EASTERBROOK, Circuit Judge.

Janet Lever, an assistant professor of sociology at Northwestern University, was considered for tenure during the 1979-80 academic year, her sixth with the University. Her department unanimously recommended promotion to associate professor, but the Committee on Promotions and Tenure of the College of Arts and Sciences fell one vote short of the two-thirds needed to recommend tenure. On May 5, 1980, Rudolph Weingartner, Dean of the College, wrote to Lever: "I am sorry that I must inform you that I shall not recommend to the Provost that you be promoted to the rank of associate professor. I therefore offer you a final year at Northwestern as assistant professor of sociology, terminating August 31, 1981."

Despite reviews and re-reviews, the Dean adhered to his decision, and on February 12, 1981, Provost Raymond Mack wrote Lever that "I do not feel that I am justified in reversing the decision of the Dean not to recommend your promotion to tenure at Northwestern University." On June 15, 1981, Lever filed with the EEOC a charge that Northwestern discriminated against her because she is female. Illinois is a "deferral" state, so Lever had 300 days from the discriminatory act to file a charge. 42 U.S.C. § 2000e-5(e). The charge is timely if measured from the Provost's letter, but not if measured from the Dean's. For eleven years, the parties have been locked in combat about which letter started the time. After receiving the report of magistrate judge Weisberg, 1991 U.S.Dist. LEXIS 15311, who held a two-day trial as a special master, the district judge concluded that the Dean's decision was definitive, with later steps merely appeals and requests for review. 1991 WL 206066, 1991 U.S.Dist. LEXIS 14353. Under Delaware State College v. Ricks, 449 U.S. 250, 101 S.Ct. 498, 66 L.Ed.2d 431 (1980), and Chardon v. Fernandez, 454 U.S. 6, 102 S.Ct. 28, 70 L.Ed.2d 6 (1981), neither appeals nor a deferred final date of employment postpone the time within which the employee must make a charge.

Time starts to run with "the discriminatory act, not the point at which the consequences of the act become painful." Chardon, 454 U.S. at 8, 102 S.Ct. at 29 (emphasis in original). See also Cada v. Baxter Healthcare Corp., 920 F.2d 446, 449-50 (7th Cir. 1990). When the discriminatory act occurs, no less than whether a given act was discriminatory, is a question of fact. Cf. Pullman-Standard v. Swint, 456 U.S. 273, 102 S.Ct. 1781, 72 L.Ed.2d 66 (1982); Mucha v. King, 792 F.2d 602 (7th Cir. 1986). An appellate court's role in reviewing decisions reached after trial is correspondingly small. The question is not whether we would have agreed with the district judge's decision were we the triers of fact, but whether the district judge's decision is clearly erroneous. Fed.R.Civ.P. 52(a); Anderson v. Bessemer City, 470 U.S. 564, 573-76, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985). Lever presents arguments that might have swayed the trier of fact but did not.

Northwestern's rules shroud the identity of the decisionmaker. Lever (joined by the EEOC and the American Association of University Professors as amici curiae) contends that ambiguity necessitates generosity — that the court should choose the last of the plausible dates, to preserve the claims of persons who believe that they are victims of discrimination. Doubtless courts should consider the effects of uncertainty. Still, a legal rule starting the time on the latest possible date cuts against the grain of Ricks and other decisions we are bound to follow. Congress chose a short period of limitations for employment-discrimination cases, providing employers with a valuable entitlement offsetting the portions of Title VII that assist employees, and courts may not second-guess that decision. No rule of law says that employees win all close cases. Accordingly, we have a factual rather than a legal question, and in an appellate court factual ambiguity ensures affirmance: it means that the district judge's choice is not clearly erroneous. Lever's contention that all debatable questions about timeliness should be resolved in favor of the plaintiff is in the end either a plea to abandon the "clearly erroneous" standard when the district judge has ruled in favor of the employer or an expression of disagreement with Chardon and Ricks.

Did the Dean make a recommendation (as Lever believes) or a decision (as the district court concluded)? To put this differently, did the Dean recommend that Lever not be promoted, or did he decide not to recommend her promotion? His letter of May 5 reads more like a decision, an interpretation fortified by the offer of a terminal contract. That concrete act smacks of finality. You would expect such a contract and a decision to go together. On this understanding, the Dean had the power to conclude the matter adversely to a faculty member, but a final decision in favor of tenure required the approval of the Provost (and ultimately the Board of Trustees).

To be sure, one could believe that Northwestern gave the Dean only a power to recommend in either direction. Page 10 of the handbook for faculty in Northwestern's College of Arts and Sciences provided:

A decision by the Dean to recommend against the granting of tenure leads to an offer of a terminal year. . . . When the Dean sends the Provost a negative recommendation about a candidate, the Dean invites the candidate to a private meeting in order to discuss the reasons for the negative recommendation. If, after that meeting, the candidate writes to the Dean requesting a written statement of those reasons, the Dean writes such a statement. The Dean's recommendations are subject to review by the Provost. A candidate's appeal from a recommendation by the Dean is addressed to the University Faculty Reappointment, Promotion, and Tenure Appeals Panel. The procedure is outlined in the NU Faculty Handbook, pages 13-14.

Page 13 of the faculty handbook said that:

[T]he dean of a school may decide to recommend against reappointment, promotion or tenure for a faculty member. . . . [The faculty member] may, within six months from the date of initial notification, file an appeal with the University Faculty Reappointment, Promotion, and Tenure Appeals Panel. He must at the same time notify the Provost that he has filed such an appeal. Upon receiving this notification, the Provost will defer his own decision on the matter until he has received the recommendation of the University Faculty Reappointment, Promotion, and Tenure Appeals Panel.

One could read these passages as saying that the Dean does nothing but recommend, leaving the decision to the Provost. Yet one also could read them as giving the Dean's "negative recommendation" the status of a final decision (hence the terminal contract), subject only to an "appeal" to a committee and "review" by the Provost. Then the Dean's decision is final in the same sense as a judgment of a district court: it is self-effectuating and stands unless reversed, which can occur only on a request for review.

Testimony by the Dean and the Provost led the district court to conclude that the Dean's decision was indeed self-effectuating, as the award of the terminal contract implies. The Provost testified that he never saw the file unless the faculty member appealed. More, the district court concluded that Lever knew this to be so — which may explain why, when filing a charge with the EEOC, she characterized as the discriminatory decision not the Provost's letter of February 12, 1981, but the Dean's letter of January 15, 1981, reiterating his letter of May 5, 1980.

Lever falls back to an argument that the Dean rescinded his own decision in May 1980, making his letter of January 15, 1981, the "real" initial action. Both Lever and the Sociology Department protested the Dean's action of May 5. After receiving a delegation of disgruntled sociologists, the Dean, a philosopher, wrote to the Chairman of the Sociology Department on May 14, 1980:

Your request [that I reconsider] has led me to spend considerable additional time in reading Miss Lever's promotion dossiser . . . and in reflecting on her case. These considerations have not enabled me to find sufficient ground to counter the recommendations of the CAS Promotion and Tenure Committee, so as now to recommend Miss Lever's promotion to the Provost. These same considerations have, however, led me to conclude with some clarity that the issue of Miss Lever's promotability at Northwestern . . . rests above all on the quality of her unfinished manuscript, Soccer in Brazil — in itself and as a sign of the nature of her future work . . . .

My letter to Miss Lever of May 5 continues to constitute formal notice to the effect that the academic year 1980-81 will be her terminal year at Northwestern. If the final draft of the manuscript in question, Soccer in Brazil, is completed by November 1, 1980 and submitted to the Dean's Office by that date, I will undertake to have a further evaluation made of [it]. On the basis of the information I receive in this way, I will either reverse or confirm my recommendation to the Provost.

Lever submitted the manuscript on time. (It was published in 1984, as Soccer Madness, by the University of Chicago Press.) As he promised, the Dean obtained evaluations of the manuscript. They were not sufficiently favorable, and the Dean adhered to his decision. (The delay made it hard for Lever to secure appointment elsewhere for the 1981-82 academic year, so the Dean extended her terminal contract through August 31, 1982.) The Provost backed up the Dean in February 1981. Lever appealed to the Faculty Reappointment, Promotion, and Tenure Appeals Panel in April 1981. The Panel's adverse decision came in February 1982.

Lever's contention that the Dean's willingness to receive further evidence rescinded his decision of May 5 is inconsistent with the Dean's own description, in the letter of May 14, of the effect of his undertaking: "My letter to Miss Lever of May 5 continues to constitute formal notice to the effect that the academic year 1980-81 will be her terminal year at Northwestern." It may well be that Lever did not want to cry "discrimination!" while there was still an opportunity to obtain a favorable decision from the University. But Professor Ricks had the same legitimate concern, explaining that he did not want to commence adversarial proceedings until the College's Board of Trustees had acted on his request for reconsideration. The Court replied: "It is true that `the filing of a lawsuit might tend to deter efforts at conciliation.' Johnson v. Railway Express Agency, Inc., 421 U.S. [454] at 461 [ 95 S.Ct. 1716 at 1720, 44 L.Ed.2d 295 (1975)]. But this is the `natural effect of the choice Congress has made,' ibid., in explicitly requiring that the limitations period commence with the date of the `alleged unlawful employment practice,' 42 U.S.C. § 2000e-5(c)." Ricks, 449 U.S. at 259 n. 11, 101 S.Ct. at 505 n. 11. If the Dean was holding Lever's sex against her, the unlawful practice occurred on May 5, 1980, when the Dean turned her down for promotion.

An employer's refusal to undo a discriminatory decision is not a fresh act of discrimination. If it were, then an employee could avoid the 300-day limit by filing a series of appeals or fresh requests; Ricks and Chardon hold that adverse decisions on appeals do not re-start the time for filing a charge. See also, e.g., Malhotra v. Cotter Co., 885 F.2d 1305 (7th Cir. 1989). Lever's situation is well removed from Webb v. Indiana National Bank, 931 F.2d 434 (7th Cir. 1991), in which a person applied for jobs with the same employer, was turned down twice, and filed a charge after the second denial. We held that the applicant was entitled to an opportunity to show that the second decision was independent of the first, in which case she could challenge the second (and only the second) decision. An applicant does not have to sue about the first wrong to be entitled to contest a second. Cf. Selan v. Kiley, 969 F.2d 560, 563-67 (9th Cir. 1992). But when the first decision is connected to and implies the second — when, in other words, a single discriminatory decision is taken, communicated, and later enforced despite pleas to relent — the time starts with the initial decision. Webb, 931 F.2d at 437 (distinguishing two ad hoc decisions from a "permanent exclusion" such as "a denial of tenure").

At last we come to Lever's contention that Northwestern is equitably estopped to assert the statute of limitations and that the time is also equitably tolled. Equitable estoppel arises, we held in Cada, 920 F.2d at 450-51, when the employer "takes active steps to prevent the plaintiff from suing in time, as by promising not to plead the statute of limitations." Equitable tolling, by contrast, comes into play "if despite all due diligence [the employee] is unable to obtain vital information bearing on the existence of his claim." Id. at 451. Nothing in Northwestern's avenues of redress kept "vital information" out of Lever's hands; if the Dean's decision is best explained by her sex rather than her scholarship, that was equally apparent on May 5, 1980, and January 15, 1981. An employer need not proclaim its bias as a precondition to the start of the period of limitations. As for equitable estoppel: Northwestern did nothing to dissuade Lever from filing a charge. True, it offered many forums for persuasion, and Lever's desire to pursue these with a minimum of friction likely explains her delay (although Lever could have had her cake and eaten it too: 17 of the 300 days remained on February 12, 1981, when the Provost said no). To treat such opportunities as a source of equitable estoppel would reverse the Court's conclusion in Ricks and Chardon that appeals and requests for reconsideration do not permit delay.

Lever's strategy is to depict the opportunities as snares for the unwary. When the Dean wrote on May 14 that he would take a new look at things if she submitted her manuscript by November 1, Lever became concerned that this might interfere with her appeal to the Faculty Reappointment, Promotion, and Tenure Appeals Panel, which had to be initiated within six months of the Dean's decision. She went to see Provost Mack, himself a sociologist. Mack assured her that he would see to it that the Panel considered any appeal filed promptly after the Dean's decision on reconsideration. Mack was true to his word, insisting that the Panel convene and decide even though Lever's eventual appeal was technically late. What Mack said yields a classic case for equitable tolling — of the time limit to appeal to the Panel, not the limit to file a charge with the EEOC. Mack said nothing about the EEOC and hadn't an inkling that Lever was contemplating such a step. Excessive kindness in providing many and varied opportunities for internal review is not the sort of deception that supports equitable estoppel. It certainly does not compel a conclusion that the district judge abused her discretion.

Northwestern could do its faculty (and the courts) a favor by writing accurate descriptions of its procedures. The handbooks are vague and can be understood to describe steps inconsistent with the process that, according to both Dean and Provost, Northwestern actually follows. Professors can cope with some ambiguity, however, anyone able to obtain a Ph.D. in sociology has read (or written) much worse. The district court found that the Dean had, and exercised, the authority to terminate a career at Northwestern — and that Lever knew it. That conclusion is not clearly erroneous.

AFFIRMED.