He did not ignore the trade practices offered by Lever to support its argument. Rather, he found it significant that a history of combined sales of approximately sixty-three million food items during three years had not yielded even one documented instance of actual confusion. From this he inferred no likelihood of confusion. We agree. Affiliated Hosp. Prods., Inc. v. Merdel Game Mfg. Co., 513 F.2d 1183, 1188 (2d Cir.1975).

(4) Quality of Defendant's Product

Lever does not address this issue in its brief, apparently agreeing with the district court that "[t]he satisfactory quality of AUTUMN GRAIN bread is rather solidly demonstrated by American's sales of some 50 to 60 million loaves having a sales value of about $43,000,000, since the middle of 1977 when distribution began," 537 F. Supp. at 255. Because the purpose of protecting a user's mark on noncompeting goods is, in part, to guard against tarnishing the reputation of the mark, S.C. Johnson & Son v. Johnson, supra, 175 F.2d at 179-80, proof that American's bread had proved highly successful with consumers weighs heavily against Lever. See King Research, Inc. v. Shulton, Inc., supra, 454 F.2d at 69; see also Chandon Champagne Corp. v. San Marino Wine Corp., 335 F.2d 531, 536 (2d Cir.1964).

(5) Bridging the Gap

This element of the Polaroid test is directed at Lever's interest in preserving the possibility it might one day in the future expand into the market American now occupies, using the AUTUMN mark. Although Lever presented evidence that it had bridged the gap with other products, and that its parent, Unilever, had introduced a non-grain-type bread in Europe, no evidence of substance was adduced that Lever was considering, or had ever considered, a bread line extension using the mark AUTUMN. Indeed, Lever's own witness, John Budd, testified that when capitalizing on the success of a Unilever product, Lever generally uses the mark which thrived in Europe, in this case FRESHMAN.10 Moreover, Lever has never, in its eighty-five year history, entered the baked goods market. It was not error to conclude that the likelihood of its bridging the gap between margarine and bread was remote.

(6) Good Faith

American officials denied under oath they were aware of Lever's mark AUTUMN when the original decision was made to use the mark AUTUMN GRAIN. Judge Neaher credited this testimony, in light of the evidence that the margarine was primarily being sold in west coast stores during the period when American was considering trademarks, while the bread was developed to meet a competitor's success with a grain-type bread in North Carolina. The search report, of course, revealed the existence of Lever's mark, but also disclosed a number of other registrations which used the word "autumn" on a variety of food and drink products. Furthermore, American rejected the use of another mark, HARVEST GRAIN, because of the existence of prior registrations which were, in the opinion of counsel, too close for comfort. The choice not to employ one of the marks under consideration "indicates [American's] process of decision was reasoned and in good faith." 537 F. Supp. at 256. Judge Neaher reasonably concluded that American had acted in good faith. See Kiki Undies Corp. v. Promenade Hosiery Mills, Inc., supra, 411 F.2d at 1101.

(7) Sophistication of the Buyer

This element of the Polaroid test was the only one decided against American. The ordinary purchaser of bread and margarine is a casual buyer, and the bustling, self-service atmosphere of a typical supermarket makes careful examination of products unlikely, and Judge Neaher so concluded. 537 F. Supp. at 255. No single Polaroid factor is determinative of the overall question. All elements of the Polaroid test must be considered. In light of our conclusion that the district court properly resolved the others, we are of the view that the balance of interests tips in favor of American. We affirm the judgment of the district court.

Affirmed.