This diversity case presents questions of federal jurisdiction and Illinois tort law in an unseemly setting of bank fraud and attorney misconduct. In the early 1970s Michael Sullivan, a successful real estate lawyer and broker and a stockholder of the Ford City Bank, would refer clients to William Quinn, an officer of the bank, for loans and other bank services. In exchange Quinn would refer legal business to Sullivan. (Sullivan may have violated the Illinois Code of Professional Responsibility, which in Canon 2 provides that “a lawyer shall not promise or give another person anything of value to initiate a contact with a prospective client on behalf of the lawyer.” Ill.Rev.Stat. ch. 110A, foil. II774, Rule 2 — 103(d).) One of Quinn’s customers was Richard Orlak. He was a tile contractor with ambitions to be something grander, and a client of Sullivan. In February 1970 he wanted to borrow $6,000, but he could not obtain a loan in that amount from any bank because of his poor credit history, which included a bankruptcy. He turned to Sullivan for help, and Sullivan, although he knew about the bankruptcy, recommended Orlak to Quinn, who approved Or-lak’s application for the loan. Orlak did not disclose his bankruptcy on the application, although the application form required him to list all bankruptcy proceedings in which he had been involved.
The $6,000 loan inaugurated a substantial relationship between Orlak and the Ford City Bank. By the summer of 1972, *379Orlak, through a series of loans approved by Quinn and by another bank officer, Bruce Beede, owed the bank some $500,-000. Sullivan did not participate in the negotiations for these loans, but, as Orlak’s lawyer, was aware of them.
That summer Orlak got wind of an unimproved 17-acre parcel of land, later dubbed Neuport Estates, that was for sale for $190,000. Beede, Quinn, Orlak — and Sullivan — hatched a fraudulent plan to obtain financing for the purchase and development of the parcel. This plan included the formation of a partnership, in which Orlak and Sullivan would be disclosed partners and Quinn and Beede undisclosed ones, to buy the land; the concealment from the bank’s executive committee not only of the undisclosed (and unlawful) partnership interests of the bank officers but also of Orlak’s bankruptcy and of the purchase price; and the submission of an appraisal of the property as worth $540,000 developed, without revealing that it was as yet undeveloped. By means of these representations the partnership obtained a loan from the bank for $250,000, later raised to $270,000. The loan was secured by the land, which was placed in a land trust (“Land Trust No. 300”). An Illinois land trust is like a mortgage, so the bank corresponded to a mortgagee and the partnership to a mortgagor. Sullivan signed a personal guarantee of the trust’s debts. The guarantee figured in the trial, but as our disposition renders the guarantee issues moot we shall omit them to simplify the opinion.
For the next two years Orlak continued to borrow money fraudulently from the bank, using the good offices of Quinn and Beede. None of the loans was intended for the Neuport Estates project, and Sullivan, although he continued to represent Orlak, was not involved in negotiating any of them. However, Orlak used some of the proceeds of the loans to pay interest on the Neuport Estates loan — and Sullivan knew this, as he admitted in his deposition, and he also knew that Orlak was in financial trouble. Also during this period Sullivan received covert, plainly irregular cash payments from Beede. The source of these funds was the other loans, but Sullivan testified that he thought the source was the Neuport Estates loan, and the judge believed him.
Orlak’s house of cards collapsed in June 1974, when Quinn, realizing that Orlak was insolvent and hoping to save his own hide, ’fessed up to the president of the Ford City Bank. Orlak shortly defaulted, owing the bank more than $2.5 million. The bank demanded that Sullivan sign over his interest in Neuport Estates, and he complied. Although the bank ended up some $1.5 million in the hole after foreclosing on Orlak’s loans, Neuport Estates turned out to be one of his happier projects; the bank was able to complete and sell the project for more than the unpaid balance of the $270,000 loan and the costs of completion. It applied the proceeds of the sale to Orlak’s other loans.
Hartford Accident and Indemnity Company had insured Ford City Bank against defalcations by its officers. Hartford settled Ford City’s indemnity claim against it for $1.25 million plus an assignment of Ford City’s legal rights. Hartford is not a citizen of Illinois, so it was able to sue the bank officers, Orlak, and Sullivan — all of whom are citizens of Illinois, as is Ford City Bank — in federal district court. The suit charged a civil conspiracy. Only Sullivan fought the suit; the others, who unlike Sullivan had been prosecuted criminally, either settled or defaulted. Sullivan filed a third-party complaint against Ford City, seeking to recover the amount that Ford City had realized from the sale of Neuport Estates over and above the amount owed the bank under Land Trust No. 300. Sullivan’s principal ground was that the bank had breached a fiduciary duty to him by applying the proceeds to Orlak’s other loans. He also made this ground the basis of a counterclaim (really a set off) against Hartford.
Hartford’s claim against Sullivan, and Sullivan’s counterclaim against Hartford and third-party complaint against Ford City, were tried together, resulting in a decision by Judge Leighton that Sullivan was not liable to Hartford — because his *380participation in the conspiracy had been limited to the Neuport Estates loan, on which Ford City had lost no money — but that Ford City was liable to Sullivan to the tune of $280,000. The judge dismissed Sullivan’s counterclaim against Hartford. Hartford and Ford City appeal from the judgments against them. Sullivan has not cross-appealed from the dismissal of the counterclaim.
It is convenient to take up the question of federal jurisdiction over the third-party complaint first, because that question raises in turn a question about jurisdiction over the main complaint. Since Sullivan and Ford City are citizens of the same state and their claims present no federal questions, the suit between them can be maintained in federal court, if at all, only under the doctrine of ancillary jurisdiction. Had Ford City agreed to indemnify Sullivan for any judgment that Hartford might obtain against him, Sullivan could cite much authority for deeming his third-party complaint within the federal courts’ ancillary jurisdiction by reason of the great economy of trying such closely related claims in the same court. See, e.g., Revere Copper & Brass Inc. v. Aetna Casualty & Surety Co., 426 F.2d 709 (5th Cir.1970); Pennsylvania R.R. v. Erie Avenue Warehouse Co., 302 F.2d 843, 844-45 (3d Cir.1962). But, even more clearly, if Hartford had had claims against both Sullivan and Ford City arising out of Orlak’s fraudulent course of borrowing, it could not have used the doctrine of ancillary jurisdiction to get around the requirement of complete diversity by first suing Sullivan and then impleading Ford City. See Owen Equipment & Erection Co. v. Kroger, 437 U.S. 365, 374, 98 S.Ct. 2396, 2402, 57 L.Ed.2d 274 (1978); American National Bank & Trust Co. v. Bailey, 750 F.2d 577, 583 (7th Cir.1984); cf. Acton Co. v. Bachman Foods, Inc., 668 F.2d 76, 79-80 (1st Cir.1982). This case is somewhere in between these examples of proper and improper invocation of ancillary jurisdiction, but we think it closer to the improper pole and therefore order the dismissal of the third-party action for lack of federal jurisdiction.
Since Hartford was suing Sullivan by virtue of an assignment of Ford City’s rights, Sullivan could set up by way of defense to the suit any defenses he would have had if Ford City had sued him. And he did; he pleaded his claim to the return of his interest in Neuport Estates as a set off to Hartford’s claim for the damages that he had caused the bank by his participation in a conspiracy to defraud it. However, he had no right to an affirmative judgment against Hartford, which had taken an assignment of Ford City’s rights but not an open-ended assignment of Ford City’s liabilities. His only method of obtaining an affirmative judgment was to sue Ford City Bank. But he could have done this (had Hartford not sued him first in federal court) only in state court, and if Hartford had later sued him in federal court he could not have consolidated his state court action with his federal court defense. It is only because Hartford sued him before he got around to suing Ford City in state court that he can argue that since he was hauled into federal court he ought to be able to try his entire “defense” there, one part of which is an offense against Ford City Bank (but as the saying goes, a good offense is the best defense).
The issues in the two suits are closely related, but they are not identical, nor logically entwined in the sense that the decision in one suit would necessarily decide the other by operation of collateral estoppel or otherwise. None of the parties argues that if Hartford wins, Ford City must win, or that if Hartford loses, Ford City must lose. Hartford might for example have won only on Sullivan’s personal guarantee, in which event it would be entirely possible for Ford City to lose on Sullivan’s claim against it. Ancillary jurisdiction should be narrowly interpreted, in recognition that what is at stake is the sovereign right of the states to confine (so far as possible) to their own courts the litigation of issues of state law in disputes between their own citizens. The doctrine should not be used merely to enable state and federal suits to be consolidated in federal court whenever ordinary notions of judicial economy would make that a desirable result, but instead *381should be reserved for cases where failure to exercise federal jurisdiction would prevent a party from obtaining justice. This is not such a case. Having to bring parallel suits may be a hardship, but it is not an injustice. Had it not been for the accidents of timing, Sullivan would have found himself a plaintiff in state court, forced to litigate parallel but distinct lawsuits in different court systems as soon as Hartford sued him in federal court.
Many courts asked to establish the boundaries of ancillary jurisdiction in the setting of third-party claims draw an analogy to the distinction between compulsory and permissive counterclaims, the former arising out of the same “transaction or occurrence that is the subject matter of the” main claim, Fed.R.Civ.P. 13(a), the latter not, Fed.R.Civ.P. 13(b). Compulsory counterclaims are within the ancillary jurisdiction of the federal courts; permissive counterclaims are not. American National Bank & Trust Co. v. Bailey, supra, 750 F.2d at 583; EVRA Corp. v. Swiss Bank Corp., 673 F.2d 951, 959-60 (7th Cir.1982); Finkle v. Gulf & Western Mfg. Co., 744 F.2d 1015, 1018-19 (3d Cir.1984); H.L. Peterson Co. v. Applewhite, 383 F.2d 430, 433 (5th Cir.1967); Moor v. County of Alameda, 411 U.S. 693, 714-15, 93 S.Ct. 1785, 1798-99, 36 L.Ed.2d 596 (1973) (dictum); 13 Wright, Miller & Cooper, Federal Practice and Procedure § 3523, at pp. 106-09 (2d ed. 1984). The analogy is particularly apt in this case, given the close relationship between Ford City and Hartford as assignor and assignee; if they were one, Sullivan’s third-party claim would be a counterclaim.
Rule 14(a), under which Sullivan necessarily proceeded in impleading Ford City Bank, appears to cut across the distinction between compulsory and permissive counterclaims; it allows the defendant to im-plead a third party “who is or may be liable to the third-party plaintiff [i.e., the original defendant — Sullivan] for all or part of the plaintiffs claim against the third-party plaintiff.” The appearance may be deceptive, for it is hard to see how the condition we have just quoted could be fulfilled unless the third-party’s liability was derivative, Barab v. Menford, 98 F.R.D. 455, 456 (E.D.Pa.1983), and thus was readily seen to arise from the same dispute that gave rise to the plaintiff’s claim. This is the usual situation. Revere Copper & Brass Inc. v. Aetna Casualty & Surety Co., cited earlier, illustrates it. The defendant, Aet-na, had agreed to insure the plaintiff, Revere, against a breach by Fuller (later the third-party defendant) of his performance bond. Fuller in turn had agreed to indemnify Aetna for any losses to Aetna resulting from the contract of suretyship. Fuller breached his bond, Revere sued Aetna on the suretyship, so Aetna impleaded Fuller. Although Fuller may have had defenses based solely on the terms of his agreement to indemnify Aetna, the dispositive issues in the third-party claim would probably be the same as in the main claim: had Fuller breached the performance bond, and if so what had this cost Revere? So ancillary jurisdiction of the third-party claim was upheld, by explicit analogy to compulsory counterclaims. See 426 F.2d at 716; Mayer Paving & Asphalt Co. v. General Dynamics Corp., 486 F.2d 763, 772 (7th Cir.1973); United States v. City of Twin Falls, 806 F.2d 862, 867 (9th Cir.1986).
The analogy between counterclaims and impleader is not exact, however, and the result in these cases can therefore be questioned. A defendant who fails to plead a compulsory counterclaim in a federal suit may be barred from bringing a state suit based on the same claim. See, e.g., McDonald’s Corp. v. Levine, 108 Ill.App.3d 732, 743, 64 Ill.Dec. 224, 236, 439 N.E.2d 475, 487 (1982). Thus the filing of such a counterclaim, unlike the filing of a third-party claim, is in a sense involuntary, and it would be an injustice to the counterclaim-ant if the district court in effect barred his claim forever by refusing to exercise jurisdiction oyer it. But we are painting in too vivid colors. A defendant prevented by a jurisdictional ruling from filing a compulsory counterclaim is not likely, in fact, to be held to be barred from maintaining an independent suit on the same claim. A state court would have to be awfully severe, or worse, to bar a claim that the plaintiff *382could not have asserted as a counterclaim, no matter how hard he had tried, because the court in which the main claim had been filed would not have exercised jurisdiction over the counterclaim. The forfeiture of a claim that should have been but was not raised as a compulsory counterclaim is based on principles of or akin to res judica-ta, and the analogy implies that if the claimant had no actual chance to bring a compulsory counterelaim the claim should not be forfeited. Cf. Dindo v. Whitney, 451 F.2d 1, 3 (1st Cir.1971); LaFollette v. Herron, 211 F.Supp. 919 (E.D.Tenn.1962). Indeed, it seems odd even to describe as a counterclaim a claim that cannot be brought as a counterclaim because it is outside the jurisdiction of the court in which the main claim is pending.
At all events, impleader clearly is voluntary. If Aetna had not impleaded Fuller, it could still have sought indemnity from him if it lost against Revere.. Impleader was just a convenience. Maybe that should not be enough to allow a federal court to decide an issue of state law between two parties who are citizens of the same state, even if the issue arises out of the same transaction or occurrence that is the subject of the claim over which the court has unquestioned federal jurisdiction. So Judge Lumbard argued forcefully in dissent in Dery v. Wyer, 265 F.2d 804, 810-13 (2d Cir.1959), but the argument thus far has fallen on deaf ears. We need go no deeper into the question. The proper analogy to Sullivan’s claim against Ford City Bank is permissive, not compulsory, counterclaim. Hartford’s claim against Sullivan arises out of the fraudulent scheme by Orlak and others to defraud the bank, while Sullivan’s claim against the bank arises out of the bank’s efforts to cut its losses by foreclosing the Neuport Estates loan after the frauds were discovered. The claims involve considerable factual overlap, of course, but they arise at different times from different events, and thus flunk the test of same transaction or occurrence. That is the proper test to use in determining which Rule 14(a) claims are within the federal courts’ ancillary jurisdiction if the doctrine is narrowly construed in the im-pleader setting, as we think it should be, since the grounds for ancillary jurisdiction (a purely judge-made doctrine) are rather tenuous in that setting as we have seen. Dicta in a few cases can be read to suggest that any 14(a) claim is within the ancillary jurisdiction of the federal courts, see Moor v. County of Alameda, supra, 411 U.S. at 715, 93 S.Ct. at 1798; H.L. Peterson Co. v. Applewhite, supra, 383 F.2d at 433, but it is doubtful that the suggestion was intended, and if it was we respectfully disagree with it.
If the two suits had been so closely related that Sullivan v. Ford City Bank was within the federal district court’s ancillary jurisdiction, a question would arise whether Ford City’s assignment to Hartford was effective in creating diversity jurisdiction, for it might then be viewed as a device for getting around the requirement of complete diversity. Assignments designed to confer diversity jurisdiction, normally for purposes of facilitating the collection of a debt, are collusive and ineffective. See 28 U.S.C. § 1359 (“A district court shall not have jurisdiction of a civil action in which any party, by assignment or otherwise, has been improperly or collusively made or joined to invoke the jurisdiction of such court”); Kramer v. Caribbean Mills, Inc., 394 U.S. 823, 89 S.Ct. 1487, 23 L.Ed.2d 9 (1969); Steele v. Hartford Fire Ins. Co., 788 F.2d 441, 444 (7th Cir.1986). But there is no evidence that at the time of assignment Ford City Bank anticipated being named a third-party defendant, or that the assignment was anything more than a natural sequel to the insurance contract; so the principle that collusive assignments do not invoke the federal diversity jurisdiction is not in play. As the wording of section 1359 indicates, if the assignment is not collusive or otherwise improper, the assignee’s citizenship determines whether his suit is within the diversity jurisdiction. See, e.g., Westinghouse Credit Corp. v. Shelton, 645 F.2d 869, 870 (10th Cir.1981).
We can therefore proceed to the merits of Hartford’s claim, which is that the undisputed facts establish that Sullivan was a participant in the overall conspiracy with *383Orlak, Quinn, and Beede to defraud the Ford City Bank of millions of dollars, and not just in a lesser conspiracy to obtain a $270,000 loan, eventually repaid in full with interest, for the Neuport Estates project.
Relatively few Illinois, cases deal with civil conspiracy, but enough to make clear that Illinois courts do not take an idiosyncratic view of this tort. See Miller v. John, 208 Ill. 173, 70 N.E. 27 (1904); Wolf v. Liberis, 153 Ill.App.3d 488, 496,106 Ill.Dec. 411, 417, 505 N.E.2d 1202, 1208 (1987); Dymek v. Nyquist, 128 Ill.App.3d 859, 866, 83 Ill.Dec. 52, 58, 469 N.E.2d 659, 665 (1984); Reiter v. Illinois National Casualty Co., 328 Ill.App. 234, 258, 65 N.E.2d 830, 942 (1946), rev’d on other grounds, 397 Ill. 141, 73 N.E.2d 412 (1947). So we can draw on general principles. A further simplification is made possible by the fact that courts treat civil and criminal conspiracy alike — apart of course from standard of proof and other respects in which civil and criminal procedure differ — so that the abundant precedents on the meaning of criminal conspiracy are available for use in the civil context.
A conspiracy is an agreement, United States v. Borelli, 336 F.2d 376, 384 (2d Cir.1964) (Friendly, J.), but often an implicit one, and the scope of an implicit agreement — the issue here, as in so many other conspiracy cases, both civil and criminal — is hard to determine with precision. The best formulation may be Learned Hand’s in United States v. Andolschek, 142 F.2d 503, 507 (2d Cir.1944):
“a party to a conspiracy need not know the identity, or even the number, of his confederates; when he embarks upon a criminal venture of indefinite outline, he takes his chances as to its content and membership, so be it that they fall within the common purposes as he understands them. Nevertheless, he must be aware of those purposes, must accept them and their implications, if he is to be charged with what others may do in execution of them.”
See also Bell v. City of Milwaukee, 746 F.2d 1205, 1255 (7th Cir.1984); Halberstam v. Welch, 705 F.2d 472, 481 (D.C. Cir.1983).
The precise issue here is whether there were two conspiracies — one centering on the loan for Neuport Estates, the other on the other loans to Orlak — or one giant conspiracy to defraud the bank, with the loan for Neuport Estates just one component of it. These characterizations are not mutually exclusive, but only if the second is permissible can Sullivan be held liable for the losses inflicted on the bank (and hence on Hartford) by virtue of the other loans to Orlak. Everything depends on whether it can be said that the entire course of dealings with Orlak had a common purpose to which Sullivan knowingly subscribed, though he need not have been continuously involved in the fraudulent enterprise. See, e.g., United States v. Davis, 838 F.2d 909, 913 (7th Cir.1988); United States v. Noble, 754 F.2d 1324, 1329 (7th Cir.1985).
We consider this not only a permissible but an inescapable characterization, so that the district court clearly erred in rejecting it. The loan for Neuport Estates was the first in which Quinn and Beede were bribed. Sullivan was in the thick of this fraud; he was not only present at the creation, but instrumental in it. He was instrumental in a further sense. It was he who first introduced Orlak to the bank, and it was he who enabled Orlak to obtain a loan to which Orlak was not entitled; for he recommended the loan to Quinn without disclosing Orlak’s previous bankruptcy, which Sullivan knew about. Sullivan was directly involved in two fraudulent loans; this is an admitted, not a contested, fact.
This participation would not by itself make him liable for all the subsequent acts of his fellow conspirators. There is no evidence that the agreement to obtain the loan for Neuport Estates was part of a broader agreement to siphon money from the bank to Orlak. Cf. United States v. Melchor-Lopez, 627 F.2d 886, 891-92 (9th Cir.1980). Such an agreement undoubtedly was formed at some point in what became a continuing, indeed accelerating, scheme of fraudulent borrowing, but it was formed in the first instance among Orlak and the two bank officers, and the question is whether Sullivan joined the agreement. *384a only question is whether Sullivan joined it. See, e.g., United States v. Melchor-Lopez, supra, 627 F.2d at 891. The evidence leaves no doubt that he did. Cf. United States v. Percival, 756 F.2d 600, 610-11 (7th Cir.1985); United States v. Marks, 816 F.2d 1209, 1212 (7th Cir.1987). Sullivan admits that he knew both that Orlak was to obtain fraudulent loans from the bank and that some of this money was being used to pay interest due on the Neu-port Estates loan. Thus, Sullivan not only launched what turned out to be a continuing scheme for defrauding the bank but also knew that the scheme was continuing, and knowingly benefited from the continuation. He also benefited indirectly from the continuing fraud, for by expanding Orlak’s business it enlarged Orlak’s demands for Sullivan’s legal services. Though we certainly do not suggest that a lawyer by rendering services becomes a partner in his client’s misdeeds, cf. United States v. Anderson, 542 F.2d 428, 436 (7th Cir.1976), membership in the legal profession is not a shield against liability for conduct in excess of professional right or duty, see Celano v. Frederick, 54 Ill.App.2d 393, 399-400, 203 N.E.2d 774, 778 (1964); Wahlgren v. Bausch & Lomb Optical Co., 68 F.2d 660, 664 (7th Cir.1934); Ill.Rev.Stat. ch. 110A, foll. ¶ 774, Rule 7-102 (“Representing a Client Within the Bounds of the Law”).
The facts that we have recited were established by clear and convincing evidence, as required by Illinois courts in civil conspiracy cases. See, e.g., Rosee v. Board of Trade, 43 Ill.App.3d 203, 239, 1 Ill.Dec. 730, 744, 356 N.E.2d 1012, 1036 (1976). The question, also factual, is whether they establish Sullivan’s participation in the broader conspiracy. Misled by a dictum in Lektro-Vend Corp. v. Vendo Co., 660 F.2d 255, 263 (7th Cir.1981), Sullivan makes the unguarded concession, which is incorrect and to which we shall not hold him, that “mixed” questions of fact and law, such as whether the facts add up to conspiracy, are not governed by the clearly-erroneous rule. The current view of this circuit is that, with immaterial exceptions, such “mixed” or “ultimate” questions are facts for purposes of the rule. See Mucha v. King, 792 F.2d 602, 605 (7th Cir.1986); Davenport v. DeRobertis, 844 F.2d 1310, 1311-1312 (7th Cir.1988).
A passage from an early Illinois case on civil conspiracy is apropros: “Abraham L. Day, although ignorant of what was intended, was indifferent and careless as to everything except the compensation promised him and his personal safety, and, by assisting to carry out the conspiracy, became a party to it.” Santee v. Dag, 111 Ill.App. 495, 502 (1903). Sullivan assisted the broader conspiracy by introducing Or-lak to the bank in circumstances redolent of fraud and unethical conduct and by hatching the first full-scale loan fraud, that involving the loan for Neuport Estates, in which Sullivan was a principal. He did not blow the whistle-a clean way of terminating his participation in the fraud. Maybe as an officer of the court, and a stockholder of the victim of the fraud, he could not avoid complicity in the fraud unless he did blow the whistle, though this we needn’t decide. He did not even wash his hands of the business by walking away, but instead watched contentedly as money was diverted from other loans that he knew to be a fraud on the bank of which he was a stockholder to pay interest on his fraudulent loan, the loan to Neuport Estates. The Neuport Estates enterprise was ongoing, moreover; it was not a completed fraud that preceded the broader conspiracy in which Sullivan is alleged to have participated. Sullivan joined the larger conspiracy when, having helped to launch it, he knowingly permitted it to nourish his own particular fraud.
As in Santee v. Day, we have not a case of “mere knowledge, acquiescence, or approval, without intentional participation,” which is “not enough to constitute one a party to a conspiracy,” Aaron v. Dausch, 313 Ill.App. 524, 535, 40 N.E.2d 805, 810 (1942), but instead a case governed by the rule “that all persons who aid or advise in the commission of a fraudulent act by another, or who approve of it after it is done *385for their benefit are liable in the same manner as they would be had they themselves performed the same act,” id. at 530, 40 N.E.2d at 808 (emphasis added). See also Egan v. United States, 137 F.2d 369, 378 (8th Cir.1943). Suppose, to vary the facts immaterially, that a group of people got together and agreed to distribute illegal drugs in Chicago, and each took shares in the enterprise. The enterprise spread to other cities, resulting in increased revenues for the Chicago operation, and a member of the original group, who knew about but did not participate in the expansion, obtained some of the increased revenues by virtue of his share in the Chicago branch. He would be a member of the larger conspiracy, cf. Old Security Life Ins. Co. v. Continental Illinois National Bank & Trust Co., 740 F.2d 1384, 1397 (7th Cir.1984), and this case is the same. We hold that under Illinois law one who participates actively in launching a conspiracy with limited aims, who knows that the aims have been exceeded, and who knowingly obtains direct monetary benefits from the expanded conspiracy, is a participant in that conspiracy as well as in the narrower one from which it grew. He has joined the broader conspiracy and his liability grows with it.
A couple of loose ends, and we are done. The first is, what to do about Sullivan’s set off against Hartford’s claim? Remember that Judge Leighton dismissed Sullivan’s counterclaim, and Sullivan did not file a conditional cross-appeal, as he could have done, in order to preserve the set off in the event that his suit against Ford City Bank failed, as it has. An appellee who wants the judgment of the district court modified, as distinct from merely wanting to defend the district court’s judgment on a ground ignored or rejected by the district judge, must file a cross-appeal; for the application of this rule in the context of a set off see Ayers v. United States, 750 F.2d 449, 457 (5th Cir.1985). Whether the rule is jurisdictional is a debated topic. See 15 Wright, Miller & Cooper, Federal Practice and Procedure § 3904 (1976). Whether it even applies to a case such as the present where the modification is conditional on the appellate court’s rejecting other parts of the district court’s decision is a question on which we can find no authority. But we need not penetrate any deeper into this thicket. Sullivan has not mentioned the set off in his brief in this court, and by his silence has waived any objection to the dismissal of the set off by the district court. Indeed, he asks us to affirm the district court’s decision “in its entirety,” with no ifs, ands, or buts.
Finally, we were astonished to learn at argument that disciplinary proceedings had never been initiated against Sullivan and that he remains to this day a lawyer in good standing in the Illinois bar, continuing to practice real estate law. Yet the trial of this case revealed acknowledged and very serious fraud on his part, for which he might well have been prosecuted criminally although he was not. (The other three conspirators pleaded guilty to federal criminal charges.) And this has been known for many years. We are mailing this opinion to the Illinois Attorney Registration and Disciplinary Commission for such disciplinary action against Sullivan as the Commission may deem appropriate.
The judgment against the Ford City Bank is vacated with directions to dismiss Sullivan’s claim against it for lack of federal jurisdiction. The judgment against the Hartford Accident and Indemnity Company is reversed with directions to enter judgment on liability in its favor and to conduct such further proceedings as may be necessary to determine its damages. In view of our conclusion that Sullivan is liable as a conspirator, Hartford’s claim based on his personal guarantee is moot.
Reversed with Directions.