concurring in part and dissenting in part:
I agree with the court’s able opinion except I am persuaded that “doing business” is a jurisdictional ground under clause one. This reading is consistent with the statuto*205ry language once we acknowledge the implausibility of a literal interpretation. It also has support in the legislative history, and carries out the primary purpose behind the statute. While wary of potential jurisdictional excesses and possibilities for affronts to foreign sovereigns presented by the “doing business” jurisdiction, I am persuaded that the remedy for such ills is faithful adherence to Congress’ mandate that the business carried on in the United States be “substantial,” and use of the forum non conveniens doctrine. At the same time our exercise of this jurisdictional ground ought to be limited to cases that would otherwise be cognizable in federal court absent any issue of sovereign immunity. Such a restriction fulfills the congressional purpose of avoiding use of our court system by strangers with foreign disputes.
A.
The FSIA presents a peculiarly twisted exercise in statutory draftsmanship. See Gibbons v. Udaras na Gaeltachta, 549 F.Supp. 1094, 1105, 1106 (S.D.N.Y.1982) (characterizing the FSIA as “remarkably obtuse,” “a ... statutory labyrinth that, owing to the numerous interpretive questions engendered by its bizarre structure and its many deliberately vague provisions, has during _ its lifetime been a financial boon for the private bar but a constant bane of the federal judiciary”); Texas Trading v. Federal Republic of Nigeria, 647 F.2d 300, 302 (2nd Cir.1981) (“vaguely-worded statute”).
The interpretive exercise of this panel was anticipated and virtually required. Congress chose to make the exceptions in sections 1605-07 purposefully ambiguous, having “decided to put [their] faith in the U.S. Courts,” and thus attempted to provide only “very modest guidance” to the judiciary. Jurisdiction of U.S. Courts in Suits Against Foreign States: Hearings Before the Subcomm. on Administrative Law and Governmental Relations of the House Comm, on the Judiciary on H.R. 11315, 94th Cong.2d Sess. 53 (1976) (testimony of Monroe Leigh, Legal Adviser, Department of State).
The Act’s structure complicates matters. Rather than hinging personal jurisdiction on a defendant’s contacts with the United States, the FSIA inquires whether there is subject matter jurisdiction. Then rather than mark subject matter jurisdiction by traditional indicia such as the citizenship of the parties or the substantive law under which the claim arises, the FSIA turns the inquiry to whether sovereign immunity exists. Thus, the scant language of the first clause of section 1605(a)(2) must: (1) establish the necessity that the claim somehow relate to a commercial activity, (2) establish subject matter jurisdiction with its traditional focus on the claim instead of the parties, and (3) specify the defendant’s contacts necessary for personal jurisdiction, but (4) be sufficiently malleable to accomodate the shaping wisdom of the case-by-case process. It is not surprising that this “Gordian Knot” fails to give concrete guidance to courts. See Texas Trading, 647 F.2d at 307. The language of section 1605(a)(2) therefore provides a starting-point for my interpretation, but the primary congressional purpose of subjecting foreign governments acting commercially to jurisdiction like any other commercial entity ultimately serves as my compass. See Gibbons v. Udaras na Gaeltachta, 549 F.Supp. at 1106. See also House Report at 7, reprinted in 1976 U.S.C.C.A.N. at 6605 (Act codifies restrictive theory of sovereign immunity); 28 U.S.C. § 1602 (in statute’s “findings and declaration of purpose” it is declared that “states are not immune from the jurisdiction of foreign courts insofar as their commercial activities are concerned.”).
B.
No circuit court has ever directly faced the issue of whether the first clause of section 1605(a)(2) includes “doing business” as a jurisdictional ground. Although several circuit courts have found jurisdiction based on the existence of a nexus, the majority opinion is the first to deny jurisdiction for lack of a nexus where the defendant “does business” in the United States, the Third Circuit’s “rejection” of a “doing *206business” test being dicta in my view. See Velidor v. L/P/G Benghazi, 653 F.2d 812, 820 (3d Cir.1981). Further, as the majority opinion concedes, the Second Circuit arguably recognized “doing business” as a jurisdictional ground. While the majority is correct in regarding the language in Ministry of Supply, Cairo v. Universe Tankships, 708 F.2d 80 (2d Cir.1983), as dicta, because the “doing business” issue was not directly before it, I would be equally cautious with the interpretation of the Third Circuit’s opinions.
Our panel then writes on what is essentially a clean slate. And while I share the majority’s concern for uniformity among the circuits, I understand this circuit to be the first to address the “doing business” issue.
C.
When the relevant portions of section 1603(d) and (e) are inserted into clause one of section 1605(a)(2), we are left with the task of construing:
A foreign state shall not be immune from the jurisdiction of the courts of the United States or of the States in any case in which the action is based upon a regular course of commercial conduct carried on by such state and having substantial contact with the United States.
See Matter of Rio Grande Transport, Inc., etc., 516 F.Supp. 1155, 1162 (S.D.N.Y.1981). Because this language cannot be interpreted literally, courts have been driven to substitutions for the words “based upon.” The majority’s substitution is “have a nexus with.” While plausible, the search for a nexus between the United States and a claim in the usual case is a search for a specific act or acts in the United States, rather than for a regular course of commercial conduct in the sense of a carrying on of a commercial business in the United States. See House Report at 16, reprinted in 1976 U.S.C.C.A.N. at 6614-15 (“A ‘regular course of commercial conduct’ includes the carrying on of a commercial enterprise such as a mineral extraction company, an airline or a state trading corporation.”) I would expect a search for a nexus to be a search for an act or series of acts that will be part of the conduct of the business without constituting the entire conduct of the business. It follows that allowing only claims with a nexus to the “commercial activity” carried on in the United States would not exhaust the full range of clause one when “commercial activity” is defined as a “regular course of commercial conduct.”
The “substantial contact” language of § 1603(e) also supports my reading. This language appears to draw upon the “minimum contacts” test of International Shoe v. Washington, 326 U.S. 310, 66 S.Ct. 154, 90 L.Ed. 95 (1945). That test marks off a “doing business” ground as involving “instances in which the continuous corporate operations within a state [are] thought so substantial and of such a nature as to justify suit against it on causes of action arising from dealings entirely distinct from these activities.” Id. at 318, 66 S.Ct. at 159. The FSIA’s “substantial contact” language thus does not signal a congressional intent to enact more stringent a test than the “minimum contacts” necessary for “doing business” jurisdiction. But see Maritime Intern. Nominees v. Republic of Guinea, 693 F.2d 1094, 1109 and n. 23 (D.C.Cir.1982), cert. denied, — U.S.—, 104 S.Ct. 71, 78 L.Ed.2d 84 (1983). To the contrary, the phrase could be lifted straight from International Shoe.
Furthermore, reading clause one to embody a “doing business” jurisdictional ground gives a coherent organizational structure to the three clauses of the “commercial activities” exception. If “doing business” is allowed as a basis for jurisdiction under clause one, then the three clauses of § 1605(a)(2) roughly correspond to three jurisdictional categories; personal jurisdiction exists where the defendant (1) does business in the United States, (2) commits an act within the United States but is not “doing business,” and (3) does not commit a relevant act in the United States, but *207causes a direct effect in the United States.1 These categories reflect the development of jurisdictional principles, from Pennoyer v. Neff, 95 U.S. 714, 24 L.Ed. 565 (1877) (doing business), through International Shoe (committing an act can constitute minimum contacts), McGee v. International Life Insurance Co., 355 U.S. 220, 78 S.Ct. 199, 2 L.Ed.2d 223 (1957) (effects can constitute minimum contacts), and beyond. See Note, Effects Jurisdiction Under the Foreign Immunities Act and the Due Process Clause, 55 N.Y.U.L.Rev. 474 (1980). Given these three historical grounds of jurisdiction and the appearance of § 1605(a)(2) as embodying these grounds, it seems reasonable to read clause one as embodying a “doing business” ground in addition to a “nexus” ground when the language can support this construction.
The legislative history also supports this construction, although the evidence is not conclusive. In discussing § 1330(b), the “Personal Jurisdiction” section — the most significant juncture of the section-by-section analysis for our purposes — the House Report stated:
Section 1330(b) provides, in effect, a Federal long-arm statute over foreign states (including political subdivisions, agencies, and instrumentalities of foreign states). It is patterned after the long-arm statute Congress enacted for the District of Columbia. Public Law 91-358, sec. 132(a), title I, 84 Stat. 549. The requirements of minimum jurisdictional contacts and adequate notice are embodied in the provision. Cf. International Shoe Co. v. Washington, 326 U.S. 310 [66 S.Ct. 154, 90 L.Ed. 95] (1945), and McGee v. International Life Insurance Co., 355 U.S. 220, 223 [78 S.Ct. 199, 2 L.Ed.2d 223] (1957). For personal jurisdiction to exist under section 1330(b), the claim must first of all be one over which the district courts have original jurisdiction under section 1330(a), meaning a claim for which the foreign state is not entitled to immunity. Significantly, each of the immunity provisions in the bill, sections 1605-1607, requires some connection between the lawsuit and the United States, or an express or implied waiver by the foreign state of its immunity from jurisdiction ____
House Report at 13, reprinted in 1976 U.S.C.C.A.N. at 6612. The statement that “the requirements of minimum jurisdictional contacts” are embodied in the Act, with the citations to International Shoe and McGee, supports a reading of the Act as coextensive with the limits of due process. There are other statements on which one may rely and other inferences one may draw from the legislative history, but this strong statement at the critical juncture is the more potent.
One may point to other statements in the legislative history which seem to emphasize the importance of the claim's relation to the United States as requiring a nexus, such as the above statement that sections 1605— 1607 require “some connection between the lawsuit and the United States____” But in the context of the entire quotation the connection can be equally read to be “doing business.” Significantly, Congress specifically approved “doing business” jurisdiction for § 1605(a)(3) expropriations cases. Thus “doing business” must equal “some connection between the lawsuit and the United States” for the sentence to make sense. It follows for me that the majority opinion’s reliance on the idea that “doing business” would not fulfill the goal of clause one in requiring a connection between the lawsuit and the United States is misplaced.
The focus of the House Report is not on “doing business” as a jurisdictional ground. *208Yet, it does not necessarily follow that Congress’ failure affirmatively to mention “doing business” jurisdiction evidences a congressional intent that this avenue to jurisdiction be unavailable to the plaintiff suing a foreign sovereign. “Doing business” was an established, basic jurisdictional ground, and it is unlikely that Congress would dismiss it without a clearer indication of its intent to do so. Cf. Texas Trading v. Federal Republic of Nigeria, 647 F.2d at 313 (“Before the FSIA plaintiffs enjoyed a broad right to bring suits against foreign states, subject only to State Department intervention and the presence of attachable assets. Congress in the FSIA certainly did not intend significantly to constrict jurisdiction; it intended to regularize it.”). At the same time it was reasonable for Congress to focus on a nexus ground in its explanations, as the experience before passage of the FSIA was that the vast majority of claims specifically related to the contacts within the forum that gave rise to jurisdiction. See L. Henkin, R.C. Pugh, O. Schacter and H. Smit, International Law 514 (2d ed. 1979).
In light of the pre-FSIA history it is then understandable that the focus would be on the dispute’s connection with the United States. In a direct sense a dispute with an entity “doing business” in the United States does have a relation to the United States, although the simple fact of “doing business” may not be enough for the suit to survive a motion for dismissal on forum non conveniens grounds. Finally, if the allowance of “doing business” jurisdiction meant that all suits between foreigners could get into U.S. courts so long as there was a “doing business” connection, then the legislative history would most certainly support a construction of the language as rejecting “doing business.” However, as I explain below, “doing business” would not alone create jurisdiction.
My final argument is of the “forest for the trees” variety. We should not, in parsing the legislative history, overlook Congress’ overall purpose. By codifying the restrictive theory of sovereign immunity, Congress recognized that those aggrieved by the acts of governmental commercial entities ought to be able to sue those governmental entities just as other, private, commercial entities could be sued. It seems the most probable congressional purpose was to extend “doing business” jurisdiction to foreign governments’ commercial entities, thus placing those entities in positions to be sued not dissimilar from other foreign-owned companies.
At the same time the FSIA’s “doing business” jurisdiction ought to be strictly limited to what I perceive to be Congress’ purpose in its enactment. I think it self-evident that Congress did not intend to extend “doing business” jurisdiction to suits between foreigners unless, as in this case, the jurisdiction of the federal courts would extend to cover the case absent sovereign immunity. Notwithstanding that under the FSIA Congress did in general extend federal jurisdiction to suits between foreigners as long as one of the exceptions to immunity is present, see Verlinden B.V. v. Central Bank of Nigeria, 461 U.S. 480, 103 S.Ct. 1962, 76 L.Ed.2d 81 (1983), I think Congress would have been explicit had it meant to extend United States jurisdiction to all suits between foreigners and foreign states without any nexus at all between the cause of action and the United States. Extending “doing business” jurisdiction to any claim arising anywhere in the world between foreigners and foreign states would be an unprecedented and radical extension of United States jurisdiction; surely it was not contemplated by the FSIA. My reading of the FSIA as allowing for “doing business” jurisdiction allows for “doing business” jurisdiction only in cases that would otherwise be cognizable in federal courts under this jurisdictional ground absent any issue of sovereign immunity. Congress was unquestionably wary lest our federal courts become international courts of claims. With the interpretive burden Congress has given the courts in administering the statute and the strong congressional purpose pushing toward allowing “doing business” as a ground, a halfway construction is justified.
*209Even if we allay the concern that the FSIA could comprehend suits between foreigners whenever a foreign state is “doing business” in the United States, “doing business” is a more controversial jurisdictional ground to assert over foreign entities than is a nexus ground. See Carl, Suing Foreign Governments in American Courts: The United States Foreign Sovereign Immunities Act in Practice, 33 Sw.L.J. 1009, 1058-61 (1979). But c.f. Restatement of Foreign Relations Law of the United States (Revised), Tent.Draft No. 2 § 403(2)(b) (1981) (suggesting “links such as nationality, residence, or economic activity, between the regulating state and the persons principally responsible for the activity to be regulated” as considerations recognized by international law that are relevant to a country’s assertion of “jurisdiction to prescribe” (emphasis supplied)). Given some judicial readings of “doing business” jurisdiction, any queasiness of those at the international bar over our “doing business” jurisdiction is not without some justification. See, e.g., Bryant v. Finnish National Airline, 15 N.Y.2d 426, 260 N.Y.S.2d 625, 208 N.E.2d 439 (1965) (“doing business” jurisdiction found on basis that defendant maintained a one-and-one-half room office for publicity purposes in New York). Yet I am confident that U.S. courts could responsibly assert “doing business” jurisdiction in the FSIA context in such just ways as to avoid affronts to foreign sovereigns. By adhering strictly to the “substantial contact” standard, and by use of forum non conveniens dismissals, our courts could screen out those cases that do not belong here. My construction would thus not threaten to turn our courts into international courts of claims. The power to hear a claim and the willingness to do so are quite different.
Here, although the case would likely succumb to a motion to dismiss on forum non conveniens grounds, see, e.g., Perusakaan Umum Listrik Negara v. M/V Tel Aviv, 711 F.2d 1231 (5th Cir.1983); Veba-Chemie A.G. v. M/V Getafix, 711 F.2d 1243 (5th Cir.1983), absent CNAN’s status as an instrumentality of a foreign government the admiralty jurisdiction of the federal courts over the claim against CNAN would be unquestionably established, notwithstanding that both Vencedora and CNAN are foreigners. Involved are time-worn elements of a complaint charging a salvor with misdeeds: that the salvor without good reason prevented the distressed vessel’s master for rejoining his ship,2 that the salvor without good reason did not comply with the request of the owner to bring the salved property into a nearby, safe port, and that these misdeeds resulted in the destruction of the vessel. See 3A Benedict on Admiralty §§ 125, 119 at 8-39 (7th ed. revised 1980); Byrne v. Johnson, 53 F. 840 (5th Cir.1893). Absent in rem jurisdiction, all Vencedora would have to establish would be that jurisdiction over CNAN would be consistent with due process. See The Belgenland, 114 U.S. 355, 5 S.Ct. 860, 29 L.Ed. 152 (1884). Thus since this case would otherwise be cognizable in our admiralty courts absent a claim of sovereign immunity, the purpose behind the FSIA pushes toward an interpretation that would allow for our exercise of jurisdiction.
D.
The majority opinion frets over Vencedora’s foreign status. But the tenuousness of any connection between this particular dispute and the United States is no reason to disallow “doing business” as a jurisdictional ground in all cases. Had an American company initiated this suit the equities might well lie in favor of U.S. jurisdiction. Disallowing “doing business” jurisdiction under the FSIA may adversely affect not just foreigners whose suits clearly do not *210belong in the U.S., but also U.S. citizens whose suits would otherwise be cognizable in U.S. courts if the defendant was not a sovereign instrumentality. This inquiry is relevant to our central effort of plumbing congressional intent. I am persuaded that had Congress anticipated the issue it would not wish to deny access to U.S. citizens injured abroad by a foreign sovereign entity doing business in the United States.
It might be argued that clause three of § 1605(a)(2) would prevent U.S. citizens in this situation from being denied access to U.S. courts. Clause three allows for jurisdiction over claims based upon acts outside the U.S. in connection with commercial activity outside the U.S. where there are direct effects in the United States. Although there is little precedent on the scope of the “direct effects” clause, it is doubtful that this clause would avoid the loss caused to U.S. plaintiffs by the disallowance of “doing business” jurisdiction.
Clause three lays out two relevant limitations on jurisdiction: (1) the effect must be “direct,” and (2) the effect must be “in the United States.” Though Congress gave little hint of what is meant by “direct,” a thoughtful commentator points out that not all injuries proximately caused by a tort will amount to a “direct effect.” See Note, Effects Jurisdiction Under the Foreign Sovereign Immunities Act and the Due Process Clause, 55 N.Y.U.L.Rev. 474, 506 (1980). Thus, the limitation that an injury be “direct” in all likelihood restricts the reach of U.S. jurisdiction over actions by U.S. plaintiffs based on acts outside our territory, in comparison to “doing business.” Under “doing business” jurisdiction, if the defendant “does business” in the U.S. the fact that the injury suffered in the United States was not the primary injury would not defeat jurisdiction.
It is generally easy to determine whether a personal injury has occurred “in the United States.” Courts have naturally adopted a test of looking to see where the injury occurred, rather than whether or not the injury has occurred to a person who lives in the United States. If section 1605(a)(2) does not include a “doing business” test, then use of this “direct effects” test alone would of course cut off a number of U.S. plaintiffs from U.S. courts in suits based on torts occurring abroad where the defendant does business in the U.S.
It is more problematic to decide whether a financial injury has occurred “in the United States.” Nevertheless, it is likely that U.S. business plaintiffs suffering financial injury would be foreclosed from bringing suit in U.S. courts in a significant number of cases unless U.S. courts took the unlikely and arguably improper course of adopting a test for “directness” that went beyond primary injury as well as a broad test for location of injury. In sum, “direct effects” jurisdiction would not completely mitigate the loss by U.S. plaintiffs of forum selection caused by ridding the courts of “doing business” jurisdiction over sovereigns. I think this concern is relevant to congressional purpose and supports my view that “doing business” jurisdiction should be recognized under the FSIA.
E.
Underlying the court’s opinion is the fear of affronts to foreign sovereigns hauled into our courts over disputes having nothing to do with the United States. Undoubtedly many cases over which United States courts could have jurisdiction only because of an FSIA “doing business” exception ought not be heard in the United States. But an over-broad jurisdictional ban is not the best way to exclude such cases. Perhaps I am more confident in our ability to weed out cases on an individualized basis. But, more importantly, I believe those cases that could survive a motion for forum non conveniens dismissal — especially such cases involving U.S. plaintiffs — ought to be heard because Congress meant for commercial, governmental entities to be subject to suit like any other commercial entity.
. The relevant portion of § 1605(a)(2) provides: (a) A foreign state shall not be immune from the jurisdiction of courts of the United States or of the States in any case ... (2)[ (i) ] in which the action is based upon a commercial activity carried on in the United States by the foreign state; [ (ii) ] or upon an act performed in the United States in connection with a commercial activity of the foreign state elsewhere; [ (iii) ] or upon an act outside the territory of the United States in connection with a commercial activity of the foreign state elsewhere and that act causes a direct effect in the United States ....
. Vencedora alleges that CNAN refused to allow the master of the KAPETAN MARCOS to accompany the tugs on their salvage mission. Vencedora says that the survivors of the crew of the KAPETAN MARCOS arrived in Bejaia before CNAN ordered its tugs from Algerian ports to salvage the KAPETAN MARCOS. Vencedora states that the port authorities at Annaba led the master of the KAPETAN MARCOS to believe that he would be permitted to board one of the tugs and return to his ship, but that the CNAN tugs left without him.