UNITED STATES of America, Appellant, v. Stephen J. FLEMMI, Defendant, Appellee, 225 F.3d 78

SELYA Circuit Judge. This appeal arises out of a relentless effort by the Federal Bureau of Investigation (FBI) to infiltrate, and eventually to smash, the New England branch of La Cosa Nostra (LCN). To achieve its goal, the FBI struck a Faustian bargain with two reputed organized crime figures, James ‘Whitey” Bulger and Stephen Flemmi. We concentrate on Flemmi, because he is the protagonist in this appeal. For thirty years, Flemmi — one of Boston’s most notorious gangsters— functioned behind the scenes as an FBI informant. Eventually, however, the government severed the tie.

It later indicted him (along with several others) on multiple counts of racketeering and kindred offenses. Flemmi’s double life began to emerge during the protracted pretrial proceedings that ensued. Once the cat was out of the bag, Flemmi sought to turn his informant status to his advantage. His efforts were rewarded when the district court found that his FBI handlers had promised him use immunity in respect to the fruits of electronic surveillance conducted at three specified locations. See United States v. Salemme, 91 F.Supp.2d 141, 164-65, 329, 400 (D.Mass.1999).

Based on this finding, the court prohibited the government from using that evidence against Flemmi. See id. at 400. This interlocutory appeal followed. In its present posture, the case turns principally on an important question of first impression in this circuit: Do FBI agents, acting independently, have the authority to confer use immunity on a confidential informant? After carefully considering the arguments advanced by the parties, we conclude that they do not.

Thus, the alleged promise of use immunity, even if made, was unauthorized (and, therefore, unenforceable), and the district court erred in suppressing the evidence in question. I. BACKGROUND We recount the background facts, focus; ing on the circumstances relevant to this appeal. We refer the reader who hungers for greater insight into the seamier side of law enforcement to the district court’s more exegetic treatment. See id. at 148-315.

Flemmi long has been a fixture in Boston’s organized crime hierarchy, reputedly engaging in (or overseeing) activities as varied as loan-sharking, extortion, gambling, drug trafficking, and homicide. For much of that period, he and Bulger, as the leaders of the Winter Hill Gang (also known as the Irish Mob), did extensive business with LCN. In the 1960s, Flemmi and Bulger saw a chance to hamstring their competitors and simultaneously ingratiate themselves with the authorities. Accordingly, they began talking to FBI sources about LCN activities.

By 1967, the FBI had designated both men as top-echelon informants — a term defined at the time, according to a knowledgeable witness, as encompassing individuals who “could provide a continuous flow of quality criminal intelligence information regarding the leaders of organized crime.” The data that Flemmi and Bulger provided enabled the FBI to make significant progress in its investigation and prosecution of major LCN figures. Many of the particulars of this uneasy alliance are disputed, and Flemmi often attributes promises and assurances to FBI agents who deny having made them.

For present purposes, we accept the district court’s resolution of these conflicts — but we do so arguendo, without critical examination of the supportability of the court’s findings. Flemmi claims that his initial FBI handler, Agent Paul Rico, promised him protection against prosecution. When a state grand jury indicted Flemmi in 1969 for a ear bombing and a murder, Rico supposedly suggested that he flee. Flemmi took the advice and remained a fugitive for over four years, returning only when Rico assured him that he would be released on bail and that the indictments thereafter would be dismissed.

These predictions proved to be prescient. Once back in town, Flemmi continued to provide information to the FBI. Agent John Connolly became his handler, and Flemmi developed a cordial (some might say cozy) relationship with Connolly and Connolly’s supervisor, John Morris. Flemmi testified that both men afforded him “protection” in various ways: they warned him about electronic surveillances and wiretaps, interceded with the authorities to fend off charges, and stonewalled investigators who were looking into Flem-mi’s activities. A prime example of this protection occurred in 1979.

Jeremiah O’Sullivan, a federal prosecutor who headed the Organized Crime Strike Force in Boston, spearheaded an investigation into a race-fixing scheme. When indictments seemed imminent, Morris and Connolly divulged to O’Sullivan that Flemmi was an FBI informant. Others were indicted, but Flemmi was not. In 1980, the FBI purposed to introduce an electronic listening device into a redoubt on Prince Street, reputed to be LCN’s regional headquarters. Morris and Connolly asked Flemmi and Bulger to visit the site and gather information regarding alarms, locks, and other security devices.

The agents allegedly assured them that, once the bug became operational, nothing on the ensuing tapes would be used against them. The two informants carried out this mission and, in addition, supplied information on which the government relied to establish probable cause for the necessary warrant. See 18 U.S.C. § 2518(1), (3). Although Morris and Connolly warned Flemmi to avoid the Prince Street location, and he did so, the electronic surveillance yielded taped conversations that implicated Flemmi in multifarious criminal activity.

In 1986, the FBI obtained information from Flemmi that established probable cause for electronic surveillance of Vanessa’s Restaurant (where LCN meetings supposedly were taking place). Flemmi claims that Morris and Connolly asked him to provide a diagram of the meeting room. Although he received no express assurances, he asserts that he “reasonably understood” that the same promises applied here as at Prince Street.

Flemmi also asserts that he understood the Prince Street promises to pertain to his 1989 role in securing a “roving bug” that memorialized an LCN induction ceremony at 34 Guild Street in Medford, Massachusetts. In 1990, the FBI “closed” Flemmi as an informant. Not too long thereafter, a federal grand jury began probing the activities of Flemmi, Bulger, and others. In August 1994, Bulger warned Flemmi that ah indictment was imminent, and the two men fled. Bulger remains at large and has not testified in the proceedings below. Conversely, Flemmi returned a few months later and the FBI arrested him.

On January 10, 1995, a federal grand jury handed up an indictment against Flemmi, Bulger, and several other defendants. The indictment charged Flemmi with suborning perjury, Hobbs Act extortion, conspiracy to commit extortion, racketeering, and racketeering conspiracy. See 18 U.S.C. §§ 1512, 1951, 1962(c)-(d). The case was assigned to Judge Wolf. For the next two years, Judge Wolf was unaware that Flemmi had been an FBI informant. The facade crumpled in March of 1997, when the judge came across sealed submissions identifying Flemmi as such.

Judge Wolf spoke privately with Flemmi on April 16, 1997, and soon thereafter Flemmi disclosed his informant status to his counsel and his codefendants. The government, responding to the district court’s order, then confirmed that Flemmi had been an FBI source. In due course, Flemmi filed a series of motions seeking to dismiss the charges against him due to government misconduct or, alternatively, to suppress evidence derived directly or indirectly from statements he had provided to the government. The district court convened an evidentiary hearing and took testimony from January 6 to October 30,1998.

On September 15, 1999, Judge Wolf issued a 661-page opinion. He concluded, inter alia, that the FBI had promised Flemmi both use immunity and derivative use immunity anent the conversations intercepted at Prince Street, Vanessa’s Restaurant, and Guild Street. See Salemme, 91 F.Supp.2d at 164-65, 329, 400. In reaching this conclusion, the judge found that Morris and Connolly had expressly promised Flemmi that none of the evidence overheard at Prince Street would be used against him “directly or indirectly.” Id. at 400.

Based on this assurance, the judge reasoned that, although Morris and Connolly had not made any equivalent promises with respect to Vanessa’s Restaurant and Guild Street, Flemmi “had an agreement implied in fact from the promise concerning 98 Prince Street and the conduct of the government that there would be no direct or indirect use against him” of the evidence intercepted at those locations. Id. at 329. Having determined that promises were made, Judge Wolf ruled that these promises were enforceable.

He grounded this ruling in the thesis that FBI agents have authority to promise immunity as part of their power “to investigate federal crime” and “the[ir] power to develop and utilize informants.” Id. at 331. He also suggested — but did not adjudicate — two other bases on which Flemmi might prevail even if the agents’ assurances were unauthorized. First, the judge asserted that, in some circumstances, due process might require the government to perform a promise by an agent who exceeded his actual authority. See id. at 346-47.

Second, he mused that if the agents’ promises were unauthorized, Flemmi’s statements might be deemed involuntary. See id. at 347-50. Finally, Judge Wolf proposed holding further hearings to determine whether evidence drawn from the intercepted conversations had been presented to the grand jury and, if so, whether the indictment against Flemmi should be dismissed. See id. at 400. The government sought refuge in this court without waiting for the second shoe to drop. II. APPELLATE JURISDICTION Deeming the government’s appeal premature, Flemmi sought to dismiss it for want of appellate jurisdiction.

On February 16, 2000, we denied his motion in an unpublished order. Before addressing the merits of the district court’s suppression ruling, we think it is appropriate to explicate our rationale for accepting jurisdiction.

The government premises appellate jurisdiction on 18 U.S.C. § 3731, which provides in pertinent part: An appeal by the United States shall lie to a court of appeals from a decision or order of a district court suppressing or excluding evidence or requiring the return of seized property in a criminal proceeding, not made after the defendant has been put in jeopardy and before the verdict or finding on an indictment or information, if the United States attorney certifies to the district court that the appeal is not taken for purpose of delay and that the evidence is a substantial proof of a fact material in the proceeding....

The provisions of this

section shall be liberally construed to effectuate its purposes. In this instance, however, some aspects of the pending motions remain to be decided. See Salemme, 91 F.Supp.2d at 400. The question, then, is whether the district court’s ruling constitutes “a decision or order ... suppressing or excluding evidence” within the meaning of

section 3731. We believe that it does. To begin, the fact that the district court has not yet transformed its ruling into a separate order does not place the ruling beyond the reach of

section 3731. The district court announced with unmistakable clarity that all direct and derivative evidence stemming from the three electronic surveillances would be suppressed. See Salemme, 91 F.Supp.2d at 338-39. In light of

section 3731’s disjunctive reference to a “decision or order,” such an announcement plainly triggers the statute’s prophylaxis. Second, Judge Wolfs intimation that he did not intend his ruling to be immediately appealable, see id. at 400-01 (“[T]he court will hold the hearings necessary to determine whether this case must be dismissed and, if not, the scope of the evidence to be excluded at trial before entering any Order that may be appealable by the United States .... ”), does not defeat our jurisdiction.

After all, a trial court’s characterization of its own ruling neither determines nor controls the classification of that ruling in respect to appellate jurisdiction. See United States v. Scott, 437 U.S. 82, 96, 98 S.Ct. 2187, 57 L.Ed.2d 65 (1978); United States v. Jorn, 400 U.S. 470, 478 n. 7, 91 S.Ct. 547, 27 L.Ed.2d 543 (1971); United States v. Kennings, 861 F.2d 381, 384 n. 5 (3d Cir.1988); United States v. Schwartz, 785 F.2d 673, 677 (9th Cir.1986); United States v. Kehoe, 516 F.2d 78, 82 (5th Cir.1975).

The pertinent question is not whether the trial court intended its decision to be immediately appealable, but, rather, whether that decision, viewed objectively, satisfies the criteria that Congress established in

section 3731. Answering this question requires us to apply

section 3731 in a pragmatic, common-sense manner. Under such an approach, “pretrial orders that have the practical effect of excluding material evidence at trial are appealable under

section 3731, regardless of nomenclature.” United States v. Brooks, 145 F.3d 446, 454 (1st Cir.1998); accord United States v. Parks, 100 F.3d 1300, 1304-05 (7th Cir.1996); United States v. Drogoul, 1 F.3d 1546, 1551 n. 13 (11th Cir.1993); In re Grand Jury Empanelled, 597 F.2d 851, 855-56 (3d Cir.1979). Moreover, the language of

section 3731 makes pellucid Congress’s desire that, as opposed to other jurisdiction-conferring statutes, this statute should be construed liberally. These guideposts point us to a finding of jurisdiction. The practical effect of the ruling sub judice is to suppress a body of evidence. The fact that other, closely related issues remain to be decided in the district court neither distorts that effect nor divests the ruling of its appealable character. See United States v. Ienco, 126 F.3d 1016, 1017 (7th Cir.1997); In re Grand Jury Subpoena, 646 F.2d 963, 967-68 (5th Cir.1981). Thus, consistent with the generous compass of

section 3731, we conclude that we have jurisdiction to hear and determine this interlocutory appeal. III. THE MERITS We find ourselves able to resolve this appeal by focusing primarily on one issue: assuming, arguendo, that FBI agents actually promised Flemmi use and derivative use immunity, were they authorized to make such a promise? This is a pure question of law and, as such, engenders de novo review. See United States v. Igbonwa, 120 F.3d 437, 442 (3d Cir.1997). We start with first principles. A formal grant of use immunity, conferred pursuant to 18 U.S.C. § 6002, must be honored. See Kastigar v.

United States, 406 U.S. 441, 460, 92 S.Ct. 1653, 32 L.Ed.2d 212 (1972); United States v. McLaughlin, 957 F.2d 12, 16 n. 4 (1st Cir.1992). A prosecutor’s promise, accepted by a defendant in consideration of a change of plea, likewise must be honored. See Santobello v. New York, 404 U.S. 257, 262, 92 S.Ct. 495, 30 L.Ed.2d 427 (1971); United States v. Alegría, 192 F.3d 179, 182 (1st Cir.1999). Flemmi’s case differs from these prototypes in that he did not receive immunity by way of either a statutory grant or a plea bargain.

To bridge this gap, he argues that assurances of immunity made incident to cooperation agreements should be honored because such agreements are analogous to plea agreements. There is, of course, one basic similarity (though not necessarily the similarity that Flemmi wishes us to emphasize): a defendant who seeks specifically to enforce a promise, whether contained in a plea agreement or a freestanding cooperation agreement, must show both that the prom-isor had actual authority to make the particular promise and that he (the defendant) detrimentally relied on it. See San Pedro v.

United States, 79 F.3d 1065, 1068 (11th Cir.1996); Thomas v. INS, 35 F.3d 1332, 1337 (9th Cir.1994); United States v. Streebing, 987 F.2d 368, 372 (6th Cir.1993). If either part of this showing fails, the promise is unenforceable. See San Pedro, 79 F.3d at 1068; Streebing, 987 F.2d at 372. We turn, then, to the question of authority. As a general rule, doctrines such as estoppel and apparent authority are not available to bind the federal sovereign. See Igbonwa, 120 F.3d at 443; United States v. Holmquist, 36 F.3d 154, 161 n. 7 (1st Cir.1994).

Thus, the inquiry into the scope of the FBI agents’ authority here must be framed in terms of actual authority. See Federal Crop Ins. Corp. v. Merrill, 332 U.S. 380, 384, 68 S.Ct. 1, 92 L.Ed. 10 (1947); Dresser Indus. v. United States, 596 F.2d 1231, 1236 (5th Cir.1979). Actual authority may be conferred either expressly or by necessary implication. See Carr v. Runyan, 89 F.3d 327, 331 (7th Cir.1996); United States v. Greene, 995 F.2d 793, 800 (8th Cir.1993); see generally Restatement (Second) of Agency § 7 cmt. c, at 29-30, § 8 cmt. a, at 30 (1958).

The possibility of express authority need not occupy us for long. Virtually by definition, a government agent possesses express authority to bind the government if — and only if — the Constitution, a federal statute, or a duly promulgated regulation grants such authority in clear and unequivocal terms. Flemmi points to no constitutional provision, federal statute, or binding regulation that cedes to the FBI explicit authority to promise use immunity to an informant, and we are aware of none. Thus, the lens of inquiry narrows to a consideration of implied authority.

In a civil case, generally speaking, an agent’s authority to perform

an act and thereby bind his principal includes the authority to do other things “which are incidental to [the act], usually accompany it, or are reasonably necessary to accomplish it.” Restatement (Second) of Agency § 35, at 123; accord ICC v. Holmes Transp., 983 F.2d 1122, 1129 n. 10 (1st Cir.1993). Without deciding the matter, we will assume, arguendo, the core premise of Flemmi’s argument: that the same principles apply equally to the federal government in a criminal case when the issue is a constraint on the government’s powers to prosecute. Thus, in the case of a federal agent, authority to do

an act may be implied when that act is integral to the tasks assigned to him or otherwise necessary for the due accomplishment of those tasks. See H. Landau & Co. v. United States, 886 F.2d 322, 324 (Fed.Cir.1989). This black-letter statement of the controlling rule is uncontroversial, but how the rule operates in practice is far less cut and dried. This uncertainty frequently manifests itself when the authority to do a particular task can be characterized as incidental to the duties assigned to two separate government agencies.

In that type of situation, the issue presented often involves whether one agency can exercise its “incidental” power to preempt the other and thereby bind the government. A classic example involves the power to withhold deportation. The Eighth Circuit has taken a relatively expansive view of the “incidental to duty” approach in that context.

Reasoning that the “authority to prosecute implies the power to make plea agreements incidental to the prosecution,” that court has held that an Assistant United States Attorney possesses actual authority to assent to a plea agreement condition anent deportation and thereby bind the Immigration and Naturalization Service (INS). See Margalli-Olvera v. INS, 43 F.3d 345, 353-54 (8th Cir.1995) (citation and internal quotation marks omitted). The Ninth Circuit may share this view. See Thomas, 35 F.3d at 1339. But cf. United States v.

Cordova-Perez, 65 F.3d 1552, 1554 (9th Cir.1995) (stating that INS agent may not promise there will be no federal prosecution unless United States Attorney agrees). Several other courts have found this approach overbroad and have declared that, in such circumstances, a specific source of authority to do a particular task trumps a more general source of authority. The Third Circuit, using this approach, has held that an Assistant United States Attorney lacks unilateral authority to promise a criminal defendant freedom from deportation as part of a plea agreement. See Igbonwa, 120 F.3d at 443-44.

The Eleventh Circuit shares this view. See San Pedro, 79 F.3d at 1070-72. This court has not yet had occasion to choose between these modes of analysis, and we see no need to make that choice today. As we shall explain, under either approach (and continuing to assume the validity of Flemmi’s core premise, see supra), FBI agents lack the authority to promise an informant use immunity.

The United States Attorney’s implied authority to offer assurances of immunity is closely connected with, and arises out of, Congress’s express grant of authority to prosecute, see 28 U.S.C. § 547 (providing that, with exceptions not relevant here, “each United States attorney, within his district, shall ... prosecute for all offenses against the United States”), and to extend formal use immunity, see 18 U.S.C. § 6002. This fit is much tighter than that between the FBI’s authority to investigate and its supposed ability to promise immunity.

After all, use immunity is at bottom immunity from the use of evidence in a criminal prosecution, see Black’s Law Dictionary 751 (6th ed.1990), and plea bargaining is part of the warp and woof of the prosecu-torial process. The idea that the authority to promise use immunity is linked to the FBI’s responsibility to develop informants (and, thus, more efficiently investigate crimes) requires a much greater leap of faith.

The test is not whether such a power might from time to time prove advantageous, but, rather, whether such a power usually accompanies, is integral to, or is reasonably necessary for the due performance of the task. See Thomas, 35 F.3d at 1340; Holmes Transp., 983 F.2d at 1129 n. 10; see also Restatement (Second) of Agency § 35, at 123. Here, there is no sound reason to believe, either from empirical data or practical experience, that this standard has been met.

The record contains no evidence that the FBI regularly promises immunity, or even that the Bureau regards the ability to promise immunity as a part of its armamentarium. Given the many other avenues that exist for the development of informants (e.g., money, promises of good words at sentencing), we view the connection between a promise of immunity and the FBI’s duty to investigate crimes as far too attenuated to pass the “incidental to duty” screen. Were the law otherwise, the concept of “incidental to duty” would stretch so interminably as to become entirely unworkable as a limiting principle.

Leaving the tenuousness of the connection to one side, the implication of authority also would fail under a structural, “specific source of authority” analysis. The FBI is part of the Department of Justice, see 28 U.S.C. § 531, and its officials are appointed by the Attorney General, see id. §§ 532-533. By statute, the Attorney General may conduct, or direct another officer of the Department of Justice to conduct, hny kind of civil or criminal proceeding which United States Attorneys are authorized by law to conduct. See id. § 515(a).

Exercising her power of appointment, see id. § 533, she has delegated investigation of possible violations of the law to the FBI, see 28 C.F.R. § 0.85. United States Attorneys also are part of the Department of Justice, but they are appointed directly by the President, with the advice and consent of the Senate. See 28 U.S.C. § 541. Their authority to prosecute, which dates back almost to the birth of the Republic, see Judiciary Act of 1789, ch. 20, § 35, 1 Stat. 73, 92 (1789), derives directly from Congress, see 28 U.S.C. § 547.

Importantly, a United States Attorney’s decision to prosecute (or, conversely, to forbear) is largely unreviewable by the courts. See Dresser Indus., 596 F.2d at 1237. Viewed against this backdrop, it is clear that the FBI’s performance of its investigatory duties is meant to complement, not curb, the United States Attorneys’ performance of the prosecutorial function. See Weinberg v. United States, 126 F.2d 1004, 1008 (2d Cir.1942); Cooper v. O’Connor, 99 F.2d 135, 139 (D.C.Cir.1938). This circumstance has decretory significance in a “specific source of authority” analysis.

A United States Attorney’s authority to grant use immunity is implied from her statutory authority to make decisions anent prosecution pursuant to 28 U.S.C. § 547. A claim that FBI agents possess similar authority also rests on an implication — but one that is several steps removed from the original statutory grant. The former implication lies much closer to the specific source of authority — Congress-—than the latter.

The short of it is that the power to prosecute plainly includes the power not to prosecute (and, thus, the power to grant use immunity), whereas the power to investigate does not necessarily encompass (or even reasonably imply) the power to grant use immunity. Because the more specific source of authority must prevail when such a clash occurs, United States Attorneys’ power to make promises of immunity trumps the FBI’s more generalized claim.

The end result brings coherence to the law: just as applications for formal grants of use immunity are the exclusive prerogative of United States Attorneys under 18 U.S.C. § 6002, see United States v. Graham, 548 F.2d 1302, 1315 (8th Cir.1977), so too are informal grants of use immunity under current conditions.

Not surprisingly, the case law supports this result and, at the same time, contradicts the district court’s premise that officials having lesser authority over prosecutions than United States Attorneys, such as FBI agents, may bind the United States either to dismiss an indictment or to refrain from prosecution. See, e.g., Cordova-Perez, 65 F.3d at 1554 (stating that INS agent who made a “no prosecution” promise could not bind the United States); United States v.

Fuzer, 18 F.3d 517, 520-21 (7th Cir.1994) (holding that ATF agents lacked authority to promise that defendant would not be prosecuted); Streebing, 987 F.2d at 373 (finding that FBI agent “lacked any actual or apparent authority to make the alleged promise not to prosecute”); United States v.

Kettering, 861 F.2d 675, 676 (11th Cir.1988) (holding that a DEA agent lacked authority to guarantee immunity); In re Corrugated Container Antitrust Litig., 662 F.2d 875, 888 (D.C.Cir.1981) (holding there is “no authority for ruling that oral promises of immunity by an investigator [FBI agent], not in accord with statutory requirements, bind all federal ... prosecutors”); United States v.

Hudson, 609 F.2d 1326, 1329 (9th Cir.1979) (holding that Secret Service Agent’s promise to drop charges did not bind the United States); Dresser Indus., 596 F.2d at 1237 (holding that “the SEC’s agents lacked actual authority to contractually limit the prosecutorial function of the Department of Justice, [so] any such agreement ... would be unenforceable”).

This line of cases makes intuitive sense: If the rule were otherwise, a minor government functionary hidden in the recesses of an obscure department would have the power to prevent the prosecution of a most heinous criminal simply by promising immunity in return for the performance of some act which might benefit his department. Such a result could not be countenanced. Dresser Indus., 596 F.2d at 1236-37.

Thus, the clear weight of authority buttresses the government’s position that a promise of use immunity made independently by an FBI agent exceeds the scope of his actual authority (and is, therefore, unenforceable). Notwithstanding this seemingly impregnable wall of authority, Flemmi, ably represented, advances several other arguments in support of the proposition that FBI agents may promise an informant use immunity. For the sake of completeness, we deal with the most cogent of these arguments.

Flemmi first suggests that the authority to make promises of immunity is inherent in the confidential nature of the agent-informant relationship and the concomitant duty to protect the informant’s identity. Because the FBI has the prerogative to conceal an informant’s identity from prosecutors, this thesis runs, it need not consult the prosecutor with respect to an assurance of use immunity. This suggestion mixes plums and pomegranates. A promise of confidentiality and a promise of use immunity are separate and distinct assurances.

Simply because an FBI agent appropriately may keep an informant’s identity to himself does not by some mysterious alchemy imbue the agent with the (otherwise nonexistent) power to promise use immunity. Next, Flemmi calls attention to the so-called “Levi Guidelines,” named after former Attorney General Edward Levi. He asseverates that these guidelines, which were part of the FBI Agents’ Manual (the Manual) when Morris and Connolly allegedly guaranteed Flemmi use immunity, implicitly allowed FBI agents to promise immunity to informants. Flem-mi’s asseveration runs into difficulty on two levels.

As a general matter, the Manual merely provides guidance to FBI agents and does not have the force of law. See Manual § 137-17(N) (effective Jan. 12, 1981) (“These guidelines on the use of informants and confidential sources are set forth solely for the purpose of internal Department of Justice guidance. They are not intended to, do not, and may not be relied upon to create any rights, substantive or procedural, enforceable at law by any party in any matter, civil or criminal... .”). Consequently, the Manual cannot serve as a source of delegated authority to FBI agents to promise use immunity to informants.

See Kugel v. United States, 947 F.2d 1504, 1507 (D.C.Cir.1991); cf. United States v. Craveiro, 907 F.2d 260, 264 (1st Cir.1990) (explaining that “the internal guidelines of a federal agency, ... not mandated by statute or the Constitution, do not confer substantive rights on any party”). Moving from the general to the particular, Flemmi’s position also lacks traction because he reads the Manual with much too self-interested an eye. At the relevant times, the Manual did not authorize FBI agents to promise immunity to informants, but prohibited them from doing so.

From 1955 to 1976, the Manual stated: Care must be exercised in attempting to persuade individuals to act as informants to avoid any allegations of undue influence. An individual who is in custody and who offers to furnish information generally does so in the hope that he will receive some consideration in return. Bureau agents cannot promise any immunity or any reduction in sentence to a criminal who famishes infor- motion and they must not put themselves in a situation where they might subsequently be accused of having done so.

Manual § 108(D)(4) (effective Nov. 29, 1955 through May 13,1976) (emphasis supplied). The Manual was then amended to provide that “[a]gents should not exercise undue influence in developing informants including promising immunity or reduction of sentence to criminals who furnish information.” Id. §§ 108 pt. 1(C)(6) (effective Jan. 12, 1977); 137-3(6) (effective Jan. 31, 1978) (same language recodified). That caveat was in effect in 1980 when the promise to Flemmi ostensibly occurred..

It is, therefore, unsurprising that numerous agents testified below that they had no power to confer immunity, and that no agent testified to the contrary. In a desperate effort to escape the obvious implications of this language, Flemmi labors to persuade us that it applies only to informants in custody. We are unconvinced. The relevant edition of the Manual defines an “informant” as “any person who furnishes information in a continuing and confidential relationship concerning matters within areas of FBI responsibility.” Id. § 137-1 (effective Jan. 31, 1978).

Given the breadth of this language, we reject the distinction that Flemmi proposes. In a somewhat related vein, Flemmi adverts to a portion of the Manual that permits FBI agents to authorize informants to engage in criminal behavior: The FBI shall instruct all informants it uses in domestic security, organized crime, and other criminal investigations that in carrying out their assignments they shall not 4. participate in criminal activities of persons under investigation, except insofar as the FBI determines that such participation is necessary to obtain information needed for purposes of federal prosecution.

Id. § 108 pt. IV(B)(4) (effective Apr. 2, 1979). Since this language gives the FBI discretion to determine whether participation in criminal activity is necessary to gather essential information, Flemmi theorizes, it is logical to assume that this discretion includes the power to decide that an informant will be permitted to engage in such criminal activity without fear of prosecution. This theory represents a triumph of hope over reason. The authority to promise use immunity does not flow automatically from the fact that the FBI has some limited authority to authorize participation in criminal activity.

The Attorney General made this point painfully clear in 1981 (shortly after the introduction of the electronic bug at Prince Street but before either of the other two interceptions), when he inserted the following statement in the Manual: “Each such [informant] shall be advised that his relationship with the FBI will not protect him from arrest or prosecution for any violation of Federal, State, or local law, except where the FBI has determined pursuant to these guidelines that his association in specific activity, which would otherwise be criminal, is justified for law enforcement.” Id. § 137-17(E)(1) (effective Jan. 12, 1981) (emphasis supplied).

Where “extraordinary” criminal activities are concerned — i.e., activities which involve, inter aha, “a significant risk of violence ... or severe financial loss to a victim,” id. § 137-17(F)(2) — they can be authorized only with the written approval of the United States Attorney, see id. § 137-17(F)(2) to (3). Here, Flemmi’s criminal activities assuredly were extraordinary in the requisite sense, yet no approval from the United States Attorney was either sought or obtained. Thus, these activities cannot qualify for the narrow exception limned in

section 137-17(E)(1). Flemmi next invokes the so-called Top Echelon Informant Directive, contained in the Manual from 1977 through 1980. He quotes the following sentence: “The sue-cess of the Top Echelon informant program depends on a dynamic and imaginative approach in developing quality sources who can help the Bureau in meeting its investigatory responsibilities.” Id. § 108 pt. III(B) (effective Jan. 12, 1977); id. § 137-12(2) (effective Apr. 12, 1979).

Flemmi persuaded the district court that this language reasonably could have been interpreted by Morris and Connolly as authorizing an informal promise of use immunity. See Salemme, 91 F.Supp.2d at 193, 335. We are not so sanguine. Reading the Manual as a whole, the quoted language is far too oblique for us to accompany the lower court on its interpretive journey.

This is particularly so in light of other language that appears elsewhere in the same section: “Agents should not exercise undue influence in developing informants including promising immunity or reduction of sentence to criminals who furnish information.” Manual § 137-3(6) (effective Jan. 31, 1978). In our judgment, the latter, more explicit language vitiates the inference that Flemmi seeks to have us extract from the former, more generalized language. Cf. Edmond v.

United States, 520 U.S. 651, 657, 117 S.Ct. 1573, 137 L.Ed.2d 917 (1997) (explaining that “where a specific [statutory] provision conflicts with a general one, the specific governs”). Flemmi next claims that Jeremiah O’Sullivan, the Strike Force prosecutor, ratified the promise of use immunity. This ratification occurred, Flemmi says, when Morris and Connolly met with O’Sullivan in 1979, told him that Flemmi was a clandestine informant, and requested that Flemmi be excluded from the race-fix indictment.

In principle, the government may be bound by an unauthorized agreement if a properly authorized official subsequently ratifies it. See, e.g., Harbert/Lummus Agrifuels Projects v. United States, 142 F.3d 1429, 1433 (Fed.Cir.1998); Carr, 89 F.3d at 332; Howard v. United States, 31 Fed.Cl. 297, 314 (Fed.Cl.1994). But no express ratification transpired here, and ratification can be implied only when the ratifying official knows of the agreement, fails to repudiate it in a timely manner, and accepts benefits under it. See United States v.

Beebe, 180 U.S. 343, 354, 21 S.Ct. 371, 45 L.Ed. 563 (1901); Carr, 89 F.3d at 332; Ouimette v. E.F. Hutton & Co., 740 F.2d 72, 78 (1st Cir.1984); see also Restatement (Second) of Agency § 43, at 134. In this case, O’Sullivan was told of Flemmi’s status qua informant and agreed not to charge him in the race-fix case. There is, however, no evidence that O’Sullivan knew of the promise of use immunity. Nor could he, inasmuch as the conversation with him took place before the Prince Street affair bubbled up and the agents made the alleged promise of use immunity.

This chronology is fatal to Flemmi’s ratification argument. See Inn Foods, Inc. v. Equitable Co-op. Bank, 45 F.3d 594, 597 (1st Cir.1995); Computel, Inc. v. Emery Air Freight Corp., 919 F.2d 678, 683 (11th Cir.1990); see also Irving Tanning Co. v. Shir, 295 Mass. 380, 3 N.E.2d 841, 843 (1936). Finally, Flemmi places particularly staunch reliance on United States v. Rodman, 519 F.2d 1058 (1st Cir.1975) (per curiam), a case that is easily distinguished.

In Rodman, we affirmed dismissal of an indictment where the defendant was induced to give statements to the Securities and Exchange Commission (SEC) by a promise that the SEC would strongly recommend to the United States Attorney that the defendant not be prosecuted, and the SEC not only failed to make the recommendation but was actively contemplating the preparation of a criminal reference report implicating the defendant. See id. at 1059. There is no hint that the unfulfilled promise in Rodman — to recommend to the United States Attorney that no prosecution be undertaken — was beyond the promisor’s authority.

That places Rodman at a considerable remove from the case at bar. Flemmi’s other arguments on the authority issue are bootless, and we reject them out of hand. Accordingly, we hold that FBI agents lack authority to tender a binding promise of use immunity to an informant. Absent that authority, any promise made to Flemmi was unenforceable. See Dresser Indus., 596 F.2d at 1237. Two other points demand our attention.

Flemmi argues that the lower court left open a pair of issues: whether the fruits. of the electronic surveillance should be excluded on a theory of detrimental reliance, see Salemme, 91 F.Supp.2d at 846-47, and whether the excluded evidence could be considered as having been obtained involuntarily, see id. at 347-50. He beseeches us to remand the case for further findings on those issues, should we decline to enforce the agents’ promise.

Federal appellate courts, however, are free to consider issues that were not decided below as long as the parties have had an adequate opportunity to present their proof and the record on appeal limns those issues with sufficient clarity. See, e.g., Katt v. Dykhouse, 983 F.2d 690, 695 (6th Cir.1992); Societe Des Produits Nestle v. Casa Helvetia, Inc., 982 F.2d 633, 642 (1st Cir.1992); United States v. Mora, 821 F.2d 860, 869 (1st Cir.1987). Such a course becomes compelling where, as here, the resolution of such issues will materially advance the administration of justice.

The first of Flemmi’s concerns is a non-starter. For a promise to be binding upon the government, two conditions must be met: the agent who makes the promise must have actual authority to do so, and the party who seeks to enforce the promise must have detrimentally relied on it. See San Pedro, 79 F.3d at 1068; Streebing, 987 F.2d at 372. If either condition is lacking, then the promise is unenforceable. Because Morris and Connolly lacked authority, see supra, it is immaterial whether Flemmi relied on the supposed promise. The second of Flemmi’s concerns is equally unavailing.

In this context, the question of voluntariness is a legal question, subject to de novo review. See United States v. Palmer, 203 F.3d 55, 60 (1st Cir.2000), cert. denied, — U.S. -, 120 S.Ct. 2756, 147 L.Ed.2d 1018 (2000). We have the benefit of a massive record compiled in pretrial proceedings that already have taken over five years. We are thus equipped to decide it.

Flemmi argues that even if the FBI agents lacked authority to promise him use immunity, their unauthorized promises induced his statements to the FBI in connection with the three intercepts, thereby rendering the statements involuntary and the fruits of the surveillance inadmissible. Long ago, this argument may have had some bite. At common law, statements prompted by promises not to prosecute sometimes were excluded as involuntary. See 2 Wayne R. LaFave et al., Criminal Procedure § 6.2, at 442 (1984).

In more recent times, however, “the Supreme Court has confined the voluntariness concept by holding that only [statements] procured by coercive official tactics should be excluded as involuntary.” United States v. Byram, 145 F.3d 405, 407 (1st Cir.1998) (citing Colorado v. Connelly, 479 U.S. 157, 167, 107 S.Ct. 515, 93 L.Ed.2d 473 (1986)). The mere fact that an unfulfilled promise was made in exchange for a person’s statement does not constitute coercion, rendering the statement involuntary or its fruits inadmissible. See United States v. Walton, 10 F.3d 1024, 1028 (3d Cir.1993).

In this instance, the record will not support Flemmi’s effort to treat as coercion the assurances that he would be held harmless from prosecution. The nisi pri-us roll reveals no threats of retaliation, see, e.g, Lynumn v. Illinois, 372 U.S. 528, 534, 83 S.Ct. 917, 9 L.Ed.2d 922 (1963), or of violence, see, e.g., United States v. Rosario-Diaz, 202 F.3d 54, 69-70 (1st Cir.2000), and no evidence of consciously misleading conduct on the part of the FBI agents, see, e.g., United States v. Swint, 15 F.3d 286, 290-91 (3d Cir.1994).

Flemmi was neither incarcerated nor under investigation at the time of the asserted promise, and he enjoyed a friendly — even social-relationship with Morris and Connolly. Weighting the totality of the circumstances, as we must, see Walton, 10 F.3d at 1028, we conclude as a matter of law that Flemmi’s statements with respect to the three intercepts were voluntary. IV. CONCLUSION We need go no further. We have considered all of Flemmi’s arguments and find them wanting.

We therefore conclude that the ruling suppressing evidence stemming from the electronic surveillance of Prince Street, Vanessa’s Restaurant, and Guild Street is insupportable and must be set aside. This holding will, of course, obviate any need for the district court to conduct further hearings as to whether evidence derived from the intercepted conversations was presented to the grand jury and whether the indictment should be dismissed on that basis. Reversed and remanded for further proceedings not inconsistent with this opinion. .

In a retrospective conversation that Flemmi says transpired in April 1985, Morris and Connolly supposedly confirmed that the knowledge gleaned during the Prince Street surveillance would not be used against him. Morris also allegedly stated: "You can do anything you want as long as you don't 'clip' anyone.” .

Although Flemmi dances around the point, his argument necessarily relies on the following progression: Congress ceded to the Attorney General authority to appoint officials to detect crimes against the United States, see 28 U.S.C. § 533; the Attorney General delegated this power to the FBI, see 28 C.F.R. § 0.85; the FBI’s authority to develop informants is implied from the investigatory power; and the purported right to promise use immunity is implied from the right to develop informants. .

The district court sought to distinguish cases of this stripe on the ground that they involved individuals who, at the time of the promise, were criminal defendants or targets of criminal investigations, not merely informants. See Salemme, 91 F.Supp.2d at 335-37. We see no principled basis for such a distinction. . A narrow exception to this rule exists when the government’s noncompliance with an unauthorized promise would render a prosecution fundamentally unfair. See, e.g., Streebing, 987 F.2d at 373; United States v. Williams, 780 F.2d 802, 803 (9th Cir.1985); United States v.

Costello, 750 F.2d 553, 556 (7th Cir.1984). This case lies well outside the compass of that seldom-seen exception. . Of course, trickery can sink to the level of coercion, but this is a relatively rare phenomenon. Here, there is no evidence that Morris or Connolly intended to mislead Flemmi or tried to dupe him. At any rate, "confessions procured by deceits have been held voluntary in a number of situations.” Byram, 145 F.3d at 408 (citing cases). Even if the government could be charged with deceit on this record— a matter on which we take no view — we would still regard Flemmi’s statements as voluntary.