with whom The Chief Justice, Justice Blackmun, and Justice O’Connor join, concurring in part and dissenting in part.
I agree with the Court’s analytic approach to this matter, and with its conclusion that subsection (c)(3) of the regulation, 19 CFR § 133.21(c)(3) (1987), is not a permissible construction of § 526(a) of the Tariff Act of 1930, 19 U. S. C. § 1526(a). I therefore join Parts I, II-A, and II-C of the Court’s opinion. In my view, however, subsections (c)(1) and (c)(2) of the regulation are also in conflict with the clear language of § 526(a). I therefore decline to join Parts II-B and III of Justice Kennedy’s opinion and dissent from that part of the judgment upholding subsections (c)(1) and (c)(2).
I
The Court observes that the statutory phrase “owned by” is ambiguous when applied to domestic subsidiaries of foreign corporations (case 2a). With this much I agree. It may be reasonable for some purposes to say that a trademark nominally owned by a domestic subsidiary is “owned by” its foreign parent corporation. This lawsuit would be different if the Customs Service regulation at issue here did no more than resolve this arguable ambiguity, by providing that a domestic subsidiary of a foreign parent could not claim the protection of § 526(a). In fact, however, that has never been asserted to be the theory of the regulation, and is assuredly not its only, or even its principal, effect. The authority to clarify an ambiguity in a statute is not the authority to alter even its unambiguous applications, and § 526(a) unambiguously encompasses most of the situations that the regulation purports to exclude.
Thus, the regulation excludes from § 526(a)’s import prohibition products bearing a domestic trademark that have been manufactured abroad by the trademark owner (case 2c), or by the trademark owner’s subsidiary (case 2b). But the statutory requirement that the trademark be “owned by” a *319United States citizen or corporation is unambiguous with respect to these two cases. A parent corporation may or may not be said to “own” the assets owned by its subsidiary, but no matter how that ambiguity is resolved it is impossible to conclude that a trademark owned by a United States corporation and applied abroad either by the corporation or its foreign subsidiary is “owned by” anyone other than a United States corporation.
Five Members of the Court (hereinafter referred to as “the majority”) assert, however, that the regulation’s treatment of situations 2b and 2c is attributable to the resolution of yet another ambiguity in § 526(a). See ante, at 292-293 (opinion of Kennedy, J.); ante, at 299 (opinion of Brennan, J.). The statute excludes only merchandise “of foreign manufacture,” which the majority says might mean “manufactured by a foreigner” rather than “manufactured in a foreign country.” I think not. Words, like syllables, acquire meaning not in isolation but within their context. While looking up the separate word “foreign” in a dictionary might produce the reading the majority suggests, that approach would also interpret the phrase “I have a foreign object in my eye” as referring, perhaps, to something from Italy. The phrase “of foreign manufacture” is a common usage, well understood to mean “manufactured abroad.” Hence, when statutes and regulations intend to describe the universe of manufactured goods, they do not refer to goods “of foreign or citizen manufacture,” but to goods “of foreign or domestic manufacture.” See, e. g., 19 CFR § 133.21(a) (1987). I know of no instance in' which anyone, anywhere, has used the phrase with the meaning the majority suggests — and the majority provides no example.
In the particular context of the present statute, however, the majority’s suggested interpretation is not merely unusual but inconceivable, since it would have the effect of eliminating § 526(a)’s protection for some trademark holders in case 1 — which contains what the Court describes as the “proto*320typical” gray-market victims. Not uncommonly a foreign trademark owner licenses an American firm to use its trademark in the United States and also licenses one or more other American firms to use the trademark in other countries. In this situation, the firm with the United States license could not keep out gray-market imports manufactured abroad by the other American firms, since, under the majority’s interpretation, the goods would not be “of foreign manufacture.” Thus, to save the regulation, the majority proposes an interpretation that undermines even the core of the statute.1
The majority does not insist that this queer reading is the best interpretation of “of foreign manufacture,” but only that the Customs Service has adopted this construction of the statute as the basis for its regulation. That will come as a surprise to the Customs Service. The Government’s petition for writ of certiorari in this very case states that § 526(a) deals not with goods manufactured by foreigners, but rather with “goods manufactured abroad,” “genuine foreign-made goods,” “[gjenuine goods manufactured abroad,” “goods produced abroad.” Pet. for Cert. in No. 86-625, p. 3. As far as I can discern, that accords with the absolutely uniform Customs Service interpretation. For example, the Customs *321Service cites (Brief for Federal Petitioners 38, n. 46) the 1951 correspondence from the Commissioner of Customs to Senator Douglas describing § 526(a):
“As interpreted by the [Customs] Bureau, section 526 prohibits the importation of genuine articles of foreign origin bearing a genuine trade-mark. . . . For example: if the foreign owner of a trade-mark applied to articles manufactured in a foreign country assigns the United States rights [to a United States citizen] no articles of foreign origin bearing such mark . . . may be imported.” App. 53 (emphasis added).
Perhaps most telling of all is the Commissioner’s description, in this letter, of the common-ownership exception:
“However, if the United States trade-mark owner and the owner of the foreign rights to the same mark are one and the same person, articles produced and sold abroad by the foreign owner may be imported by anyone. ...” Ibid, (emphasis added).
The Commissioner’s reference to articles produced and sold abroad was not original, but paraphrased the language of the earliest regulation articulating the common-ownership exception to § 526(a) (“merchandise manufactured or sold in a foreign country”) which was reiterated in regulations promulgated in 1937, 1943, 1947, and 1969. See Cust. Reg. 1931, art. 518(b), as amended, T. D. 48537, 70 Treas. Dec. 336-337 (1936); Cust. Reg. 1937, art. 537(b); Cust. Reg. 1943, 19 CFR § 11.14(b) (1943); Cust. Reg. 1947, 19 CFR § 11.14(b) (1947); 19 CFR § 11.14(b) (1969). It is a strange sort of deference to agency interpretation which adopts a view of the statute that the agency clearly rejects.
If it were, as Justice Kennedy believes, “the current interpretation of the regulations we are sustaining,” ante, at 293, n. 4, one would expect there to be in place some mechanism that enables the Customs Service to identify goods that are not only manufactured abroad but also (as the majority’s *322interpretation requires) manufactured by foreigners. Acquiring this knowledge cannot be easy, since the importer of merchandise will often not know the manufacturer’s identity, much less its corporate pedigree. International corporate ownership, not a matter of public record, is often a closely guarded secret. Yet although there is in place a regulation requiring the country of origin (i. e., whether “manufactured abroad”) to be plainly indicated on all imports, see 19 CFR §§ 134.0-134.55 (1987), there is none requiring the nationality of the manufacturer to be stated. After today’s decision, of course, the Customs Service, if it would not rather amend its regulations, will presumably have to devise means to enforce what we say it has been enforcing.
Which suggests one of the most important reasons we defer to an agency’s construction of a statute: its expert knowledge of the interpretation’s practical consequences. Since the Government has never advocated the interpretation proposed by the majority (although these cases have been argued twice), and since we did not so much as ask for additional briefing after conceiving of the novel interpretation, we cannot be sure what other difficulties it will create. It might, for example, conflict with mutually accepted understandings of our commercial treaty commitments to foreign countries, such as the provision in our Treaty of Friendship, Commerce and Navigation with the Federal Republic of Germany, October 29, 1954, that “[njationals and companies of either Party shall be accorded national treatment and most-favored-nation treatment by the other Party with respect to all matters relating to importation and exportation.” [1956] 7 U. S. T. 1839, T. I. A. S. No. 3593, art. XIV, ¶4. I doubt, in any case, that our trade partners will look favorably upon a regulation which, as now interpreted, treats goods manufactured by American companies on their soil more favorably than goods manufactured there by their own nationals.
I find it extraordinary for this Court, on the theory of deferring to an agency’s judgment, to burden that agency with *323an interpretation that it not only has never suggested, but that is contrary to ordinary usage, to the purposes of the statute, and to the interpretation the agency appears to have applied consistently for half a century.
h-i ► — I
Section 526(a) also unambiguously embraces a third situation that the regulation excludes — namely, the situation (case 3) in which a domestic trademark owner and registrant authorizes a foreign firm to use its United States trademark abroad. There, the United States trademark is unambiguously “owned by” a United States firm, and registered by a firm “domiciled in the United States,” and the goods sought to be imported are “of foreign manufacture.” According to Justice Brennan, however, thus reading the words to mean what they say “bespeaks stolid anachronism not solid analysis,” because “[a]ny prescient legislator who could have contemplated that a trademark owner might license the use of its trademark would almost certainly have concluded that such a transaction would divest the licensor not only of the benefit of §526’s importation prohibition, but of all trademark protection.” Ante, at 312, 315.
There may be an anachronism here, but if so it is the statute itself — which Congress has chosen not to update — and not the faithful reading of it to cover what it covers. Justice Brennan characterizes his view as the resolution of textual “ambiguity,” but it has nothing to do with that. A 19th-century statute criminalizing the theft of goods is not ambiguous in its application to the theft of microwave ovens simply because the legislators enacting it “were unlikely to have contemplated” those appliances; and a 1922 (or 1930) statute covering a “corporation . . . organized within, the United States” unambiguously includes a United States corporation that has licensed its trademark abroad, whether or not a United States corporation with that characteristic ex*324isted at the time.2 Justice Brennan is asserting that we have the power — indeed, the obligation, lest we commit a “stolid anachronism” — to decline to apply a statute to a situation that its language concededly covers, not on the ground that the enacting Congress actually intended but failed to express such an exception, nor even on the ground that failure to infer such an exception produces an absurd result, but on the ground that, if the enacting Congress had foreseen modem circumstances, it would have adopted such an exception, since otherwise the effect of the law would extend beyond its originally contemplated purpose. I confess never to have *325heard of such a theory of statutory construction. The principle of our democratic system is not that each legislature enacts a purpose, independent of the language in a statute, which the courts must then perpetuate, assuring that it is fully achieved but never overshot by expanding or ignoring the statutory language as changing circumstances require. To the contrary, it seems to me the prerogative of each currently elected Congress to allow those laws which change has rendered nugatory to die an unobserved death if it no longer thinks their purposes worthwhile; and to allow those laws whose effects have been expanded by change to remain alive if it favors the new effects. In the last analysis it is Justice Brennan’s approach that “bespeaks stolid anachronism,” because in theory it requires judges to rewrite the United States Code to accord with the unenacted purposes of Congresses long since called home. (The reality, I fear, may be even worse than the theory. In practice, the rewriting is less likely to accord with the legislative purposes of yesteryear than the judicial predelictions of today.)
But it is not really necessary to conduct the discourse at such a philosophical level in order to reject what Justice Brennan proposes. For even those who would support the power of a court to disregard the plain application of a statute when changed circumstances cause its effects to exceed the original legislative purpose would concede, I must believe, that such power should be exercised only when (1) it is clear that the alleged changed circumstances were unknown to, and unenvisioned by, the enacting legislature, and (2) it is clear that they cause the challenged application of the statute to exceed its original purpose. Here both of those conditions, far from being clearly satisfied, are almost certainly not met.
Justice Brennan asserts that legislators in 1922 or 1930 were unlikely to have contemplated that a trademark owner could assign his trademark unless he simultaneously conveyed the goodwill and business associated with the mark. *326Ante, at 313-314. But the prohibition on assigning a trademark apart from its associated goodwill has not been eliminated. See 15 U. S. C. § 1060. And no more in 1922 than today did it preclude assignment of the trademark and goodwill on a region-by-region basis. By 1920, it was firmly established that unrelated businesses could own and use an identical trademark so long as the uses were confined to different and distinct regions. See United Drug Co. v. Theodore Rectanus Co., 248 U. S. 90, 100-101 (1918); Hanover Star Milling Co. v. Metcalf, 240 U. S. 403, 415 (1916). As a consequence, a trademark holder doing business in two distinct territories was free to assign the business, goodwill, and rights to the trademark in one of the regions. See, e. g., Scandinavia Belting Co. v. Asbestos & Rubber Works of America, Inc., 257 F. 937, 953-956 (CA2 1919); Battle Creek Toasted Corn Flake Co. v. Kellogg Toasted Corn Flake Co., 54 Ont. L. Rep. 537, 546, 550 (1923); see also Apollinaris Co. v. Scherer, 27 F. 18, 19-20 (CC SDNY 1886) (dicta); cf. Saxlehner v. Eisner & Mendelson Co., 179 U. S. 19 (1900) (a trademark owner does not abandon his trademark if he continues to use it domestically while granting another party the exclusive right to sell the product in certain foreign countries).3 Similarly, a firm that used its trademark in one *327business, say manufacturing cola syrup, could transfer rights to use the trademark in another business, such as bottling cola-flavored soda. See Coca-Cola Bottling Co. v. Coca-Cola Co., 269 F. 796, 806-808 (DC Del. 1920). It was also well established that different parties using an identical trademark in different regions, or for different purposes, could enter into a consent agreement authorizing each party to continue the nonconflicting uses. • See Waukesha Hygeia Mineral Springs Co. v. Hygeia Sparkling Distilled Water Co., 63 F. 438, 441 (CA7 1894). Justice Brennan correctly notes that trademark law now recognizes, as it had only begun to recognize in 1930, that a trademark may be licensed for use by different firms in the same or overlapping regions, ante, at 314-315. That change in the law, however, plays almost no part in the application of § 526(a). Since international trademark licensing is interregional, a statute that applies only to imported goods is hardly affected by a change in trademark law concerning intraregional licensing. Finally, there is direct proof that Congress appreciated the possibility of territorial assignment of trademarks. Justice Brennan acknowledges that the 1922 Congress was well aware of, and indeed was motivated by, the case of A. Bourjois & Co. v. Katzel, 275 F. 539 (CA2 1921), which presented a textbook example of an assignment of the right to use a trademark in a distinct market. Although Congress understood that a United States trademark owner could authorize the use of its mark abroad, Congress nonetheless chose not to create an exception to § 526(a) for that situation.
Nor does it seem to me that the second condition for disregarding the words of the statute is met: that the originar legislative purpose is not served by its text. I cannot agree that “the equities in case 3 . . . differ significantly from the *328equities that motivated Congress to protect the prototypical gray-market victim (case 1),” the United States assignee of a foreign trademark. Ante, at 316. The United States as-signee’s innocent vulnerability to gray-market imports is no greater than that of the United States trademark owner who assigns the right to use his trademark abroad — and whom Justice Brennan would deprive of § 526(a)’s protection. I cannot understand why the latter victim “does not have the same sort of investment at stake,” ante, at 317. If anything, his investment may be even greater, consisting of the entire goodwill associated with his trademark in this country. Nor do I understand why he has more “direct control” over the harm, ibid. The means of control available to the United States assignor are precisely those available to the United States assignee: he can either decline to participate in the assignment from the beginning, or contractually preclude the other party to the assignment from parallel importation. The latter is as unlikely to be effective in the one case as in the other since the bulk of the gray market is attributable to third parties that are unaffiliated with either the manufacturer or the trademark holder. That same phenomenon renders inexplicable Justice Brennan’s perception that all affiliated trademark holders are less in need of, or less deserving of, § 526(a) protection against the products of their foreign affiliates. It is not the affiliates who are doing the damage but third parties.
In sum, while congressional attention to the problem addressed by § 526(a) may have been prompted by the gray-marketeering represented by A. Bourjois & Co. v. Katzel, supra, the language of the statute goes well beyond that narrow case to cover the same inequity in other contexts. Even if Congress could not have envisioned those other contexts I would find no reasonable basis to disregard what the statute plainly says; but to make the case complete, it surely must have envisioned them.
*329* * *
I of course agree that to the extent § 526(a) is ambiguous we need only determine whether the Customs Service’s interpretation of the statute is reasonable, see Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 842-843 (1984). But we owe no deference to a construction that is contrary to the interpretation of the agency. I would therefore hold invalid, in addition to subsection (c)(3) of the regulation, subsections (c)(1) and (c)(2).
Justice Kennedy suggests that “the regulation speaks to [this] hypothetical situation,” since it “allows a company justifiably invoking the protection of the statute to bar the importation of goods of foreign or domestic manufacture. 19 CFR § 133.21(a) (1987).” Ante, at 293, n. 4. This suggestion is puzzling. If, as the majority believes (or as it believes the Customs Service believes), the statute does not exclude the goods in this situation, it is hard to understand how the regulation could do so. The reality, in any case, is that subsection (a) of § 133.21 has nothing to do with § 526(a), but rather implements § 42 of the Lanham Trade-Mark Act, 15 U. S. C. § 1124, which prohibits importation of goods of foreign or domestic manufacture bearing not genuine trademarks identical to a United States trademark, but trademarks that “copy or simulate” a recorded trademark. It is subsection (b) of § 133.21 that implements § 526(a), and which, consistent with that statute, only prohibits importation of “[f]oreign-made [but not domestic-made] articles bearing a trademark identical with one owned” by a United States trademark holder.
Justice Brennan responds with the example of an old statute requiring agency inspection of “all ovens” for their propensity to spew flames. This statute is asserted to be ambiguous with respect to modem microwave or electric ovens because its purpose would not be served. Ante, at 315-316. With respect to microwave ovens there may indeed be an ambiguity — not because the purpose would not be served but because the term “oven” connotes “a heated enclosure,” Webster’s Third New International Dictionary 1605 (1981), and may or may not embrace microwave ovens. With respect to electric ovens, there seems to me no ambiguity at all. Perhaps the statute should not be interpreted to cover electric ovens, but if so it would not be because of ambiguity but because (1) electric ovens are incapable of spewing flames, (2) it is therefore absurd to inspect electric ovens for that propensity, and (3) it is a venerable principle that a law will not be interpreted to produce absurd results.
“The common sense of man approves the judgment mentioned by Puffendorf, that the Bolognian law which enacted ‘that whoever drew blood in the streets should be punished with the utmost severity,’ did not extend to the surgeon who opened the vein of a person that fell down in the street in a fit. The same common sense accepts the ruling, cited by Plowden, that the statute of 1st Edward II, which enacts that a prisoner who breaks prison shall be guilty of felony, does not extend to a prisoner who breaks out when the prison is on fire — ‘for he is not to be hanged because he would not stay to be burnt.’” United States v. Kirby, 7 Wall. 482, 487 (1869) (citations omitted).
Nothing in Justice Brennan’s example suggests that we can simply disregard a phrase, such as “corporation . . . organized within . . . the United States,” whose unambiguous application produces a result that is not at all absurd but merely (in Justice Brennan’s estimation) beyond the contemplation of the enacting Congress.
Justice Brennan’s only attempt to provide case-law support for the proposition that regional trademark licensing was impermissible consists of a reference to dicta in Independent Baking Powder Co. v. Boorman, 175 F. 448 (CC NJ 1910), and Eiseman v. Schiffer, 157 F. 473 (CC SDNY 1907), see ante, at 313-314, n. 7. The latter case contains nothing more than statements of the principle that a trademark cannot be assigned separately from the business to which it pertains — which says nothing about whether business and trademark can be conveyed on a regional basis. The former case does contain the seemingly pertinent remark, quoted by Justice Brennan, that “the assignor cannot, after the assignment, continue the same identical business and at the same places as before, under unassigned trade-marks, and at the same time authorize his assignee to conduct the same business elsewhere under an assigned trade-mark.” 175 F., at 454 (emphasis added). On the facts of the case, however, “elsewhere” was elsewhere in the same market in which the licensor continued to do busi*327ness. The basis of the holding and of the dictum was, once again, that a trademark is not assignable separately from the business to which it pertains. There is no reason to believe that the court, even in dictum, was addressing the question of regional licensing.