dissenting.
Because I believe that the Board’s conclusion of likelihood of confusion was erroneous on the evidentiary record before it, I respectfully dissent. I would remand the case for new findings, based on record evidence instead of rank speculation, or reverse because there is no likelihood of confusion under the marketing circumstances established by the prior registration, the application-in-suit, and the other documentary evidence of record. Evidence subsequently submitted only to us may not properly be taken into account.
I.
The Board’s finding that “[cjustomers for registrant’s services would include, among others, service stations, automotive repair and body shops, as well as retailers of automotive parts,” In re Shell Oil Co., Serial No. 73/753,045, slip op. at 8 (TTAB Dec. 16,1991) (emphasis added), was unsupported by any evidence of record. Although the Board may draw reasonable inferences from the evidence, it may not simply assume whom a registrant’s customer base “would include.” Because there was no record evidence identifying registrant’s customers, the Board can only have based its finding entirely on its own speculative assumptions. Even if these assumptions coincidentally are correct, a finding based only on assumptions is, by definition, clearly erroneous.
Inasmuch as the Board’s conclusion was based substantially on this clearly erroneous finding, it, too, must be erroneous. The parties’ services are different and non-competing, with different channels of trade. Therefore, if, instead, the finding about customers had been that registrant sold and advertised only to jobbers (who, in turn, sold to retailers), the customers of registrant who might *1210go to applicant’s service stations would all be sophisticated professionals.1 Such a finding would have changed the entire analysis — by the Board and by us. In addition, such a finding would have blocked the majority’s attempt to distinguish Electronic Design & Sales, Inc. v. Electronic Data Systems Corp., 954 F.2d 713, 21 USPQ2d 1388 (Fed.Cir. 1992) (EDS), and Astra Pharmaceutical Products, Inc. v. Beckman Instruments, Inc., 718 F.2d 1201, 220 USPQ 786 (1st Cir.1983), from the present case on the basis of disparity in sophistication.
Because assumptions can skew the legal analysis, the law is well settled that the Board’s legal conclusion of likelihood of confusion must be based on more than just theoretical possibilities and speculation. EDS, 954 F.2d at 717, 719, 21 USPQ2d at 1391, 1393 (quoting Witco Chem. Co. v. Whitfield Chem. Co., 418 F.2d 1403, 1405, 164 USPQ 43, 44-45 (CCPA 1969) and citing Carter-Wallace, Inc. v. Procter & Gamble Co., 434 F.2d 794, 804, 167 USPQ 713, 720 (9th Cir.1970), respectively); In re Bed & Breakfast Registry, 791 F.2d 157, 159, 229 USPQ 818, 819 (Fed.Cir.1986) (Newman, J.) (“This theory, the nub of the Board’s opinion, is not supported by evidence.... We are left with speculative assumption, an inadequate basis for the legal conclusion reached by the Board.”). What we condemned in Bed & Breakfast, however, we bless here.
II.
Even assuming, arguendo, that the Board’s speculative finding as to registrant’s customer base was correct, in my view even unsophisticated motorists would not be confused. While Shell’s application for its RIGHT-A-WAY AND DESIGN mark originally defined the services used in connection with that mark merely as “Lubrication Oil Services,” see Appendix for Appellant at 77, the amended application limits the mark’s use to “Service Station Oil and Lubrication Change Services.” Id. at 83 (emphasis added). It is common knowledge, indisputable, and verifiable — and hence subject to our judicial notice — that all Shell service stations have signs bearing the Pecten design. It follows, therefore, that Shell’s oil change and lubrication services always have been and, per the application, always will be offered at its stations and hence in view of Shell’s famous Pecten design and the name “Shell.” Thus, it is not likely that any motorist would be confused as to the source or sponsorship of these services. Even those very few motorists who were aware of registrant’s services and service mark would intuitively know that oil change and lubrication services advertised by the mark at Shell stations are sponsored by Shell and Shell only. Therefore, the Board’s conclusion that purchasers of registrant’s services would likely be confused upon coming to Shell stations is fanciful and contrary to common sense. It ignores the realities, well known to the motoring public, including registrant’s customers, that no oil company offers oil change or lube (or other such) services sponsored by anyone else at its stations. Marketplace realities, however, are supposed to control likelihood of confusion analysis. See 2 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 23.17 (3d ed. 1992).
III.
Likewise, there is no likelihood of reverse confusion because no one, registrant’s customers least of all, would think that Shell would distribute automotive parts to retail automotive parts stores, auto repair shops, body shops, or, indeed, non-Shell service stations. For Shell to do so would be to go into business against itself, because all of those entities compete with Shell’s auto repair bays. Nor, for the very same reason, would Shell own a subsidiary, or be owned by a parent company, that did so. Thus, registrant’s customers, even if unsophisticated or employing lesser care, would not likely believe that Shell and registrant were either related or owned in common by another company. Gas station patrons, including registrant’s customers, understand that major oil *1211companies operating service stations do not compete against themselves. Therefore, the majority’s conclusion is erroneous under a reverse confusion analysis as well.
CONCLUSION
In my view, at the very least this case should be remanded to the Board for a proper finding of the facts by reopening the record to consider documents such as those offered on motion to our court but not previously to the Board. Better yet, our court should simply reverse, in view of how the realities of the relevant marketplace make confusion of the marks for these dissimilar services decidedly unlikely. In my view, the potential number of relevant customers — customers of both companies — is minuscule. According to Shell’s brief, they consist entirely of sophisticated buyers for jobbers. The services, moreover, are neither closely related nor competing and use different channels of trade. Yet the Board minimized these factors, established by the documentary evidence, and ignored realities in the relevant marketplace while indulging in far-ranging and implausible speculation about registrant’s customers and their perceptions. This we should not condone. Only by a reversal or remand can we honor the core command of In re E.I. DuPont DeNemours & Co., 476 F.2d 1357, 1362, 177 USPQ 563, 567-68 (CCPA 1973)—that each case of alleged likelihood of confusion must be decided on the particular facts of the case.
. Some of the new documents offered our court on motion so indicated, although there was also a contrary indication, which is perhaps explainable (as Shell suggested) as post-decision marketing — quite different from marketing at the time of this application. Unfortunately, neither category of this evidence was before the Board.