By late 1984 New World Entertainment, Ltd. (New World) had experienced some success making and distributing low-budget movies. New World approached Balcor Entertainment Company, Ltd. (BEC), a subsidiary of Shearson Lehman/American Express, Inc., to solicit working capital. New World wanted to expand; BEC wanted some of the profits to be made in movies. There were the makings of a deal: BEC would obtain financing for New World's films in exchange for part of the films' profits. Because BEC didn't want to accept the risk of being the sole source of capital, it recruited outside investors. Balcor Film Investors (BFI) is a limited partnership formed to raise money for eight to twelve low budget movies that New World would produce and distribute.

Early in 1985 BFI registered partnership interests under the Securities Act of 1933 and commenced public solicitation. Subscriptions received were to be held in escrow until $50 million had been secured. Failing to raise this amount by the deadline, BFI reduced the minimum to $35 million, offered to refund the investors' money, and began soliciting anew. By the end of 1985 $48 million was on deposit, and BFI closed the offering. During the first two years BFI supplied the tax benefits for which the investors had hoped. But New World's films flopped. In 1988 BFI told its limited partners that they were likely to lose some of their capital. Investors then filed class action suits against Shearson (now known as Shearson Lehman Hutton, Inc.), BFI, BEC, the other Balcor entities that had put together the offering, and their officers and directors (collectively Balcor). There are two groups of plaintiffs: a class of investors who read the prospectus (the Majeski plaintiffs) and a class of those who did not (the Eckstein plaintiffs). The Majeski plaintiffs assert a standard fraud theory: they purchased the limited partnership interests in reliance on the misrepresentations in the prospectus and its omissions of material facts. The Eckstein plaintiffs, whose defining characteristic is failure to read the prospectus, contend that but for the misrepresentations and omissions the offering would not have been successful. This group, in other words, asserts causation in lieu of reliance.

Although errors and omissions in the offering documents usually lead to liability under §§ 11 and 12(2) of the '33 Act, 15 U.S.C. §§ 77k, l (2), the plaintiffs feared that their suits would be untimely under § 13 of that Act, the statute of limitations applicable to actions under §§ 11 and 12. So they invoked § 10(b) of the Securities Exchange Act of 1934, 15 U.S.C. § 78j(b), and the SEC's Rule 10b-5, 17 C.F.R. § 240.10b-5. This avenue is available even though the investors complain about an initial public offering rather than a transaction in the aftermarket. See Herman & MacLean v. Huddleston, 459 U.S. 375, 103 S. Ct. 683, 74 L. Ed. 2d 548 (1983). Section 10(b) and Rule 10b-5 require the investors to show fraud, not just material errors and omissions. In exchange they get a longer statute of limitations--or so they thought.

The Majeski plaintiffs filed in October 1988 and the Eckstein plaintiffs in February 1989. When they began the litigation, courts throughout the nation derived from state law the periods of limitations in § 10(b) cases. On July 30, 1990, this court overruled opinions that had looked to state law and announced that § 13 of the '33 Act supplies the statute of limitations. Short v. Belleville Shoe Manufacturing Co., 908 F.2d 1385 (7th Cir.1990). On June 20, 1991, the Supreme Court agreed with Short that the federal securities laws are the source of the period of limitations, but the Court selected § 9(e) of the '34 Act, 15 U.S.C. § 78i(e), as the most appropriate rule. Lampf, Pleva, Lipkind, Prupis & Petigrow v. Gilbertson, --- U.S. ----, ---- n. 9, 111 S. Ct. 2773, 2782 n. 9, 115 L. Ed. 2d 321 (1991). Both § 13 of the '33 Act and § 9(e) of the '34 Act give an investor one year from discovering the facts constituting the violation, but no more than three years from the violation, to begin suit. The advantage of using § 10(b) disappeared.

Congress responded to Lampf by enacting stopgap legislation. A new § 27A of the '34 Act, 15 U.S.C. § 78aa-1(a), provides that "[t]he limitation period for any private civil action implied under section [10(b) of the '34 Act] that was commenced on or before June 19, 1991, shall be the limitation period provided by the laws applicable in the jurisdiction, including principles of retroactivity, as such laws existed on June 19, 1991." Plaintiffs believe that this law saves their suits. The district court disagreed, dismissing both as untimely. 786 F. Supp. 1458 (E.D.Wis.1992). Although § 27A avoided Lampf, the district court held, it did not disturb Short, which was the law in the seventh circuit on June 19, 1991. And Short, the district court held, is fully retroactive. It concluded that Short governs the Eckstein plaintiffs, who filed in California, as well as the Majeski plaintiffs, who filed in Wisconsin. The Panel on Multidistrict Litigation had transferred the Eckstein case to Wisconsin under 28 U.S.C. § 1407 for consolidated pretrial proceedings, and the district court then completed the transfer by invoking 28 U.S.C. § 1404(a) to make the transfer permanent. After the transfer the whole case was in Wisconsin, which the district court believed is "the jurisdiction" to which § 27A refers. Both groups of plaintiffs have appealed.

* Whether the Eckstein plaintiffs have appealed correctly is another matter. The Majeski plaintiffs may have destroyed the Eckstein plaintiffs' notice of appeal, leaving us without appellate jurisdiction to review the appeal of the Eckstein class. At the heart of the matter is the question: was the Eckstein case fully consolidated with the Majeski case? If the cases were fully consolidated the Eckstein appellants have problems.

Here is what happened. On March 11, 1992, the district court entered judgments dismissing both actions. The Majeski plaintiffs filed a timely motion under Fed.R.Civ.P. 59. Shortly thereafter, the Eckstein plaintiffs filed a notice of appeal. The district court denied the Rule 59 motion, and the Majeski plaintiffs filed a timely notice of appeal. Then the Eckstein plaintiffs filed what they called a "Supplemental Notice of Appeal". The clerk of this court treated that document as another notice of appeal and assigned it a new number. It did not last long: the Eckstein plaintiffs failed to pay the docketing fee, and as a notice of appeal this paper was at all events untimely, coming 34 days after the district court denied the Rule 59 motion--four more than Fed.R.App.P. 4(a)(1) permits. Thus our jurisdiction over the Eckstein plaintiffs' case depends on their notice of appeal filed before the district court acted on the Majeski plaintiffs' Rule 59 motion.

If the district court fully consolidated the cases, the Ecksteins have no valid notice of appeal and we have no appellate jurisdiction. Rule 4(a)(4) specifies the "time for appeal for all parties" when "any party" files a timely Rule 59 motion. See Polara v. Trans World Airlines, Inc., 284 F.2d 34 (2d Cir.1960); Continental Casualty Co. v. United States, 167 F.2d 107 (9th Cir.1948). If consolidation produced a single "case," then the Eckstein plaintiffs' first notice of appeal had "no effect" because it was filed before disposition of the Majeski plaintiffs' Rule 59 motion. Fed.R.App.P. 4(a)(4).

Did the district court consolidate the cases? Yes and no. As often happens, the district court did not clearly explain the extent to which the actions were consolidated. See Ivanov-McPhee v. Washington National Insurance Co., 719 F.2d 927 (7th Cir.1983). The district court ordered "the Eckstein and Majeski actions ... consolidated for the remainder of the pretrial proceedings and trial" and certified each set of plaintiffs as a subclass of a single class of all investors in BFI. A month later the court added: "the actions are not consolidated for 'all purposes' nor are they merged into a single cause of action" (emphasis in original). What can this mean when the court had certified a single class spanning both sets of plaintiffs? Because neither case contained the entire class, if the cases were not fully consolidated, there could not have been subclasses. In Ivanov-McPhee, 719 F.2d at 930, we held that cases were fully consolidated because they could have been brought as a single unit and there was no purpose in the cases retaining separate identities. Sandwiches, Inc. v. Wendy's International, Inc., 822 F.2d 707, 710 (7th Cir.1987), adds that courts should keep in mind the relation between consolidation and appeal. Preserving formally separation may multiply the number of appeals, which should not occur when there is only one nucleus of facts. In order for plaintiffs to make up a single class their claims must have a great deal in common. Fed.R.Civ.P. 23(b)(3). We should not have to review twice claims common enough to be classed together. By all lights, the Majeski and Eckstein cases should have been consolidated for all purposes.

Nonetheless, the district court entered two judgments, just as if the suits were separate. Different plaintiffs appealed from each judgment. Appellate jurisdiction follows the judgment, providing the certainty that is essential when a misstep may forfeit a valuable right. Separate judgments are independently appealable, and no one need appeal until the formal judgment under Fed.R.Civ.P. 58 has been entered. United States v. Indrelunas, 411 U.S. 216, 220-22, 93 S. Ct. 1562, 1564-65, 36 L. Ed. 2d 202 (1973). Sometimes a party may appeal from a final decision not embodied in a judgment, see Bankers Trust Co. v. Mallis, 435 U.S. 381, 98 S. Ct. 1117, 55 L. Ed. 2d 357 (1978), but it is always entitled to wait for (and rely on) the separate Rule 58 judgment. Shalala v. Schaefer, --- U.S. ----, ----, 113 S. Ct. 2625, 2632, 125 L. Ed. 2d 239 (1993); Mallis, 435 U.S. at 386, 98 S.Ct. at 1120; In re Kilgus, 811 F.2d 1112, 1117 (7th Cir.1987). Anything else baits a trap for unwary litigants--for wary ones, too, the kind who pay particular attention to judgments. The Eckstein plaintiffs therefore were entitled to rely on the fact that the district judge issued two judgments. They appealed in a timely manner from theirs, and the Majeski plaintiffs' motion did not affect that appeal.

BFI used a single prospectus for the two offerings of its partnership interests. When BFI re-offered these interests in the fall of 1985 it furnished investors with a supplement describing recent events. The plaintiffs' claims center on this supplement, which they say left out information about business reverses New World suffered in 1985. New World was making low-budget films that had limited attendance at the box office and reaped most of their profits from television and from the sale and rental of video cassettes for home viewing. A firm called Worldvision had agreed to distribute New World's films outside the theaters. According to plaintiffs, the supplement to the prospectus fraudulently omitted the fact that Worldvision and New World had had a falling out--that Worldvision, dissatisfied with the quality of the films, was attempting to withdraw as distributor and had filed a suit against New World. Plaintiffs also contend that the offering materials concealed the fact that New World was to keep the lion's share of the profits from VCR distribution.

These adverse events occurred before the investors sent in their money between October 11, 1985, and December 31, 1985. But, plaintiffs insist, they were unaware that New World had encountered troubles until BFI told them in mid-1988 that they probably would not get all of their money back. The Majeski plaintiffs filed suit in the Eastern District of Wisconsin in October 1988, while the Eckstein plaintiffs filed suit in the Central District of California in February 1989. Because different issues determine the applicable limitations period for each group of plaintiffs, we discuss them separately.

* The Eckstein plaintiffs' case moved from California to Wisconsin under 28 U.S.C. § 1404(a). Section 27A instructs us to use the "laws applicable in the jurisdiction" on June 19, 1991. The district in which the court that decides the case sits must be "the jurisdiction" for purposes of § 27A. But what is the law of that jurisdiction? Short identifies the statute of limitations in force within the seventh circuit on June 19, 1991. Choice-of-law rules are part of any jurisdiction's whole law, however, a principle important in diversity cases. See Klaxon Co. v. Stentor Electric Manufacturing Co., 313 U.S. 487, 61 S. Ct. 1020, 85 L. Ed. 1477 (1941). Usually we do not think of federal courts as having choice-of-law rules for cases decided under federal law, which is supposed to be uniform. Yet there is one distinctively federal choice-of-law problem. What happens when a federal court transfers the case to another under § 1404(a)? Van Dusen v. Barrack, 376 U.S. 612, 84 S. Ct. 805, 11 L. Ed. 2d 945 (1964), and Ferens v. John Deere Co., 494 U.S. 516, 110 S. Ct. 1274, 108 L. Ed. 2d 443 (1990), hold that such a transfer leaves the law unaffected; the original forum's rules (including the choice-of-law rules of the state in which that court sits) are unaffected by the movement.

Are different circuits like different states for the purposes of Van Dusen and Ferens? Usually not. In re Korean Air Lines Disaster, 829 F.2d 1171 (D.C.Cir.1987) (Ruth B. Ginsburg, J.), affirmed on other grounds under the name Chan v. Korean Air Lines, Ltd., 490 U.S. 122, 109 S. Ct. 1676, 104 L. Ed. 2d 113 (1989). A single federal law implies a national interpretation. Although courts of appeals cannot achieve this on their own, the norm is that each court of appeals considers the question independently and reaches its own decision, without regard to the geographic location of the events giving rise to the litigation. This meant to the judges in Korean Air Lines that a transfer under § 1404(a) leaves each court to work out the problem for itself rather than to guess how the circuit comprising the original district court would reason. Congress might require one federal court to apply another's interpretation of federal law, but § 1404(a) does not itself do so. Id. at 1178 (D.H. Ginsburg, J., concurring). See also Richard L. Marcus, Conflicts Among Circuits and Transfers Within the Federal Judicial System, 93 Yale L.J. 677, 702 (1984).

We agree with Korean Air Lines that a transferee court normally should use its own best judgment about the meaning of federal law when evaluating a federal claim, but § 27A instructs us to act differently. Section 27A recognizes that different circuits had taken different approaches to the appropriate statute of limitations in suits under § 10(b), and it codifies this fractured nature of federal law for cases filed before June 20, 1991. Congress requires us to apply federal law as courts understood it at a point in the past rather than to make an independent judgment about what that law actually is. And the law in use on June 19, 1991, was not nationally uniform. By then three circuits had adopted § 13 of the '33 Act for suits under § 10(b). One applied this rule retroactively, another prospectively (in the main), and this circuit had not ruled on retroactivity. In all other circuits the courts derived the period of limitations from state law--with different circuits looking to different kinds of state laws. See Pommer v. Medtest Corp., 961 F.2d 620, 627-28 (7th Cir.1992); Norris v. Wirtz, 818 F.2d 1329, 1331-33 (7th Cir.1987).

Recently the second circuit concluded that because "federal law (unlike state law) is supposed to be unitary", a transferee court should apply the law of its circuit and ignore the law of the transferor court when determining "the jurisdiction" under § 27A. Menowitz v. Brown, 991 F.2d 36, 40 (2d Cir.1993). Menowitz held that Van Dusen and Ferens apply only in diversity cases. We believe that this conclusion disregards both the language of § 27A (whose reference to "the jurisdiction" implies a non-uniform federal law) and the holdings of Van Dusen and Ferens, which construed § 1404(a) rather than any principle of state law. Although both of those cases arose under the diversity jurisdiction, their references to Erie R.R. v. Tompkins, 304 U.S. 64, 58 S. Ct. 817, 82 L. Ed. 1188 (1938), do not imply a ruling limited to state law. Erie is itself part of national law, interpreting the Rules of Decision Act, 28 U.S.C. § 1652. Van Dusen and Ferens accordingly apply whenever different federal courts properly use different rules. Erie is not unique in requiring federal courts to apply disparate norms. Consider 42 U.S.C. § 1988, which leads federal courts to use state law as the basis of a federal period of limitations. Wilson v. Garcia, 471 U.S. 261, 105 S. Ct. 1938, 85 L. Ed. 2d 254 (1985); Owens v. Okure, 488 U.S. 235, 109 S. Ct. 573, 102 L. Ed. 2d 594 (1989). Or consider the law of federal contracts, which is largely borrowed from state law. United States v. Kimbell Foods, Inc., 440 U.S. 715, 99 S. Ct. 1448, 59 L. Ed. 2d 711 (1979); United States v. Einum, 992 F.2d 761 (7th Cir.1993). When the law of the United States is geographically non-uniform, a transferee court should use the rule of the transferor forum in order to implement the central conclusion of Van Dusen and Ferens: that a transfer under § 1404(a) accomplishes "but a change of courtrooms". Van Dusen, 376 U.S. at 639, 84 S.Ct. at 821. Section 27A presents such a situation. Accordingly we respectfully disagree with Menowitz.

So we must examine how a judge in the Central District of California would have viewed the limitations question on June 19, 1991. At the time, courts within the ninth circuit used the most analogous statute of limitations under the law of the state in which the district court sat. Stitt v. Williams, 919 F.2d 516, 522 (9th Cir.1990). California's three year statute of limitations for fraud, Cal.Civ.Proc.Code § 338, was the appropriate one to use in securities fraud cases. Ibid. The period under that statute "commences when the plaintiff discovered or could have discovered the fraud with the exercise of reasonable diligence". SEC v. Seaboard Corp., 677 F.2d 1301, 1309 (9th Cir.1982). See also Miller v. Bechtel Corp., 33 Cal. 3d 868, 875, 191 Cal. Rptr. 619, 623, 663 P.2d 177, 181 (1983); Mosesian v. Peat, Marwick, Mitchell & Co., 727 F.2d 873, 877 (9th Cir.1984).

The district court concluded that the Eckstein plaintiffs' claim would be barred by that rule as well as the federal rule adopted in Short. It concluded that the time for bringing suit started to run when the investors purchased their limited partnership interests because the dire warnings of risk in the prospectus should have put the investors on notice of fraud. 786 F. Supp. at 1466. This logic is flawed. A warning about risk does not give notice of fraud. Every prospectus is filled with advice about the perils of the business, because every business faces its own set of risks. Giving general advice about the industry does not relieve an issuer of the need to be truthful about itself. Warnings of the kind that appeared in BFI's prospectus put investors on notice that there is a chance things may not go as hoped in the future; they do not put investors on notice that statements made in the prospectus are untrue at the time, or that important facts have been left out of the prospectus. Plaintiffs allege, for example, that New World and Worldvision had come to blows, and that this gravely affected the opportunity for profitable investment. A prospectus stating a risk that such a thing could happen is a far cry from one stating that this had happened. The former does not put an investor on notice of the latter.

Plaintiffs' argument that they "could [not] have discovered the fraud with the exercise of reasonable diligence" until Balcor told them in 1988 they probably would lose money is equally flawed. Discovering that one has lost money is not the same as discovering that one has been defrauded. Most losses occur without fraud of any kind. DiLeo v. Ernst & Young, 901 F.2d 624, 627-28 (7th Cir.1990). And victims of fraud usually discover the problem long before the wrongdoer stands up and confesses. Plaintiffs' position amounts to an assertion that the time to sue does not start until the extent of the injury becomes clear. But the question under California law is when the investors discovered (or should have discovered) the deceit, not when the full consequences of that deceit are felt. An investor can be defrauded before any loss is realized. Discovery of the fraud means the discovery of the misrepresentation. Howard v. Haddad, 962 F.2d 328, 330 (4th Cir.1992); Volk v. D.A. Davidson & Co., 816 F.2d 1406, 1412-13 (9th Cir.1987). Investors must bring their claims as soon as they become aware of misrepresentations or omissions, instead of waiting "while avoidable damages accrue." Volk, 816 F.2d at 1412.

It is conceivable, we suppose, that until receiving the advice in mid-1988 reasonable investors would not have bestirred themselves to find out about New World. (The Eckstein plaintiffs, after all, invested in BFI without reading the prospectus; apparently they do not think information very useful.) Many reasonable investors--the kind who read the prospectus?--would have pricked up their ears sooner. The fact that Worldvision filed suit against New World in the fall of 1985 to rescind its contract may have been sufficient to trigger an investigation. The suit itself was a piece of information missing from the prospectus; from one angle its omission was the fraud, yet the fact of the suit was public knowledge in 1985. If that should have triggered an investigation, then California's statute of limitations expired before the Eckstein plaintiffs filed their suit in February 1989. That plaintiffs may have been ignorant of Worldvision's suit should not matter, because they, not the defendants, bear the burden of their own ignorance. See United States v. Kubrick, 444 U.S. 111, 124, 100 S. Ct. 352, 360, 62 L. Ed. 2d 259 (1979); Norris, 818 F.2d at 1335-36. It may be possible to decide what reasonable investors knew, or should have known, without submitting the question to a jury, but the subject is one for the district court to address in the first instance.

Unlike the Eckstein plaintiffs, the Majeski plaintiffs have clearly satisfied the relevant state statute of limitations. Wisconsin gives investors three years from the sale, Wis.Stat. § 551.59, and the Majeski plaintiffs filed before the end of 1988, at a time when we would have used that law as the period of limitations. Cahill v. Ernst & Ernst, 625 F.2d 151 (7th Cir.1980). But § 13 of the '33 Act gives investors only one year from the time the fraud was or should have been discovered, and the considerations we discussed immediately above imply that the Majeski plaintiffs would have a hard time satisfying that requirement. The district court held that § 13 applies to the Majeski plaintiffs because they filed suit within the seventh circuit after Short had been decided, and that the suit is untimely under § 13.

The district court concluded that Short is fully retroactive. Under current law, it would be. James B. Beam Distilling Co. v. Georgia, --- U.S. ----, 111 S. Ct. 2439, 115 L. Ed. 2d 481 (1991), holds that decisions are retroactive if the court applies the new rule to the parties. See also Harper v. Virginia Department of Taxation, --- U.S. ----, 113 S. Ct. 2510, 125 L. Ed. 2d 74 (1993). Short applied its new rule to the litigants in that case. But § 27A tells us to use the law, including principles of retroactivity, in force on June 19, 1991, and Beam, like Lampf, was decided on June 20, 1991. We concluded in McCool v. Strata Oil Co., 972 F.2d 1452, 1458-59 (7th Cir.1992), that Beam changed the law of retroactivity, which means that on June 19, 1991, Chevron Oil Co. v. Huson, 404 U.S. 97, 92 S. Ct. 349, 30 L. Ed. 2d 296 (1971), still controlled the approach to that subject. McCool holds that § 27A requires us to employ Chevron's approach. To apply Chevron courts must balance several factors. After a consideration of these factors McCool concluded that Short does not apply retroactively when the plaintiff relied on the pre-Short limitations period. 972 F.2d at 1459.

The dispositive question under McCool thus is: if the plaintiffs had known that Short would apply to their claim, could and would they have filed within the period of limitations? The Majeski plaintiffs say that they filed suit within two months of discovering the fraud, so it is hard to see how any delay can be chalked up to reliance on the availability of Wisconsin's law. But the belief that state law would provide three years may have influenced the investors in other ways. Perhaps it offered assurance that they need not investigate expeditiously. Perhaps the Majeski plaintiffs relied by filing in Wisconsin when they could have chosen California. Although Balcor insists that filing in one forum as opposed to another is not a legitimate form of reliance, we disagree. McCool held that the plaintiffs had relied on the former status of the law by dismissing a suit they had filed in state court. 972 F.2d at 1459. We do not believe that relying by foregoing the opportunity to pursue a claim in a state forum should differ from foregoing the opportunity to pursue a claim in another federal forum.

Ultimately, the determination whether the plaintiffs did rely is for the district court. Chevron creates a balancing approach, and the court of first instance does the balancing with deferential appellate review. Indeed if material factual questions about reliance are controverted the matter cannot be decided on summary judgment at all. Berning v. A.G. Edwards & Sons, Inc., 990 F.2d 272, 276-77 (7th Cir.1993).

Despite our disagreement with the district court's reason for dismissing the suits, a remand does not necessarily follow. We may affirm its judgment on any properly preserved ground that the record supports. Massachusetts Mutual Life Insurance Co. v. Ludwig, 426 U.S. 479, 96 S. Ct. 2158, 48 L. Ed. 2d 784 (1976). A timely but non-meritorious suit should be cut off at the first opportunity. We therefore inquire whether the plaintiffs have a good claim under the securities laws.

* Because they never read the prospectus, the Eckstein plaintiffs encounter difficulty in establishing that they relied to their detriment on the seller's statements, a component of a claim under § 10(b) and Rule 10b-5 according to the canonical formulation. See List v. Fashion Park, Inc., 340 F.2d 457 (2d Cir.1965). Although courts often refer to "reliance" as an element of the claim under § 10(b), we have held that reliance is a means by which the plaintiff may establish that material misstatements or omissions caused him injury, rather than an indispensable element. Flamm v. Eberstadt, 814 F.2d 1169, 1173 (7th Cir.1987). See also Rowe v. Maremont Corp., 850 F.2d 1226, 1233 (7th Cir.1988). The Supreme Court's adoption of the fraud-on-the-market doctrine in Basic, Inc. v. Levinson, 485 U.S. 224, 108 S. Ct. 978, 99 L. Ed. 2d 194 (1988), shows that reliance is not essential; although Basic continued to use that word, it allowed an alternative method of establishing causation--an effect on the market price--to support recovery by investors who never read the supposedly deceitful statement. Id. at 243-47, 108 S.Ct. at 989-92. See also Affiliated Ute Citizens v. United States, 406 U.S. 128, 152-54, 92 S. Ct. 1456, 1471-72, 31 L. Ed. 2d 741 (1972).

When "the market"--that is, the outcome of trading by persons who are well-informed about what the issuer is doing and saying--translates a lie or omission from voice to price, it is easy to see how injury can befall a person who is unaware of the deceit. The price in an open and developed market usually reflects all available information, because the price is an outcome of competition among knowledgeable investors. See generally Sanford J. Grossman, The Informational Role of Prices (1989); James H. Lorie, Peter Dodd & Mary Hamilton Kimpton, The Stock Market: Theories and Evidence (2d ed. 1985). Competition among savvy investors leads to a price that impounds all available information, even knowledge that is difficult to articulate. We call a market "efficient" because the price reflects a consensus about the value of the security being traded--not necessarily because the price captures the true value of the firm's assets but because the price is the best available device to assess the significance of additional bits of information. Investors who trade at the market price are affected, for good or ill, by the information underlying that price.

Not all stocks are eagerly followed by astute investors with the capital to turn their views into movements in price. The more thinly traded the stock, the less well the price reflects the latest pieces of information. "Efficiency" is not an all-or-nothing phenomenon. See Comment, Sufficient Efficiency: Fraud on the Market in the Initial Public Offering Context, 58 U.Chi.L.Rev. 1393 (1991). Prices of even poorly followed stocks change in response to news, including statements by the issuers, and these changes may be better indicators of causation than litigants' self-serving statements about what they read and relied on and about what they would have paid (or whether they would have bought at all) had the issuer said something different. Even an "inefficient" market price is objective and contemporaneous with events, not plagued by lapses of memory or the cognitive dissonance that influences what witnesses may remember years later. Still, at some point the market process peters out and the litigation process offers superior information about causation. The Eckstein plaintiffs may have reached that point.

BFI issued its interests as part of an initial public offering at a fixed price, $1,000 per unit (with a minimum of three units per investor). No trading market valued these interests; only the investors could do so. No trading market developed afterward, so we cannot combine the Capital Asset Pricing Model with the tables in the Wall Street Journal to see what effect Worldvision's suit, or the other information that slowly came to light about New World, had on price. It would be revealing if news about Worldvision's withdrawal as New World's distributor caused the price to fall by 20% (using the CAPM to hold the market constant). It would be equally revealing if the news had no effect on price, which would imply either that other distributors were readily available (so that information about Worldvision was not material) or that investors had fully taken this fact into account back in 1985 (so the statute of limitations would bar these suits). Alas, no such luck, because there are no such prices. This does not mean that the Eckstein plaintiffs cannot show causation, but they must carry the greater burden of proving the causal links that an efficient secondary market establishes automatically.

The Eckstein plaintiffs try to do so via a theory we could call "fraud-created-the-market." BFI's offering was conditioned on its ability to raise at least $35 million. A minimum sales requirement may serve two functions: it ensures that the venture has sufficient capital to function, and it provides a form of vicarious protection to ignorant investors who assume that the condition will be met only if a significant number of informed buyers think the project a good investment. Plaintiffs allege that, if BFI had made complete and truthful statements, it would not have been able to sell interests to investors who did read the prospectus. Without those investors, BFI would not have been able to sell the minimum amount, and thus would have returned the plaintiffs' tendered funds. Thus, say the Eckstein plaintiffs, the misstatements and omissions in the prospectus caused their losses.

The fifth circuit adopted a variant of this approach by a vote of 12 to 10 in Shores v. Sklar, 647 F.2d 462 (1981) (in banc). Shores held that an investor may maintain an action under § 10(b) by establishing that the fraud permitted the securities to exist in the market--that but for the fraud the securities would have been "unmarketable"--and that the investor relied on their existence. The tenth and eleventh circuits follow modified versions of the Shores approach, while the sixth circuit has repudiated that case outright. Compare T.J. Raney & Sons, Inc. v. Ft. Cobb Oklahoma Irrigation Fuel Authority, 717 F.2d 1330 (10th Cir.1983), and Ross v. Bank South, N.A., 885 F.2d 723 (11th Cir.1989) (in banc), with Freeman v. Laventhol & Horwath, 915 F.2d 193 (6th Cir.1990). We agree with the sixth circuit. The existence of a security does not depend on, or warrant, the adequacy of disclosure. Many a security is on the market even though the issuer or some third party made incomplete disclosures. Federal securities law does not include "merit regulation." See The Business Roundtable v. SEC, 905 F.2d 406 (D.C.Cir.1990); § 23 of the '33 Act, 15 U.S.C. § 77w. Full disclosure of adverse information may lower the price, but it does not exclude the security from the market. Securities of bankrupt corporations trade freely; some markets specialize in penny stocks. Thus the linchpin of Shores--that disclosing bad information keeps securities off the market, entitling investors to rely on the presence of the securities just as they would rely on statements in a prospectus--is simply false.

Without the aid of Shores, the Eckstein plaintiffs have rough sledding ahead. They cannot use the incomplete or rosy nature of Balcor's pamphlets, which they may have read, as the actionable "fraud"; sales literature need not repeat the full disclosures and risk analysis in the prospectus. (We discuss this more fully below.) 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 decision-making 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. Still, the record in its current state does not doom their case, so we must remand their case to the district court for further proceedings.

The Majeski plaintiffs, who read the prospectus, may establish causation in the traditional way, by proof of their reliance on any misstatements and omissions. These plaintiffs advance a second theory--that because Balcor circulated upbeat brochures and other supplemental literature that was much easier to understand than the dense prospectus, the defendants should have put the stern warnings about risk in these documents as well. Why, one of these brochures even offered tables of possible rates of return under different assumptions, and the worst case still showed BFI able to repay the original investments (although without interest or other profits). Because these easy-to-digest brochures dominated the investors' perception of the risks, plaintiffs insist, Balcor was obligated to make additional disclosures of risk in the supplemental literature. To simplify the conduct of this litigation on remand, we now hold that such a theory cannot support a claim under § 10(b)--which, recall, depends on proof of fraud, and not just errors or omissions. Ernst & Ernst v. Hochfelder, 425 U.S. 185, 96 S. Ct. 1375, 47 L. Ed. 2d 668 (1976); Aaron v. SEC, 446 U.S. 680, 100 S. Ct. 1945, 64 L. Ed. 2d 611 (1980).

The '33 Act permits issuers, underwriters, and dealers to engage in "free writing" once the registration statement becomes effective, and to furnish promotional literature to investors provided that literature is accompanied or preceded by a prospectus. Section 2(10)(a) of the '33 Act, 15 U.S.C. § 77b(10)(a); Louis Loss, Fundamentals of Securities Regulation 119 (1983). Only the registration statement need be self-contained. A prospectus is a subset of the information contained in the registration statement, and the sales brochures are a subset of the information in the prospectus (plus the customary effort to sell the securities). If the sales literature had to contain all the warnings that appear in the prospectus, the privilege of distributing supplemental sales literature would be all but meaningless. Federal law establishes a regime in which the prospectus contains the comprehensive description of the securities. Other literature can be brief precisely because an inquiring investor has the prospectus to turn to. Federal law also establishes a rule for resolving conflicts: in the event statements in sales brochures and the prospectus do not agree, the prospectus wins. See Brown v. E.F. Hutton Group, Inc., 991 F.2d 1020, 1033 (2d Cir.1993). "If the investor already possesses information sufficient to call [a] representation into question, he cannot claim later that he relied on or was deceived by the lie." Teamsters Local 282 Pension Trust Fund v. Angelos, 762 F.2d 522, 530 (7th Cir.1985). See also, e.g., Atari Corp. v. Ernst & Whinney, 981 F.2d 1025, 1030-31 (9th Cir.1992); Zobrist v. Coal-X, Inc., 708 F.2d 1511, 1518-19 (10th Cir.1983). The Majeski plaintiffs do not contend that the risk disclosures in the prospectus were buried or indigestible; to the contrary, they were prominent and blunt. Cf. Virginia Bankshares, Inc. v. Sandberg, --- U.S. ----, ---- - ----, 111 S. Ct. 2749, 2760-61, 115 L. Ed. 2d 929 (1991); Associated Randall Bank v. Griffin, Kubik, Stephens & Thompson, Inc., 3 F.3d 208 (7th Cir.1993); Acme Propane, Inc. v. Tenexco, Inc., 844 F.2d 1317, 1322, 1325 (7th Cir.1988). Failure to disclose important things in supplemental literature is not fraud when those things appear in the prospectus.

One final observation. Many of the claims in this case arise out of predictions that did not come to pass. For example, the Majeski plaintiffs allege that Balcor stated that presales of movies would generate revenues equal to half of the movies' production costs, and that this statement is false. Yet the statement is nothing but a prediction about how much revenue New World expected to generate from preselling movies. Only statements or omissions of fact can be fraudulent. Although intentions and beliefs are "facts" for this purpose when they are open to objective verification, Sandberg, --- U.S. at ----, ----, 111 S. Ct. at 2760, 2765, an inability to foresee the future does not constitute fraud, because "[t]he securities laws approach matters from an ex ante perspective". Pommer, 961 F.2d at 623. See DiLeo, 901 F.2d at 627-28. If those statements had a reasonable basis when made, the defendants did not commit fraud. Wielgos v. Commonwealth Edison Co., 892 F.2d 509, 513 (7th Cir.1989).

The judgments are vacated, and the cases are remanded for further proceedings consistent with this opinion. As the district court dismissed the plaintiffs' claims under state law only because it had dismissed all of their claims under federal law, the state-law claims must be reinstated.