EASTERBROOK, Circuit Judge.

A decade ago Balcor Film Investors (BFI) sold $48 million of securities to the public through an offering registered under the Securities Act of 1933. Litigation commenced in 1988 and 1989 became snagged by a change of rules defining the period of limitations. When last the case was here, we held that the suit by the Eckstein class is problematic on the merits, while the suit by the Majeski class may or may not be timely. We remanded for further proceedings to address these subjects. 8 F.3d 1121 (1993). On remand the district court granted summary judgment for the defendants, holding that the Eckstein class cannot show reliance on any material misstatements or omissions and that the Majeski action is too late. Our first opinion lays out the essential facts; we restate only the minimum necessary to grasp the remaining issues.

I.

The Majeski action was filed in Wisconsin three years to the day after the registration statement became effective, posing serious problems under the one-and-three year period of limitations and repose established by Short v. Belleville Shoe Manufacturing Co., 908 F.2d 1385 (7th Cir. 1990). On the first appeal, we held that § 27A(a) of the Securities Exchange Act of 1934, 15 U.S.C. § 78aa-1(a), requires us to apply Short unless the plaintiffs relied to their detriment on pre- Short law. 8 F.3d at 1128-29. (Plaut v. Spendthrift Farm, Inc., ___ U.S. ___, 115 S.Ct. 1447, 131 L.Ed.2d 328 (1995), which holds § 27A(b) unconstitutional, does not affect § 27A(a).) One form of detrimental reliance, we concluded, could be filing in Wisconsin rather than in another jurisdiction that, under § 27A(a), would apply state law rather than the federal period of limitations established in Short. The Majeski class asserted that, had it anticipated Short, it would have filed suit in federal court in California, which would have applied California's longer period of limitations. (The only other form of reliance the class advances, that it has lavished time and money on federal litigation, is insufficient. Lewis v. Long Grove Trading Co., 13 F.3d 1028 (7th Cir. 1994).) California gives the investor three years from discovering the fraud, while federal law allows only one year. The district court accepted the Majeski plaintiffs' representation about their litigation strategy but held that it would have done them no good — because, the court concluded, a California court would have been led to the fatal one-and-three year period by California's borrowing statute:

When a cause of action has arisen in another state, or in a foreign country, and by the laws thereof an action thereon cannot there be maintained against a person by reason of the lapse of time, an action thereon shall not be maintained against him in this state.

CCP § 361. The Majeski suit is doomed if California courts would borrow the federal law that, according to Short and § 27A(a), applies to securities suits filed in Wisconsin. But, they insist, a California court — and therefore a federal court applying California's choice-of-law rules under Klaxon Co. v. Stentor Electric Manufacturing Co., 313 U.S. 487, 61 S.Ct. 1020, 85 L.Ed. 1477 (1941) — would borrow Wisconsin's state law rather than the federal law prevailing in Wisconsin.

The linchpin of this argument is the contention that "the laws thereof" in § 361 points exclusively to state or foreign law. The introductory phrase refers to actions that have "arisen in another state, or in a foreign country", so the antecedent of "laws thereof" must be state and foreign law, the argument runs. This is not, however, what § 361 says. The initial reference is to the place where the claim arose, not to the legislature or judicial system of that place. The Majeski class claim arose within the State of Wisconsin (so the class concedes). Wisconsin is both a state and a part of the United States; to say that a claim arose "in" a state is to say that it arose in the United States. Then § 361 calls on us to determine what law was in force in Wisconsin. That is a question answered long ago.

The laws of the United States are laws in the several States, and just as much binding on the citizens and courts thereof as the State laws are. The United States is not a foreign sovereignty as regards the several States, but is a concurrent, and, within its jurisdiction, paramount sovereignty. Every citizen of a State is a subject of two distinct sovereignties, having concurrent jurisdiction in the State, — concurrent as to place and persons, though distinct as to subject-matter. . . . The disposition to regard the laws of the United States as emanating from a foreign jurisdiction is founded on erroneous views of the nature and relations of the State and Federal governments.

Claflin v. Houseman, 93 U.S. 130, 136-37, 23 L.Ed. 833 (1876). Federal law — which is to say, Short — applies to claims in Wisconsin under the federal securities acts, and Wisconsin's period of limitations to claims based on state substantive law.

We have approached the interpretation of § 361 as a matter of first principles because the courts of California have never considered whether to distinguish between state and federal periods of limitations under that statute. They have never had any reason to do so, because federal periods of limitations normally are uniform across the nation, and when the law is uniform it makes no sense to think of "borrowing" one state's federal period to apply to a federal suit filed in another state. Section 27A(a) created a temporary geographic non-uniformity by requiring courts to apply the law in force in each jurisdiction on June 19, 1991, the day before Lampf, Pleva, Lipkind, Prupis Petigrow v. Gilbertson, 501 U.S. 350, 111 S.Ct. 2773, 115 L.Ed.2d 321 (1991), resolved a conflict among the circuits. Given Klaxon, the law in force on June 19, 1991, in a district court sitting in California includes CCP § 361, which sends us back to Wisconsin, which means Short.

This is not the first time a federal court has had to divine the answer to a question of state law that state courts themselves have not addressed and are unlikely ever to address. Marrese v. American Academy of Orthopaedic Surgeons, 470 U.S. 373, 105 S.Ct. 1327, 84 L.Ed.2d 274 (1985), posed the question whether a state would treat a boycott suit under state law as barring a later suit, based on federal antitrust law, arising out of the same conduct. Because the effect of a state judgment in federal court depends on the state's law of preclusion, 28 U.S.C. § 1738, the state's answer was controlling; yet given the widespread assumption that federal courts have exclusive jurisdiction over suits under the Sherman Act no state court had addressed the issue. What to do? The Supreme Court told us to make the best guess we can, drawing parallels to the state's treatment of similar issues. 470 U.S. at 381-82, 105 S.Ct. at 1332. Cf. Nolan v. Transocean Air Lines, 276 F.2d 280, 281 (2d Cir. 1960) ("Our principal task . . . is to determine what the New York courts would think the California courts would think on an issue about which neither has thought. They have had no occasion to do so.") (Friendly, J.). So we asked the parties at oral argument whether the state courts of California had addressed this or any similar issue concerning borrowing. Neither side was aware of a pertinent case. Plaintiffs scared up a few decisions by federal courts interpreting California's borrowing statute, but none attempted to predict how California's own courts would deal with geographically non-uniform federal law. The closest match we could find is not from California but from Ohio — and for reasons related to the exclusive — jurisdiction problem that set the stage for Marrese, Ohio's answer comes not from an Ohio court but from the Supreme Court of the United States.

Cope v. Anderson, 331 U.S. 461, 67 S.Ct. 1340, 91 L.Ed. 1602 (1947), was a banking suit to be decided under national law. Congress had not specified a statute of limitations, so the federal court borrowed one from state law — not as a matter of state law, but as a matter of national law. See Wilson v. Garcia, 471 U.S. 261, 105 S.Ct. 1938, 85 L.Ed.2d 254 (1985); see also North Star Steel Co. v. Thomas, ___ U.S. ___, 115 S.Ct. 1927, 132 L.Ed.2d 27 (1995). But which state's law? A suit was filed in federal court in Ohio. Following Klaxon, the Court turned to Ohio's choice-of-law rules and found its borrowing statute, similar in all interesting respects to California's. 331 U.S. at 464-65, 67 S.Ct. at 1341-42. It was met with the response that Ohio's law seemed to allow the borrowing only of another state's period of limitations for claims based on state law, yet the Court needed to select a federal period of limitations for a claim arising under federal law.

The place where the events giving rise to a cause of action occur is said to be "important only insofar as the laws of that place are controlling." Under this argument, the cause of action here could not have "arisen" in any state since the statutory obligation of shareholders was not imposed or controlled by state law. Hence, the argument runs, the Ohio law did not contemplate borrowing any state statute of limitations in a case where liability is governed by federal law. And no federal statute of limitations could be borrowed in this case for none existed.

331 U.S. at 465, 67 S.Ct. at 1342. No problem, the Court concluded; the Ohio law should not be understood to distinguish state from federal law. With the distinction cleared away, the Ohio statute pointed to Kentucky, whose five-year period the Court used as a federal statute of limitations for the case.

Cope is not a perfect parallel; the principal argument there was that a state borrowing statute should not be used when the substance of the claim depends on federal law; here the Majeski class wants to avoid borrowing the federal period of limitations that Short established for claims in Wisconsin. But in the end the principle is the same: federal law is part of the law "in" a state, and when a claim arises in that state, a borrowing rule picks up federal law just as much as it selects state law of the state. Any other understanding would create serious problems under the Supremacy Clause. Suppose Wisconsin had a period of limitations that had been preempted by a federal statute or deemed unconstitutional as discriminatory against interstate commerce. Would California really attempt to apply Wisconsin's state law, without regard to the principles of federal law that affect the outcome? We doubt it. The district court therefore was right to think that a court in California would have used the California borrowing statute to apply the one-and-three year period.

This conclusion makes the Majeski claim untimely. The class advances a different view, but the district court's conclusion was inevitable. Section 13 of the '33 Act, to which we looked in Short, gives investors the lesser of three years from the time the security was first "bona fide offered to the public" or one year from the time the fraud was discovered. The year begins when a reasonable investor knew, or through the exercise of reasonable diligence could have learned, enough to commence an inquiry into the possibility of deceit. Tregenza v. Great American Communications Co., 12 F.3d 717 (7th Cir. 1993); cf. Norris v. Wirtz, 818 F.2d 1329 (7th Cir. 1987); Lasalle v. Medco Research, Inc., 54 F.3d 443 (7th Cir. 1995). The Majeski class filed suit on October 11, 1988, barely satisfying the three-year period of repose. But unless the events of which they complain remained secret until October 12, 1987 — secret even from persons investigating with reasonable diligence — the suit is too late. The two principal matters underpinning the suit would not have remained hidden from any inquirer, even a casual one. (Plaintiffs complain of other omissions and statements, which do not require separate discussion. The time begins when the investors should have discovered the general fraudulent scheme, not when they should have discovered particular details. See DeBruyne v. Equitable Life Assurance Society, 920 F.2d 457, 466 (7th Cir. 1990); Louis Loss Joel Seligman, 10 Securities Regulation 4495-97 n. 78 (1993).)

The prospectus for the securities revealed that BFI would reinvest the money in films to be made by or through New World Entertainment, a producer of low-budget pictures. The prospectus stated that New World (and hence persons who invested through BFI) would receive profits from the movies' sale on videotape. The Majeski plaintiffs say that they relied on the anticipation of these after-market revenues when investing. The prospectus omitted the potentially significant fact that only 22.5% of the home video receipts would flow to BFI. This omission of a material fact necessary to make the statements in the prospectus non-misleading was fraud, plaintiffs contend. Yet the gory details about the 22.5% cut were in view throughout 1985 — they were in the registration statement filed with the Securities Exchange Commission. The registration statement is a public document, well known to all of the underwriters and dealers plus many of the brokers, and accessible to anyone else who wants it. The prospectus informed investors that details of the home video arrangement could be obtained from BFI; an obvious alternative was to look them up in the registration statement, which included a copy of the contract. Every investor therefore had constructive notice of the 22.5% figure on the date the securities were issued. ("Constructive notice" here is the usual shorthand, meaning "no actual knowledge, but the ability to acquire knowledge by reasonably diligent inquiry.")

Relying on Corwin v. Marney, Orton Investments, 843 F.2d 194, 198 (5th Cir. 1988), and Briskin v. Ernst Ernst, 589 F.2d 1363, 1367-68 (9th Cir. 1978), plaintiffs insist that they were free to disregard the registration statement, or at least that the extent to which information in a registration statement gives constructive notice is a question of fact that may not be resolved on summary judgment. Neither case says this. Corwin dealt with documents in the files of some bureau in Texas, not with widely available registration statements. Briskin discussed the significance of information in registration statements the issuer had used for securities other than the ones in suit. It seems to us that the whole purpose of the registration statement for an issue of securities is to provide information to the purchaser — whether directly (a few purchasers read these tomes) or indirectly (by influencing investment banks, professional investors, and ultimately the market price of the securities). A registration statement is supposed to collect the most salient information, some of which goes into the prospectus. Errors and omissions in the registration statement lead to liability under § 11 of the Securities Act of 1933, 15 U.S.C. § 77k. It would be lopsided, and no favor to investors as a whole (who in the end must pay anticipated judgments), if errors in the registration statement could produce liability but truthful information in the registration statement were legally insignificant.

Registration statements do not impart actual knowledge to the average investor. That is why the prospectus, a subset of the registration statement, is prepared and distributed. Cf. Gustafson v. Alloyd Co., ___ U.S. ___, ___, 115 S.Ct. 1061, 1071, 131 L.Ed.2d 1 (1995). A prospectus may be misleading (hence actionable under § 12 of the '33 Act) and even fraudulent (hence actionable under § 10(b) of the '34 Act), if it omits material information. The presence of that information in the registration statement does not save the prospectus from a charge of incompleteness. But the presence of the information in the registration statement does give actual notice to professional investors and constructive notice to ordinary investors. It therefore starts the one-year statute of limitations. Accord, Block v. First Blood Associates, 988 F.2d 344 (2d Cir. 1993). Notice from filing systems is an old idea — it is the basis of resolving land title disputes, and financing statements under the Uniform Commercial Code give notice to lenders, who are deemed to have received that notice whether or not they search the records. The year's breathing room under the statute of limitations gives investors a comfort unknown to lenders and buyers of real estate.

The other information that the plaintiffs say was concealed is that by late 1985 New World and its distributor Worldvision had soured on each other. Worldvision thought New World's films of poor quality (seven of New World's recent films had bombed in the theaters) and too rife with sex and violence to be shown on TV. Litigation broke out, and the distribution agreement eventually was canceled. Plaintiffs believe that information about the theatrical flops, the simmering dispute, and the litigation should have been in the prospectus. But the lawsuit, which laid bare the reasons for Worldvision's disquiet, was reported in the press in December 1985. In February 1986 BFI put details, including the fact that Worldvision had stopped paying New World, in its quarterly report on Form 8-K, filed with the SEC and sent to all of its investors. Again this report afforded constructive notice — and, to those who read it, actual knowledge. Settlement of the suit was announced in the Wall Street Journal on January 7, 1987. In the first half of 1987 BFI told the investors that New World's films were doing poorly and that they were apt to suffer losses. Still plaintiffs sat on their hands, doing nothing until receiving notice in 1988 that the probability of loss had become a certainty. As we observed in the first appeal, the notice in 1988 is unrelated to fraud; many businesses suffer losses without fraud of any kind. 8 F.3d at 1127-28. See also DiLeo v. Ernst Young, 901 F.2d 624 (7th Cir. 1990); Goldberg v. Household Bank, F.S.B., 890 F.2d 965 (7th Cir. 1989). Investors cannot sit tight waiting to be told about losses; they might as well claim that the time does not begin to run until they have been told that it would be prudent to sue. The fraud, if there was fraud, lay in omitting from the prospectus information material to the likelihood of profit. The news missing from the prospectus was in the public domain more than a year before the Majeski class filed suit.

This understanding also shows why the district court properly rejected plaintiffs' contention that BFI is equitably estopped to plead the statute of limitations. During 1986 and 1987 BFI made upbeat statements about the venture's prospects. In saying that these delayed the time for suit, the Majeski class makes the same mistake that lies behind its argument that it could not have discovered any fraud until told that loss was certain. Whether a given investment makes or loses money is by and large unrelated to the question whether the prospectus was actionable. Despite errors and omissions, things often turn out well; at the same time, the most scrupulous disclosures do not ensure business success. The securities laws ask whether the disclosures were proper at the time; they use an ex ante perspective, and how things turn out ex post does not matter to liability (although they may affect damages). See Pommer v. Medtest Corp., 961 F.2d 620, 623 (7th Cir. 1992); Wielgos v. Commonwealth Edison Co., 892 F.2d 509, 518 (7th Cir. 1989); Jordan v. Duff Phelps, Inc., 815 F.2d 429, 440-41 (7th Cir. 1987); Escott v. BarChris Construction Corp., 283 F. Supp. 643, 669-70, 675 (S.D.N.Y. 1968). BFI told the investors that it hoped, even expected, that New World's films would be commercial successes and that they would therefore make money. It did not tell the investors any lies about the state of affairs in October 1985, the only relevant time for the securities action, and therefore, it is entitled to the benefit of the statute of limitations.

After the remand from this court in 1993, plaintiffs sought leave to amend their complaint to add a claim under the Racketeer Influenced and Corrupt Organizations Act (RICO), which has a four-year statute of limitations. See Agency Holding Corp. v. Malley-Duff Associates, Inc., 483 U.S. 143, 107 S.Ct. 2759, 97 L.Ed.2d 121 (1987). The district court denied leave to amend and did not abuse its discretion in doing so. Wade v. Hopper, 993 F.2d 1246, 1248 (7th Cir. 1993). Years into a complex piece of commercial litigation, a district judge is entitled to treat the issues as frozen. The case is approaching seven years old, and the underlying events happened a decade ago. The parties and courts have invested heavily in resolving the limitations question under the securities laws; it is too late to say "Never mind!" and scoot off in a different direction. Our remand order was designed to simplify the litigation and to specify the issues requiring resolution; the district judge correctly resisted an effort to transform it into a new case, with a brand new batch of problems.

II.

The district judge certified Robert and Sylvia Eckstein as representatives of this subclass:

all persons who purchased defendant Balcor Film Investors limited partnership interests without relying on any of the public offering materials.

134 F.R.D. 240, 251 (E.D.Wis. 1991) (emphasis in original). Because the Eckstein class is defined as the class that did not rely on (and presumably did not read) the offering materials, it has a substantial problem with the reliance requirement under § 10(b) and Rule 10b-5. Investors who sue under § 11 or § 12 of the '33 Act need not establish reliance, but the Eckstein class has not pursued that remedy because it filed suit after the time provided by § 13 of the '33 Act, 15 U.S.C. § 77m. Its claim is timely only because the Eckstein class filed in California, and for reasons we need not relate the defendants do not contend that the California borrowing statute pointed back to Wisconsin for the Eckstein class. To prevail the Eckstein class must show actual fraud, in violation of § 10(b) and Rule 10b-5. The patter of reminders that this requires proof of reliance, e.g., Gustafson, ___ U.S. at ___, 115 S.Ct. at 1073; Central Bank of Denver, N.A. v. First Interstate Bank of Denver, N.A., ___ U.S. ___, ___, ___, 114 S.Ct. 1439, 1449, 1455, 128 L.Ed.2d 119 (1994), means that the class definition itself spells doom for the class. How can a class defined as those who did not rely show the manner of their reliance?

Our answer two years ago was that "reliance" is a synthetic term. It refers not to the investor's state of mind but to the effect produced by a material misstatement or omission. Reliance is the confluence of materiality and causation. The fraud on the market doctrine is the best example; a material misstatement affects the security's price, which injures investors who did not know of the misstatement. See Basic, Inc. v. Levinson, 485 U.S. 224, 108 S.Ct. 978, 99 L.Ed.2d 194 (1988). But the Eckstein class bought its securities in their original issue; a liquid trading market did not translate an omission into an effect on price. We held on the first appeal that the fraud-on-the-market doctrine cannot be directly extended to the original issue of stock. 8 F.3d at 1129-31. Nonetheless, we concluded, one avenue remains to the Eckstein subclass:

To prevail the Eckstein plaintiffs must prove that, had the prospectus been free from fraud, BFI would not have satisfied the minimum-sale requirement of $35 million. Because this is a suit under § 10(b) of the '34 Act rather than §§ 11 or 12(2) of the '33 Act, the Eckstein plaintiffs must establish this counterfactual proposition about the decisionmaking of thousands of investors using only statements or omissions amounting to fraud; other errors and omissions that might have supported liability under §§ 11 or 12(2) do not support an inference of causation that can replace direct reliance in a case under § 10(b). The difference between errors and fraud, and the fact that BFI attracted $48 million, substantially exceeding the $35 million cutoff, present the Eckstein plaintiffs with a daunting task.

8 F.3d at 1131. On remand the Ecksteins offered only their own affidavit: they would not have bought the securities had they known then what they know now about New World and Worldvision. The district judge thought this woefully deficient and ended the case. Right he was.

The Ecksteins needed to show that additional disclosures would have knocked out more than $13 million in sales. They might have gone about their task by finding out who actually bought the securities (professionals who knew about the movie industry? amateurs? people seeking tax shelters?) and asking how these people would have responded to new information. Professional investors, and those who knew about the movie business, doubtless were aware of the critical facts; if they bought in substantial quantities it would not be sensible to infer that more disclosures would have cut $13 million from the subscription. But if most buyers were amateurs who read the prospectus (an important assumption, for obvious reasons), and there were good alternative investments available at the time, then additional disclosures might have curtailed sales by enough to fall below the $35 million floor. Yet the Ecksteins did not introduce such information. We know essentially nothing about who bought the securities, what they knew when they acted, what other options were open to them, and how many of the buyers read the prospectus. This state of ignorance leads to defeat for the class.

All that the Eckstein class has to offer in response is the presumption that investors rely on material omissions. See Affiliated Ute Citizens v. United States, 406 U.S. 128, 152-54, 92 S.Ct. 1456, 1471-72, 31 L.Ed.2d 741(1972). Compare Mills v. Electric Auto-Lite Co., 396 U.S. 375, 384-85, 90 S.Ct. 616, 621-22, 24 L.Ed.2d 593 (1970), with Virginia Bankshares, Inc. v. Sandberg, 501 U.S. 1083, 1099-1102, 1106-08, 111 S.Ct. 2749, 276143, 2765-66, 115 L.Ed.2d 929 (1991). It is hard to conceive of "relying" on omitted information, and Affiliated Ute therefore devised a "presumption" of reliance — which to us is yet another indicator that the securities laws do not use "reliance" in the lay sense, that only the conjunction of materiality and causation matters. Again, however, the problem with this line of argument is the definition of the Eckstein subclass: "all persons who purchased defendant Balcor Film Investors limited partnership interests without relying on any of the public offering materials." People who did not rely on what was in the materials cannot have relied on what was omitted; to them the public offering materials were so much waste paper. That is why our first opinion directed the parties' attention to other means of showing reliance, all involving the intermediation of persons who did read and rely. The class definition bars any claim of direct reliance, and, as the effort to show intermediated effects has collapsed, the Eckstein class cannot prevail under § 10(b) and Rule 10b-5.

The district court dismissed state-law claims asserted under the supplemental jurisdiction. Both classes remain at liberty to refile in state court, if they believe that claims based on state law can avoid the problems they have encountered under the federal securities acts. They have no federal remedy, however, so the judgments are

AFFIRMED.