United States v. Hubbell, Webster L.

United States Court of Appeals FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued October 21, 1998 Decided January 26, 1999

No. 98-3080 United States of America,

Appellant v. Webster L. Hubbell, Suzanna W. Hubbell, Michael C. Schaufele and Charles C. Owen

Appellees Appeal from the United States District Court for the District of Columbia

(No. 98cr00151-01) Kenneth W. Starr, Independent Counsel, argued the cause for appellant. With him on the briefs were David G. Barger, Joseph M. Ditkoff, and Darrell M. Joseph, Associate Indepen- dent Counsels. Stephen J. Binhak, Associate Independent Counsel, entered an appearance.

John W. Nields, Jr. argued the cause for appellees Web- ster L. Hubbell and Suzanna W. Hubbell. K. Chris Todd argued the cause for appellee Michael C. Schaufele. With them on the brief were Laura S. Shores, Wan J. Kim and Drake Mann, counsel for appellee Charles C. Owen.

J. Douglas Wilson, Attorney, U.S. Department of Justice, argued the cause for amicus curiae United States acting through the Attorney General.

Before: Wald, Williams and Tatel, Circuit Judges. Opinion for the Court filed Per Curiam.* Separate opinion concurring in Part I filed by Circuit Judge Wald.

Separate opinion dissenting from Part I filed by Circuit Judge Tatel.

Separate opinion dissenting from Part II filed by Circuit Judge Williams.

Per Curiam: All defendants--Webster L. and Suzanna W. Hubbell, Michael C. Schaufele, and Charles C. Owen--moved in the district court to dismiss an indictment charging tax evasion and related crimes on the ground that the indictment was beyond the prosecutorial jurisdiction of Independent Counsel Kenneth W. Starr. In addition, Webster Hubbell moved for dismissal on the theory that prosecution necessari- ly would depend on evidence produced under compulsion and used in violation of the Fifth Amendment and the immunity granted him under 18 U.S.C. s 6002. The court granted both motions. 11 F. Supp. 2d 25 (D.D.C. 1998). We reverse both decisions and remand for proceedings consistent with this opinion.

I. Jurisdiction On August 5, 1994 this court's Special Division for the Purpose of Appointing Independent Counsels ("Special Divi- sion"), upon request from the Attorney General, appointed

Viewed as an agency, the Special Division appears to act quite like one glossing its own regulation--a situation in which we usually grant substantial deference. See Paralyzed Veterans of America v. D.C. Arena L.P., 117 F.3d 579, 584 (D.C. Cir. 1997) (deference to agency interpreting its own regulations at least equal to deference under Chevron). De- fendants and the Department of Justice as amicus would have us reject deference on the ground that the Division operates without procedures for critique and comment by outsiders, especially by adversely affected parties. But the cases fre- quently find deference in such safeguard-deprived circum- stances. See, e.g., Stinson v. United States, 508 U.S. 36 (1993) (commentary to sentencing guidelines); Consolidation Coal Co. v. Federal Mine Safety and Health Review Comm'n, 136 F.3d 819 (D.C. Cir. 1998) (interpretation implicit in Commission's decision to bring enforcement action); Nation- al Wildlife Fed'n v. Browner, 127 F.3d 1126, 1129 (D.C. Cir. 1997) (litigating position, as long as it is agency's actual and deliberated-upon view); Paralyzed Veterans, 117 F.3d at 581- 82 (supplement to DOJ's ADA Title III Technical Assistance Manual).

The Department of Justice goes on to characterize its chief, the Attorney General, as the entity actually responsible for the initial grant; from that assumption it reasons that Con- gress likely intended no deference for the Special Division's interpretations of the initial s 593 grant (which is all that s 594(e) referrals are). See Martin v. Occupational Safety and Health Review Comm'n, 499 U.S. 144, 158 (1991) (Con- gress may divide various powers as it wishes, subject to broader constitutional limitations). But the fact that the Attorney General initiates the appointment process and makes an initial suggestion of jurisdiction cannot be a basis for withholding deference to the Special Division: in the end, the order appointing the Independent Counsel and setting out his jurisdiction is articulated and issued by the Special Divi- sion as its own action. See 28 U.S.C. s 593(b)(1); Paralyzed Veterans, 117 F.3d at 585 (identity of actual regulation- drafter irrelevant once regulation is "put out by [the latter agency] as its own"). We therefore presume that the Special Division's interpretations receive deference, see Martin, 499 U.S. at 151 ("[W]e presume that the power authoritatively to interpret its own regulations is a component of the agency's delegated lawmaking powers."), and we find no intent to overcome that presumption.

Morrison, it is true, requires that "the jurisdiction that the [Special Division] decides upon must be demonstrably related to the factual circumstances that gave rise to the Attorney General's investigation and request for the appointment of the independent counsel in the particular case." 487 U.S. at 679. This sets out the constitutional boundary for the Special Division's initial action. Even if we assume its extension to the Division's later s 594(e) interpretation of that grant, nothing helpful to the Department's or defendants' position would follow. It would be a curious revival of the discredited doctrines of "constitutional fact" and "jurisdictional fact" to infer from the constitutionality of the boundary that an Article III court (as such) must draw it de novo. See John Dickinson, "Crowell v. Benson: Judicial Review of Adminis- trative Determinations of Questions of 'Constitutional Fact,' " 80 U. Pa. L. Rev. 1055, 1072-75, 1077-79 (1932) (explaining logical errors in doctrine now generally regarded as mori- bund);3 see also Oklahoma Natural Gas Co. v. FERC, 28 F.3d 1281, 1283-84 (D.C. Cir. 1994) (Chevron deference appli- cable even to questions of agency jurisdiction and preemption of state power). There is no claim by defendants, moreover, that if the disputed referral is within the original grant, it that the Independent Counsel can freely prevent. Further, a grant by one authority and denial by the other need not necessarily constitute a conflict: the phrase "may refer" appears to include some discretion to decline referral even where the authority agrees with the proposed interpretation of the initial jurisdictional grant. Indeed, this apparent dis- cretion suggests the possibility--on which we express no opinion--that Congress intended only a grant of referral to be authoritative.

Finally, the claim of zero deference would if accepted render s 594(e)'s provision for referral by the Special Divi- sion meaningless. We do not believe Congress enacted this statutory procedure simply to relieve the solitude of the Independent Counsel's office.

The search warrant analogy--brought to mind by the defendants' and the Department's stress on the absence of adversary procedures--is also instructive. Although made ex parte and resolving constitutional questions, a determination of probable cause by a federal magistrate or state judge is given "great deference." Illinois v. Gates, 462 U.S. 213, 236 (1983) (internal quotation omitted). And in this context it is quite plain that neither the allocation of the power to an array of entities nor the possibility of denial by one judge before a grant by another stands in the way of deference to any particular warrant actually granted.

Both analyses lead us to deference, but employ different linguistic formulations. An agency's interpretation of its own regulation is upheld unless "plainly erroneous or inconsistent" with the regulation, Bowles v. Seminole Rock & Sand Co., 325 U.S. 410, 414 (1945), while a search warrant is valid if the magistrate had a "substantial basis" for his finding of con- formity to the applicable standard (probable cause). Gates, 462 U.S. at 238. Such formulations do not necessarily con- say, we believe that a s 594(e) referral from the Special Division falls into the "deference" category. The common thread of deference formulations being reasonableness, see McKinney, 919 F.2d at 423,5 we believe that the Special Division's decision to refer must be upheld if reasonable and rejected if not.

* * * The statute sets a minimum on the scope of the jurisdiction the Special Division is to grant. We do not think the Special Division's referral is unreasonable even if compared to this minimum. Indeed, we find the indictments themselves within the statutory minimum jurisdiction even without deference to the referral.

The statute begins by directing the Special Division to "assure that the independent counsel has adequate authority to fully investigate and prosecute the subject matter with respect to which the Attorney General has requested the appointment of the independent counsel, and all matters related to that subject matter." 28 U.S.C. s 593(b)(3) (em- phasis added). As we shall see, there is an ambiguity in the definition of that core "subject matter," but, under the sen- tence as a whole, the Independent Counsel may prosecute anything "related to" it.

The statute continues:

Such jurisdiction shall also include the authority to inves- tigate and prosecute Federal crimes ... that may arise out of the investigation or prosecution of the matter with respect to which the Attorney General's request was Rather we think the minimum statutory space must be read, as we have said in the past, in accord with the purposes of the statute. Its "central purpose ... is to permit the effective investigation and prosecution of high level govern- ment and campaign officials." United States v. Wilson, 26 F.3d 142, 148 (D.C. Cir. 1994) (emphasis added). Discussing the "related to" language of s 593(b)(3), we noted that "the scope of a special prosecutor's investigatory jurisdiction can be both wide in perimeter and fuzzy at the borders." Id. The word "relation" thus comprises more than identical twins. And just as a person is "related" not only to his parents and children, but to grandchildren and grandparents,7 the fact that a crime is in some sense a verbal step or two away from the core crime cannot render it unrelated.

More concretely, the jurisdiction to look into matters "re- lated to" the core areas of initial inquiry must allow the Independent Counsel enough leeway to investigate and prose- cute such matters as are appropriate for him to effectively carry out his mandate. We think such effectiveness cannot be secured unless the Independent Counsel is at least able to pursue crimes ancillary to the commission or concealment of crimes in the core area.

The rationale for jurisdiction in this case is the same under either the "related to" or "arising out of" phrases in the statute. If payments Hubbell received beginning in 1994 were indeed hush money to secure Hubbell's silence vis--vis Whitewater, the possible obstruction of justice therein would certainly be a crime "relating to" Whitewater, for it would be an attempt to cover up the wrongdoing afterward. Both the Department of Justice and the defendants admit as much. They nevertheless argue that the tax charges here, as well as wire fraud and mail fraud aimed at keeping the income from the IRS and others who would have resulting claims, are not like the "arising out of" crimes specified in s 593(b)(3): "per- jury, obstruction of justice, destruction of evidence, and intim- idation of witnesses." The latter, argues DOJ, involve "con- engaged in organized criminal activities; notorious or power- ful with respect to local criminal activities," etc. Id. s 4566.1(2)(a).

That does not mean, of course, that the Independent Coun- sel is bound by the specific provisions of various executive branch manuals. The dissent faults the Independent Counsel for failure to comply with a provision of the DOJ Manual requiring IRS approval as a predicate to tax cases. See Dissent at 18-19. The dissent rests this complaint on s 594(f)(1), which requires the Independent Counsel to follow DOJ policy except where "to do so would be inconsistent with the purposes of this chapter." 28 U.S.C. s 594(f)(1). It suggests that the court wrongly "assume[s]" such inconsisten- cy. Dissent at 19. But defendants have never raised a claim under s 594(f)(1). In the absence of any effort to assert the section (and thus any opportunity for the Independent Coun- sel to defend himself), an inference that its exception applies seems fairly grounded: very little independence would be left in the office if the Independent Counsel had to run to DOJ or other executive branch agencies whenever DOJ established such sign-off procedures.11

Thus, while the Independent Counsel is differently situat- ed, other agencies' views on the links between crimes provide useful guidance. It is unreasonable that the Independent Counsel should be hamstrung by the need to prove every proposition necessary for jurisdiction by the exacting stan- dards suggested by the defense. If he were, even his investi- gations would be severely limited, for the statute gives no linguistic hook for requiring a lower standard of proof for investigatory jurisdiction. Furthermore he, unlike every oth- er prosecutor, would be unable to use a prosecution of an Finding that the Independent Counsel had developed its case solely on the basis of records that Hubbell had turned over under the grant of statutory use-immunity, the district court dismissed the indictment with respect to him. As the Inde- pendent Counsel had only discovered the extent and nature of Hubbell's alleged tax violations through his response to a government subpoena, the court concluded that the Indepen- dent Counsel had improperly turned Hubbell into the primary witness against himself. The government appeals from this ruling, asserting that the district court misconstrued the protection accorded by the federal use-immunity statute, see 18 U.S.C. ss 6002, 6003, as well as the Fifth Amendment's privilege against self-incrimination with which it is coexten- sive. Because the district court utilized an improper legal standard in assessing the scope of Hubbell's Fifth Amend- ment privilege, we vacate its decision and remand for it to conduct a hearing as to the extent of the Independent Coun- sel's knowledge of the records maintained by Hubbell at the time the subpoena issued.

A.Background

In the course of its ongoing investigation into possible criminal activity related to Madison Guaranty Savings & Loan Association and the Whitewater Development Corpora- tion, the Independent Counsel learned that Webster Hubbell had received payments from entities "associated with" Presi- dent William Jefferson Clinton for consulting work allegedly performed after Hubbell's 1994 resignation from his position as the Associate Attorney General. See In re Madison Guar. Sav. & Loan Ass'n, Div. No. 94-1 (D.C. Cir. Spec. Div. filed Dec. 31, 1997) (application for order of referral to Indepen- dent Counsel at 3). Through a preliminary investigation undertaken on its own initiative, the Independent Counsel sought to determine whether the payments were related to what it later described as Hubbell's "unwillingness to cooper- ate fully with the [Whitewater] investigation, as his plea agreement obligated him to do." Id. On October 31, 1996, the federal grand jury in the Eastern District of Arkansas issued a subpoena directing Hubbell to turn over eleven categories of business and income related documents, as well as personal records of his activities and of his family's fi- nances, covering the period from January 1, 1993 to the date of the subpoena.12 a series of questions tied to each of the eleven categories of documents requested in the subpoena. With respect to each, the Independent Counsel read the relevant paragraph (or a summary thereof) and then asked either, "Did you provide all those documents?" or "Those are all the records in your possession, custody, or control; is that correct?" Hubbell answered "Yes" to all eleven queries.13 The Independent Counsel closed the session by inquiring "have you searched or have you made a thorough search or caused a thorough search to be made in response to this Subpoena?" Hubbell again replied "Yes." 11/19/1996 Tr. at 11-12.

The Independent Counsel's search for evidence into wheth- er Hubbell might have obstructed its Whitewater investiga- tion revealed potential violations of the Internal Revenue Code. Using the contents of the documents Hubbell turned over to the grand jury, the Independent Counsel identified and developed evidence that culminated in the prosecution at issue in this case. On April 30, 1998, a federal grand jury in the District of Columbia issued a ten-count indictment alleg- ing that Webster Hubbell, Suzanna Hubbell, Michael Schau- fele and Charles Owen14 had conspired to defraud the United States Department of the Treasury and Internal Revenue Service, the State of Arkansas, the District of Columbia, and the Rose Law Firm of monies owed by Webster and Suzanna Hubbell (collectively, the "Hubbells").15 The indictment fur- examining the extent of Hubbell's Fifth Amendment privi- lege. Drawing from the framework sketched in Fisher v. United States, 425 U.S. 391 (1976), and reiterated in United States v. Doe, 465 U.S. 605 (1984) (Doe I), it focused upon the testimonial and incriminating aspects of the act of production. In light of the government's admission that it had utilized "the information provided by Mr. Hubbell pursuant to the production immunity," United States v. Hubbell, 11 F. Supp. 2d 25, 34 (D.D.C. 1998), the court found that Hubbell had not only communicated the authenticity and his possession of the documents, but also implicitly testified as to the very exis- tence of documents which added to the "sum total" of the government's information against him.19 Neither the exis- tence of the documents nor their contents was a "foregone conclusion," Fisher, 425 U.S. at 411, because the Independent Counsel had no source of knowledge independent of Webster Fisher, 425 U.S. at 410 ("The act of producing evidence in response to a subpoena nevertheless has communicative as- pects of its own, wholly aside from the contents of the papers produced."). The enforcement authority that rests behind the issuance of any subpoena provides the requisite compul- sion. See id. at 409.

Specifically, the act of production communicates at least four different statements. It testifies to the fact that: i) documents responsive to a given subpoena exist; ii) they are in the possession or control of the subpoenaed party; iii) the documents provided in response to the subpoena are authen- tic;20 and iv) the responding party believes that the docu- ments produced are those described in the subpoena. See Fisher, 425 U.S. at 410; Doe I, 465 U.S. at 614 n.13.21 lower court's factual determination that complying with the subpoena would involve testimonial self-incrimination. See id. at 614.

Against this settled backdrop, the case at bar presents a series of unsettled questions. Our sister courts have yet to reach agreement on the particular elaboration and proper application of the Fisher and Doe I framework. The degree to which a communication must be testimonial, what the Doe I Court described as its "testimonial value," 465 U.S. at 613, before it will invoke the Fifth Amendment's protections nec- essarily falls somewhere in between the poles represented by Doe I and Fisher. Precisely where on this continuum a given document production crosses the rubicon remains undeter- mined. The same can be said for the requisite quantum of incrimination. Finally, since Webster Hubbell produced the subpoenaed documents under a grant of immunity, we must also determine the extent of the protection afforded by sec- tion 6002. As Kastigar teaches, that inquiry leads us straight back to the scope of the Fifth Amendment privilege. See 406 U.S. at 453 (immunity must be commensurate with the Fifth Amendment's protections). Bearing in mind the Supreme Court's prescription that "[t]hese questions perhaps do not lend themselves to categorical answers," and that "their resolution may instead depend on the facts and circumstances of particular cases or classes thereof," Fisher, 425 U.S. at 410; Doe I, 465 U.S. at 613, we discuss each in turn.

b. Testimonial Communications The court below found that Hubbell's compelled act of production required him to make communications as to the authenticity, possession, and existence of the documents. See Hubbell, 11 F. Supp. 2d at 35.23 Sidestepping this conclusion, the Independent Counsel argues that the Fifth Amendment's protection should not attach because Hubbell's response to the subpoena had insufficient testimonial value. In its view, of accountant's work papers sought through the challenged subpoena. The taxpayer had also stipulated to both the existence of the documents and that they were those de- scribed in the subpoena. See Fisher, 425 U.S. at 430 n.9 (Brennan, J., concurring). The actual production of such records accordingly added little, independent of the docu- ments' substance, to the government's quantum of knowledge. Its testimonial value was negligible. In Doe I, by contrast, where the government sought a broad range of material which could similarly be classified as ordinary income, finan- cial and business records, the Court held that the act of production would have testimonial value meriting Fifth Amendment protection.24 While the Court left open the also emphasizes its thorough knowledge of Hubbell's financial records as a result of its investigation into charges that Hubbell committed tax and mail fraud while working at the Rose Law Firm. During the period of time with which the Independent Counsel claims familiarity--1989-1992--Hubbell worked as the billing partner at a law firm in Little Rock, Arkansas. During the period of time underlying this prose- cution, Hubbell worked in Washington, D.C. as a consultant, and served out a term in prison for the mail and tax fraud counts to which he previously pled guilty. Unless the Inde- pendent Counsel can establish its knowledge with a greater degree of specificity, the mere allegation that it was once familiar with Hubbell's finances does not make the existence or possession of the records sought a foregone conclusion. See Maggio v. Zeitz, 333 U.S. 56, 65 (1948) (orders enforcing a subpoena "should not be issued ... merely on proof that at some past time [the summoned documents were] in [the] possession or control of the accused party, unless the time element and other factors make that a fair and reasonable inference").

i. The Legal Standard To formulate an appropriate legal standard as to the de- gree of prior knowledge needed to render the existence, possession or authenticity of documents a foregone conclu- sion, it is necessary to return to first principles. See Doe v. United States, 487 U.S. 201, 209 (1988) (describing Fisher and Doe I as applying "basic Fifth Amendment principles" articu- lated in general terms) (Doe II). As the Supreme Court moved away from the doctrine articulated in Boyd v. United States, 116 U.S. 616, 634-35 (1886) ("a compulsory production of the private books and papers of the owner of goods sought to be forfeited ... [compels] him to be a witness against himself, within the meaning of the Fifth Amendment to the Constitution."), and towards a more literal interpretation of the privilege against self-incrimination, see Fisher, 425 U.S. at 401 ("We cannot cut the Fifth Amendment completely loose from the moorings of its language ..."), it jettisoned the personal privacy justification in favor of a rationale tied far more directly to the nature of government compulsion.

The core idea can be traced back at least to Justice Holmes' decision in Holt v. United States, 218 U.S. 245 (1910), in which the Court rejected a Fifth Amendment challenge to a witness' testimony establishing that the defendant had donned and fit into a blouse worn in a murder for which he was being tried. Dismissing what he characterized as an "extravagant" allegation, Justice Holmes explained that "the prohibition of compelling a man in a criminal court to be witness against himself is a prohibition of the use of physical or moral compulsion to extort communications from him, not an exclusion of his body as evidence when it may be materi- al." Id. at 252-53. Justice Holmes thereby drew a funda- mental distinction between government action that extorts communication--such action falls within the umbrella of pro- tection afforded by the Fifth Amendment--and government action that merely utilizes the body of the accused as a form of evidence--that kind of action falls outside the Amend- ment's particular orbit. Subsequent cases echo and develop this focus upon the Fifth Amendment as a barrier against compulsion that acts upon, and requires the exercise of an individual's mental faculties for communication.26 Schmerber v. California, 384 U.S. 757, 764-65 (1966) (privilege against self-incrimination does not extend to a compelled blood sam- ple), Gilbert v. California, 388 U.S. 263, 265-67 (1967) (privi- lege does not extend to compelled handwriting exemplar), United States v. Wade, 388 U.S. 218, 222-23 (1967) (privilege does not extend to compelled voice exemplar), and United States v. Dionisio, 410 U.S. 1, 5-6 (1973) (privilege does not extend to compelled voice sample), all rely upon the essential distinction between compulsion which operates upon the mind by forcing the accused to communicate information or testi- mony, and compulsion which merely requires him to produce his body for inspection. While in each instance the govern- ment draws evidentiary inferences as a result of the compul- sion, the Fifth Amendment only protects against those infer- ences which derive from compelled communication. See Pennsylvania v. Muniz, 496 U.S. 582, 593 (1990) (though videotape of the defendant exhibiting signs of intoxication does not violate the Fifth Amendment, videotape showing the defendant's inability to respond to a question about the date of his sixth birthday does); Doe II, 487 U.S. at 211 n.10 (describing the Schmerber line of cases as distinguishing between "the suspect's being compelled himself to serve as evidence and the suspect's being compelled to disclose or communicate information or facts that might serve as or lead to incriminating evidence"); Schmerber, 384 U.S. at 765 ("Not even a shadow of testimonial compulsion upon or enforced communication by the accused was involved either in the extraction or in the chemical analysis" of appellant's blood); Dionisio, 410 U.S. at 5-6 ("It has long been held that the compelled display of identifiable physical characteristics in- fringes no interest protected by the privilege against compul- sory self-incrimination.").

The dissent misreads the letter and logic of Fisher and Doe I because it fails to grasp the significance of the Supreme Court's distinction between compulsion which uses the body as evidence and that which operates upon the mind by compelling communicative acts. Instead, our colleague at- tempts to dissect the testimonial and non-testimonial ele- ments of providing blood, voice and handwriting samples. He argues that in giving blood, a person implicitly says, "This is my blood", Dissent at 3; in providing a handwriting sample, the accused admits his ability to write and that the exemplar is his. See id. at 4. But because in both these cases, it will require another witness to identify the accused's voice as that of a bank robber, or DNA testing to match the accused's blood with stains left at a crime scene, the giving of blood or an exemplar will not be considered protected testimony. See id. at 3. While undoubtedly true, the point is ultimately tangential to the proper inquiry. The real question at issue in Holt, Schmerber, Wade, and Gilbert was whether the government had merely used the accused's body as a form or piece of evidence, or whether the government had to compel communicative testimony to obtain the evidence it needed. Justice Holmes' Holt opinion likened trying on a blouse to simply sitting before the jury and allowing them to compare your features to that contained in a photograph of the perpe- trator. See 218 U.S. at 253. While it can be argued that the accused implicitly testified about something in each of the cases cited by the dissent--that this is my blood containing my unique DNA, that this is my face with all of its character- istic idiosyncracies, that this is my body with a particular shape and size which fits into this blouse--that testimony was irrelevant to the Fifth Amendment inquiry because it re- quired no act of will on his part as to what he would communicate. The same reasoning applies to a compelled submission to fingerprint analysis. The suspect can be said to be communicating that this is my hand and it contains five fingerprints unique to my person, but in reality the individual has merely been compelled to make himself available as a "source of 'real or physical evidence.' " Schmerber, 384 U.S. at 764. For purposes of Fifth Amendment analysis, it is dispositive that the government has no need to rely upon the witness's truthtelling to secure the evidence it seeks.

The dissent cites to Holt, Schmerber, Wade, and Gilbert for the proposition that where the government can draw a link between evidence and the accused, independent of the ac- cused's testimony, the Fifth Amendment does not apply. See Dissent at 2-5, 11. None of these cases, read individually or taken together, however, stands for this general proposition. All of the cases focused on whether the act in question was communicative or noncommunicative--whether it relied upon the individual's mental faculties and truthtelling capacity or merely used the body as a source of physical evidence--and not on whether the prosecution could link evidence to the accused without relying on his testimony.27 In each, the its progeny, they fall on the side of the communicative and testimonial.

In assessing the testimonial value of an act of production, it makes sense to reference the anti-extortion principle which has become the motivating force of self-incrimination doc- trine. In light of Fisher, Doe I, and Doe II, we conclude that the testimonial value varies directly with the quantum of information that the government seeks to extract through compelling the expression of the contents of an individual's mind and inversely with the quantum of information in the government's possession at the time the relevant subpoena issues.30 Cf. Muniz, 496 U.S. at 597 ("the cases upholding compelled writing and voice exemplars did not involve situa- tions in which suspects were asked to communicate personal beliefs or knowledge of facts ..."). Although the Supreme Court has not explicitly stated as much, our conclusion is fully in accord with, and even helps to explain, the Court's re- peated statement that the question of whether tacit aver- ments are sufficiently testimonial as to merit the Fifth Amendment's protection depends "on the facts and circum- stances of particular cases or classes thereof."31 Fisher, 425 location of the passport were not in question, the Second Circuit affirmed its conclusion that the act of production did not have sufficient testimonial value to implicate the Fifth Amendment. See id. at 1063.

In re Steinberg, 837 F.2d 527 (1st Cir. 1988), and United States v. Clark, 847 F.2d 1467 (10th Cir. 1988) also fall within the pattern of cases in which the government's knowledge is nearly on par with that of the subpoenaed individual. In Steinberg, the government sought a series of notebooks main- tained by members of Lyndon LaRouche's security staff in which they detailed the progress of a federal investigation into LaRouche's 1984 Presidential campaign as well as the staff's internal planning measures. As a government witness had testified to the existence and contents of the notebooks, as well as to Steinberg's possession, the court found that the subpoena did not implicate the Fifth Amendment. Steinberg, 837 F.2d at 530. In Clark, an IRS summons sought accoun- tant's work papers and the personal records that the taxpayer had given his accountant, both of which were known to have subsequently been given to the taxpayer's attorney. Finding the existence of the work papers to be a foregone conclusion, and that the underlying records would be reflected therein, the Tenth Circuit refused to recognize a Fifth Amendment privilege. See 847 F.2d at 1472-73. See also United States v. Stone, 976 F.2d 909, 911 (4th Cir. 1992) (in Department of Energy investigation into relationship between department employees and owner of business receiving DOE contracts, where government knew that appellant owned a beach house and had sufficient knowledge that he rented the house out between 1983 and 1989 to request a list of renters, subpoena seeking utility bills and rental records targets documents whose existence and possession are a foregone conclusion); In re Grand Jury Subpoena Duces Tecum Dated Oct. 29, 1992, 1 F.3d 87 (2d Cir. 1993) (where defendant has testified before SEC as to possession and use of diary, and has previously turned over a copy in which government suspects adulteration, subpoena for original does not implicate Fifth Amendment).

By contrast, those cases in which our sister courts have recognized a Fifth Amendment privilege consistently reveal a gross information asymmetry between the government and the subpoenaed party, which can be bridged only by getting at the contents of the latter's mind. In United States v. Fox, 721 F.2d 32 (2d Cir. 1983), the IRS issued a wide ranging subpoena seeking an expansive array of personal, financial and business records of Dr. Martin Fox. At the time, the IRS had copies of his 1979-1981 tax returns and a transcript of third-party payments to the Foxes during 1979. Rejecting the government's contention that the testimonial value of producing the subpoenaed documents would be minimal, the court reasoned that in seeking all of the books and records of the Fox's sole proprietorship, the government had attempted "to compensate for its lack of knowledge by requiring Dr. Fox to become the primary informant against himself." Id. at 38. Since the IRS could establish neither the existence nor the authenticity of the records sought, its subpoena implicated the Fifth Amendment's protections. In re Grand Jury Pro- ceedings on February 4, 1982, 759 F.2d 1418 (9th Cir. 1985) involved a similarly broad subpoena, demanding personal journals, files related to the purchase of fishing boats, stock transactions, and receipts. Reasoning that the production of documents belonging to and prepared by the subpoenaed party would relieve the government of the need to establish their existence, possession and authenticity, the court held that the act of production would be sufficiently testimonial to require protection if incriminating. See id. at 1421. Finally, in In re Grand Jury Proceedings, Subpoenas for Documents, 41 F.3d 377 (8th Cir. 1994), the government subpoenaed all of appellants' original records from any business in which they owned an interest as well as all evidence of their financial transactions for a four-year period. Noting that the subpoe- na did not describe the requested documents in specific terms, the court found that the government had failed to establish an independent source of authentication. Focusing on the discretionary judgments involved in responding to a subpoena, the court pointedly remarked that "the broader, more general, and subjective the language of the subpoena, the more likely compliance with the subpoena would be testimonial." Id. at 380.

Although the fit is by no means perfect, the cases assessing the testimonial value of an act of production form a pattern that correlates strongly with the government's need to draw from the mental faculties of the responding party. Where the information asymmetry is large, and where the govern- ment's prior knowledge is minimal, the act of production will likely communicate either the existence, possession or authen- ticity of the subpoenaed documents.

ii. Assessing the Government's Knowledge The case at bar highlights a further doctrinal ambiguity, which is again tied to elaborating the extent of knowledge that the government must have in order to justify a conclu- sion that the communicative aspects of the act of production are a "foregone conclusion" under Fisher and Doe I. Re- hearsing an argument discussed above, the Independent Counsel contends that government knowledge of the exis- tence and possession of documents should be assessed solely through examining categories or classes of documents. While it thus argues that the court should measure whether the information sought through a subpoena is a foregone conclu- sion at a high degree of abstraction--speaking in generalized terms of business, financial and tax records--the Independent Counsel provides no support for its contention. Moreover, any instruction to filter review of the government's knowl- edge through categories is inherently an empty one, for it fails to address the recurring level of generality problem. See generally, Laurence H. Tribe & Michael C. Dorf, Levels of Generality in the Definition of Rights, 57 U. Chi. L. Rev. 1057 (1990) (discussing the malleability and outcome- determinative nature of levels of generality); Michael H. v. Gerald D., 491 U.S. 110 (1989) (discussing the appropriate level of generality at which to define fundamental rights). While a useful method of sorting information, categories do not present themselves as Platonic forms with inherent shape or universal meaning. Rather, defined by reference to their particular applications, they can be abstracted upwards or downwards (through varying levels of generality) in order to embrace or reject concrete instances. See generally Ludwig Wittgenstein, Philosophical Investigations (G.E.M. An- scombe trans., 3d ed. 1968) at ss 137-242 (general concepts do not dictate their concrete applications, but rather are defined through them). As Doe I well illustrates, the govern- ment cannot simply subpoena business records and then claim the requisite knowledge for purposes of the Fifth Amendment by pointing to the existence of a business.33 The Independent

As the district court's fact findings relevant to the Fisher/Doe I inquiry are inextricably linked with its assess- ment of the government's substantive knowledge of the al- leged offenses, we cannot decide on the record before us whether Hubbell's act of production had sufficient testimonial value to invoke the Fifth Amendment's protections. The subpoena speaks in vague terms, and the detail with which it goes through the possible forms that the information sought could take, see supra note 12, at the very least hints that the government had no knowledge as to whether Hubbell main- tained comprehensive records of the way he allocated his time. Moreover, it is unclear how the Independent Counsel became apprized of the Pulaski bank account, the three trust accounts, the Bridgeport Group, the check swap, and Hub- bell's early withdrawals from his IRA accounts, each of which figure prominently in the indictment.36 See discussion supra pp. 25-26. The Bridgeport Group had not been organized, its account at the Pulaski Bank had not been opened, and Hubbell had neither signed a book contract with William Morrow and Company, Inc., nor received his advance on the day the subpoena issued. The extent of the Independent Counsel's knowledge of Hubbell's recordkeeping practices is also uncertain. On remand, the district court should hold a hearing in which it seeks to establish the extent and detail of the government's knowledge of Hubbell's financial affairs (or of the paperwork documenting it) on the day the subpoena issued. It is only then that the court will be in a position to nary" business and financial records can almost never be incriminating. Similarly, the implicit authentication of docu- ments would only incriminate were a subpoena to be phrased in such a way as to expressly request production of the instruments of criminality. We disagree, as both logic and Supreme Court precedent rebut the claims of any such nig- gardly interpretation. First, the Fifth Amendment's protec- tions cannot depend upon such trivial semantic distinctions that the government can sidestep its application by request- ing "all income records" instead of "all incriminating income records." Artful phrasing does not suffice. Moreover, Doe I and Doe II belie such a narrow reading of the Fifth Amend- ment's protections. In Doe I, the Supreme Court found, on the basis of the findings presented, that Doe faced a "real and substantial" risk of incrimination were he to produce the documents sought in the government's subpoena-a subpoena seeking ordinary business records, not "incriminating" busi- ness records. See 465 U.S. at 614 n.13. As the government had no independent knowledge of the existence or his posses- sion of the documents listed in the subpoenas at issue, and since the act of production would tacitly admit their existence and his control over them, and provide a source of authentica- tion, the court found the threat of incrimination to be "clear." See id. A cursory comparison of the subpoena at issue in Doe I with that of the case at bar reveals marked similarities, see id. at 607 n.1, and we think it equally clear that Hubbell faced a real and substantial threat of incrimination in responding. It is hard to divine another reason why the Independent Counsel would have sought an order of immunity in the first place.

Doe II's discussion of the incrimination requirement rein- forces our conclusion. Although the decision in Doe II turned on the question of whether signing a general release form involved an assertion of fact, the Court also discussed the proper shape of the incrimination inquiry. The Kastigar Court, it argued, "implicitly concluded that the privilege prohibits 'the use of compelled testimony, as well as evidence derived directly and indirectly therefrom.' 406 U.S. at 453. The prohibition of derivative use is an implementation of the 'link in the chain of evidence' theory for invocation of the privilege, pursuant to which the 'compelled testimony' need not itself be incriminating if it would lead to the discovery of incriminating evidence." Doe II, 487 U.S. at 208 n.6.

In the present case, it appears that Hubbell's testimony likely involved both direct and indirect incrimination. Ac- knowledging the existence of an interest-bearing checking account the income from which the subpoenaed party had failed to report on his tax returns would directly incriminate; it would inform the government of a known source of unre- ported income. See United States v. Argomaniz, 925 F.2d 1349, 1354 (11th Cir. 1991) (where defendant failed to file tax returns for a series of years, admitting the existence of documents relating to income through production would es- tablish essential elements of criminal failure to file a tax return). If Hubbell had records of that account in his possession or control, that fact could further incriminate. See Smith v. Richert, 35 F.3d 300, 304 (7th Cir. 1994) (as the mere turning over of 1099s and W-2s in response to a subpoena could eliminate defense of lack of knowledge or possession, it is incriminating). Similarly, in acknowledging the existence of the Bridgeport Group and its bank account at Pulaski Bank, Hubbell provided a link in the chain of evidence used by the Independent Counsel to substantiate the criminal charges against him--an instance of indirect incrimination.

Given the procedural posture of this case, it would be premature for us to review the incrimination question any further at this juncture. Until the district court determines on remand precisely what testimony Hubbell provided through his act of production, focusing on the extent of the government's knowledge as of the date of the subpoena and on whether Hubbell's testimony added "to the sum total of the government's case against him," United States v. Edger- ton, 734 F.2d 913, 921 (2d Cir. 1984) (quoting Fisher, 425 U.S. at 411), it cannot make the appropriate fact findings as to what extent that testimony is incriminating. See Doe I, 465 U.S. at 614 (whether a compelled act of production is incrimi- nating is a question of fact). In conducting this inquiry, the district court should be guided, as the Supreme Court coun- seled in Hoffman v. United States, 341 U.S. at 487, by its particular perceptions of the specific and unique facts of the case.

d. The Upshot of Immunity A grant of statutory immunity under 18 U.S.C. ss 6002, 6003, extends as far as the Fifth Amendment privilege it supplants. It "leaves the witness and the prosecutorial au- thorities in substantially the same position as if the witness had claimed the Fifth Amendment privilege," and is "coexten- sive" with the Fifth Amendment's protections. Kastigar, 406 U.S. at 462. Since the district court erred in assessing the testimonial value of Hubbell's document production, it con- ducted its inquiry into the effect of Hubbell's statutory immu- nity against a faulty backdrop. On remand, the court should assess the impact of the immunity order in light of its new fact findings with respect to the government's prior knowl- edge and the quantum of information it extracted from "the state of mind, memory, perception, or cognition of the wit- ness." Braswell, 487 U.S. at 126 (Kennedy, J., dissenting). Although this inquiry will be fact-intensive, the district court should bear in mind Kastigar's teaching that "a grant of immunity must afford protection commensurate with that afforded by the privilege, [although] it need not be broader." 406 U.S. at 453. The precise contours of Hubbell's Fifth Amendment rights, therefore, will be dispositive.

Intervening in this case, the United States, acting through the Attorney General, has proffered a particular reading of the Fifth Amendment's intersection with compelled produc- tion which we believe merits some discussion. Like the Independent Counsel, the United States draws a sharp dis- tinction between the testimonial components of the act of production and the contents of those documents, essentially ruling out the possibility that the prohibition on the direct or indirect use of a party's compelled testimony could extend to reach the contents of the documents he turns over. Instead, it invites the court to compare what the government learns from the act of production with what it would know if the documents in question just appeared on its doorstep. That intellectual exercise, it argues, separates the information con- veyed through the act of production with what could be deciphered from the records themselves. See Intervenor's Br. at 42; Dissent at 11. Since Kastigar instructs that the government can introduce the fruits of immunized testimony provided that it can meet "the heavy burden of proving that all of the evidence it proposes to use was derived from legitimate independent sources," 406 U.S. at 461-62, the documents themselves can serve as that independent source of the information communicated by their production. Pro- vided that the government does not mention the mechanics through which it obtained those documents, and that the documents are sufficiently self-explanatory and self- referential to establish their own nexus with the defendant, the government would be free to use the subpoenaed docu- ments in making its case against the defendant.37 We dis- agree.

Although the Fisher Court observed that "[t]he 'implicit authentication' rationale appears to be the prevailing justifica- tion for the Fifth Amendment's application to documentary subpoenas," 425 U.S. at 412 n.12, the Court explicitly and repeatedly acknowledged that the act of production also communicates existence and possession. See id. at 412; Doe I, 465 U.S. at 613; Doe II, 487 U.S. at 209; Braswell, 487 U.S. at 103. The analytic tool offered by the United States, however, reads both of these testimonial components out of existence. While the government may be able to establish the authenticity of the documents independently, whether in terms of their own self-reference or the testimony of a witness familiar with them, the magical appearance of the weapon and the accused that it can divine after the fact? The questions answer themselves. They also rebut the theory of the act of production doctrine proffered by the Justice De- partment and the dissent. Since the Self-Incrimination Clause has always been understood to refer to testimony in all of its forms, whether communicated by voice or through physical acts, see Doe II, 487 U.S. at 209-10 n.8 ("Petitioner has articulated no cogent argument as to why the 'testimonial' requirement should have one meaning in the context of acts, and another meaning in the context of verbal statements."); Muniz, 496 U.S. at 595 n.9 (definition of testimonial communi- cation "applies to both verbal and nonverbal conduct"), the protection it accords to verbal communications must extend to the testimony conveyed through a compelled act of produc- tion.38

Now suppose a variation of our hypothetical, in which the police discover a victim's body in the basement of a large apartment building, and an autopsy establishes stabbing to be the cause of death. Lacking any clues, a grand jury issues a subpoena to every resident in the building, asking each to produce "all knives and other forms of cutlery that are now, or in the preceding month have been, in your possession or control." The residents object en masse, asserting inter alia their Fifth Amendment privilege against self-incrimination. After the prosecutor obtains an order compelling production and granting immunity to the maximum extent allowed by law, the residents comply with the subpoena. Among the recovered knives, the police discover the murder weapon. The prosecutor indicts its owner for murder. The defendant ty from the use of compelled testimony, as well as evidence derived directly and indirectly therefrom, ... prohibits the prosecutorial authorities from using the compelled testimony in any respect, and it therefore insures that the testimony cannot lead to the infliction of criminal penalties on the witness."). Accordingly, should the Independent Counsel prove unable to meet the requisite evidentiary burden, the contents of those documents will be inadmissible. See Sealed Case, 791 F.2d at 182 ("Thus, if in fact appellee's privilege in the act of production cannot be protected without excluding the contents of the tapes (a point on which we express no opinion) the District Court has the authority to prevent the government from referring to or introducing those con- tents.").

"The decision to seek use immunity necessarily involves a balancing of the Government's interest in obtaining informa- tion against the risk that immunity will frustrate the Govern- ment's attempts to prosecute the subject of the investigation." Doe I, 465 U.S. at 616. Unless the Independent Counsel can establish its knowledge of the existence and possession of the documents sought in the subpoena with greater detail and particularity, it will have to live with the consequences of its decision to compel production.

III. Conclusion For the reasons above, we vacate the district court's judg- ment and remand for further proceedings in light of this decision.

So ordered.

Wald, Circuit Judge, concurring in Part I: I believe that a reasonable construction of the original mandate shows that it is "demonstrably related" to the tax evasion and asset con- cealment prosecutions in dispute here. That, however, is only because, as the panel opinion makes clear, at least some of the consulting monies Hubbell received in the years 1994- 1997 may have been tainted as hush money. "The timing, sources, and extent of the payments make the belief that they were hush money reasonable." Panel Opinion ("Panel Op.") at 20. Thus, manipulation of all monies received during that period may have been related to the concealment of any tainted money--money, that is, received in return for ob- structing the Independent Counsel's Whitewater investiga- tion. Panel Op. at 15-18. In the face of these circumstances, I disagree with the dissent's conclusion that the Independent Counsel's failure to bring an indictment on the hush money allegations, either first or contemporaneous with the tax violations, means that no "credible evidence" of any wrongdo- ing exists. Dissenting Opinion ("Diss. Op.") at 16-17.

The simple fact that no prior or simultaneous indictment for obstruction was brought down specifically alleging that monies were received as a quid pro quo for noncooperation in the main Whitewater matter does not compel a conclusion that the tax evasion and asset concealment charges are not related to the original subject matter. It would, in my view, be unreasonable to require the Independent Counsel to bring authorized prosecutions in any special order or sequence provided that they are undertaken and continued only so long as the "relatedness" requirement remains satisfied. I believe such a standard may be inferable from the panel opinion, Panel Op. at 18, but I wish to make clear here my own view that at the point that the Independent Counsel is no longer diligently pursuing his investigation of the main allegation that money was channeled to Hubbell for noncooperation or misleading testimony, with a reasonable expectation that it will prove fruitful, the other ancillary claims of manipulation or failure to report taxes cease to be "demonstrably related" to the original mandate. It is not enough that the Indepen- dent Counsel could at one time state a reasonable belief that the underlying obstruction allegations were true. In my view, the underlying investigation into the hush money charge must have been ongoing, and there must have been credible evidence that the underlying offense did indeed occur at the time the ancillary indictment was filed. To the extent that prosecution for the obstruction offense was not practicable at the time of the tax indictments, whether because the main investigation had yet to be completed, or because the Inde- pendent Counsel believed that evidentiary problems limited the prospects of obtaining a conviction on the obstruction offense at that moment, the Independent Counsel would still have jurisdiction over the ancillary charges, provided that credible evidence remained to substantiate its belief that the obstruction crime had been committed. However, if at some point during the investigation it becomes apparent that this belief is unfounded, then at that time the ancillary counts should be passed over to the Justice Department to screen and, if appropriate, to prosecute. That does not, however, appear to be the case here--we were assured at oral argu- ment by the Independent Counsel himself that the main hush money investigation was not closed, and if there is any reason to believe that it is a dead letter at the time the indictment for ancillary tax matters was brought (or even when it is reinstated on remand here), then a hearing (in camera if necessary) by the district judge on the issue would be appro- priate. But assuming, as we must on the record before us, that the principal investigation into possible obstruction was alive at the time the ancillary tax indictments were brought and remains so, I cannot but conclude that the tax indict- ments were related to the main endeavor for the reasons set out in the panel opinion.1 Panel Op. at 15-18.

Finally, while I do believe that some deference is due the Special Division's interpretation of these prosecutions as re- lated to the original one, I would not find it necessary under the facts here to decide if the appropriate level of deference is substantial, Panel Op. at 7, or only due deference. I do not find the agency analogy particularly persuasive, but on the other hand I believe that the responsibilities of the Special Division under the statute to define the Independent Coun- sel's jurisdiction initially, and subsequently to assess requests for the referral of related matters, strongly militate toward some deference; the degree of deference due will likely depend on the circumstances of each case. The extent of that deference might shift according to whether the Special Divi- sion's relatedness determination is grounded particularly on facts made known to it by the Independent Counsel or the Attorney General or on a legal or conceptual conclusion that the offenses or persons are sufficiently related to the original mandate. At any rate, I think the relatedness requirement is satisfied here under either standard of deference.

Tatel, Circuit Judge, dissenting from Part I:

This court today concludes that the indictment of Webster Hubbell for failing to pay taxes on income earned in Washing- ton, D.C., in 1994 "arises out of" or "relates to" the Indepen- dent Counsel's investigation of various Arkansas land transac- tions in the mid-1980s known as Whitewater. Because this result expands independent counsel authority at the expense of the Executive Branch, I believe it violates the constitution- al principle of separation of powers. By deferring to the Special Division and by adopting virtually limitless theories of "relatedness," the court fails to police the boundaries that Morrison v. Olson deemed essential to the constitutionality of the independent counsel statute. See 487 U.S. 654 (1988). Mindful of these boundaries, which guarantee political ac- countability for the prosecutorial function, and of "the duty of federal courts to construe a statute in order to save it from constitutional infirmities," id. at 682, I would find that the tax indictment is not "demonstrably related to the factual circum- stances that gave rise to the Attorney General's ... request for the appointment of the independent counsel," id. at 679. I would therefore affirm the district court's order quashing the indictment.

I Vesting the executive power in the President of the United States, the Constitution directs that "he shall take Care that the Laws be faithfully executed." U.S. Const. art. II, s 3. To aid in this task, Congress specifically delegated to the Attorney General and her subordinates in the Department of Justice the power to "conduct any kind of legal proceeding, civil or criminal," 28 U.S.C. s 515(a) (1994); see id. s 516, including the prosecution of tax crimes, see 28 C.F.R. s 0.70 (1998).

The Ethics in Government Act of 1978 carved a narrow exception to the Attorney General's power to enforce federal criminal law. In the wake of the "extraordinary sequence of events" of the Watergate scandal--in particular, the firing of special prosecutor Archibald Cox--Congress saw the need for "the appointment of an independent temporary special prose- cutor for certain limited cases where the Department of Justice may have a conflict or interest with respect to a particular investigation." S. Rep. No. 95-170, at 6, 34 (1977), reprinted in 1978 U.S.C.C.A.N. 4126, 4222, 4250. Aware of the constitutional implications of creating an independent prosecutorial function outside the Executive Branch, Con- gress adopted an elaborate array of procedures controlling the appointment, powers, and jurisdiction of independent counsel. See 28 U.S.C. ss 592-596. Independent counsel may be removed by the Attorney General for "good cause," may perform only certain limited duties, and may act only within the scope of prosecutorial jurisdiction ceded by the Attorney General. See S. Rep. No. 95-170, at 56, 1978 U.S.C.C.A.N. at 4272 ("The prosecutorial jurisdiction of the special prosecutor is one of the most important devices for the control ... and the accountability of such a special prosecutor."). Absent expansion by the Attorney General under 28 U.S.C. s 593(c), the jurisdiction of an independent counsel is limited to matters related to the Attorney General's original request for appointment of an independent counsel under 28 U.S.C. s 592(d).

In Morrison, the Supreme Court found these constraints essential to the statute's constitutionality. Without them, the Court could not have characterized the independent counsel as an "inferior officer" under the Appointments Clause of the Constitution. See 487 U.S. at 670-77. Once an independent counsel investigates or prosecutes matters beyond the juris- diction ceded by the Attorney General, the independent coun- sel sheds his "inferior" status and becomes a "principal officer" requiring Presidential appointment and Senate confir- mation. See Buckley v. Valeo, 424 U.S. 1, 132 (1976) (per curiam). Although "the power to appoint inferior officers such as independent counsel is not in itself an 'executive' function in the constitutional sense," Morrison, 487 U.S. at 695, the power to appoint principal officers surely is. See U.S. Const. art. II, s 2, cl. 2. For that reason, the Morrison Court thought it constitutionally important that "the jurisdic- tion of the independent counsel is defined with reference to the facts submitted by the Attorney General" and that "the Act [28 U.S.C. s 594(f)] requires that the counsel abide by Justice Department policy unless it is not 'possible' to do so." 487 U.S. at 696 (emphasis added). These limitations, among others, "give the Executive Branch sufficient control over the independent counsel," id., and ensure that the Act does not " 'impermissibly undermine[ ]' the powers of the Executive Branch," id. at 695 (citation omitted). An independent coun- sel who exceeds the jurisdiction conferred by the Attorney General usurps the President's constitutional power to main- tain control of the inherently "executive" function of law enforcement. See id. at 691, 694.

Morrison expressed similar constitutional concerns about the role of the Special Division of this court. The statute allows the Special Division to appoint an independent counsel only after the Attorney General makes a preliminary investi- gation, determines that further investigation is warranted, prescribes a subject matter for investigation and potential prosecution, and "applies" to the Special Division for an appointment. See 28 U.S.C. ss 592(c)-(d), 593(b). In defin- ing an independent counsel's prosecutorial jurisdiction, the Special Division has only one function:

[to] assure that the independent counsel has adequate authority to fully investigate and prosecute the subject matter with respect to which the Attorney General has requested the appointment of the independent counsel, and all matters related to that subject matter. Such jurisdiction shall also include the authority to investigate and prosecute Federal crimes ... that may arise out of the investigation or prosecution of the matter with re- spect to which the Attorney General's request was made, including perjury, obstruction of justice, destruction of evidence, and intimidation of witnesses.

Id. s 593(b)(3). Describing the Special Division's power to define jurisdiction as "incidental" to its power to appoint, Morrison made clear that the Special Division's authority is quite limited:

[Congress] may vest the power to define the scope of the office in the [Special Division] as an incident to the appointment of the officer pursuant to the Appointments Clause. That said, we do not think that Congress may give the Division unlimited discretion to determine the independent counsel's jurisdiction. In order for the Divi- sion's definition of the counsel's jurisdiction to be truly "incidental" to its power to appoint, the jurisdiction that the court decides upon must be demonstrably related to the factual circumstances that gave rise to the Attorney General's investigation and request for the appointment of the independent counsel in the particular case.

487 U.S. at 679 (second emphasis added). To be sure, the statute also provides that the Special Division may, upon a request by an independent counsel, "refer to the independent counsel matters related to the independent counsel's prosecu- torial jurisdiction." 28 U.S.C. s 594(e). But Morrison left no doubt that referrals may not expand the original scope of an independent counsel's jurisdiction. See 487 U.S. at 680 n.18. Consistent with the separation of powers principle, only the Attorney General may expand an independent counsel's prosecutorial jurisdiction. See 28 U.S.C. s 593(c).

Thus, the Attorney General's "initial suggestion of jurisdic- tion," Majority Opinion ("Maj. Op.") at 8--or more precisely, as Morrison put it, "the factual circumstances that gave rise to the Attorney General's investigation and request for the appointment of the independent counsel," 487 U.S. at 679-- serves as the ultimate baseline for assessing the legality of any definition of jurisdiction by the Special Division as well as any exercise of authority by an independent counsel. Consti- tutionally required, this baseline reconciles the principle of separation of powers with the vesting of prosecutorial author- ity in an official independent of the Executive Branch. See id.

II In her June 30, 1994 application to the Special Division, the Attorney General prescribed the subject matter of the Inde- pendent Counsel's jurisdiction as follows: whether any individuals or entities have committed a violation of any federal criminal law, ... relating in any way to James B. McDougal's, President William Jeffer- son Clinton's, or Mrs. Hillary Rodham Clinton's relation- ships with Madison Guaranty Savings & Loan Associa- tion, Whitewater Development Corporation, or Capital Management Services, Inc.

Application for Appointment of Independent Counsel, In re Madison Guar. Sav. & Loan Ass'n (June 30, 1994). The application also authorized prosecution of any conduct ob- structing the investigation of this core subject matter. See id. Exercising its power under 28 U.S.C. s 593(b)(1), the Special Division on August 1, 1994 appointed Kenneth W. Starr as independent counsel and granted him prosecutorial jurisdiction over matters set forth in the Attorney General's application. See Maj. Op. at 3-4 (describing the Special Division's original grant of jurisdiction). Responding to sub- sequent requests from the Independent Counsel, the Special Division twice exercised its referral power under 28 U.S.C. s 594(e). See Maj. Op. at 4-5 (describing the September 1, 1994 and January 6, 1998 referrals).

The second referral arose from an application in which the Independent Counsel stated:

In the course of its investigation, this Office received information regarding payments to Mr. Hubbell from individuals and entities associated with the Clinton Ad- ministration. These payments were made starting in 1994, when Mr. Hubbell was publicly known to be under criminal investigation. This Office initiated a prelimi- nary investigation into whether these payments might be related to Mr. Hubbell's unwillingness to cooperate fully with the investigation, as his plea agreement obligated him to do. The grand jury has heard evidence related to the payments, including evidence that Mr. Hubbell may have committed fraud and tax crimes in connection with them.

Application for Order of Referral, In re Madison Guar. Sav. & Loan Ass'n (Dec. 31, 1997), at 3-4. In response, the Special Division's referral authorized investigation and prose- cution of crimes, including tax offenses, associated with Hub- bell's income since 1994, as well as other crimes such as obstruction of justice and perjury "related to payments that Mr. Hubbell has received from various individuals and enti- ties" since 1994. Order Granting Application for Order of Referral, In re Madison Guar. Sav. & Loan Ass'n (Spec. Div. D.C. Cir. Jan. 6, 1998). The Independent Counsel then brought a 10-count, 42-page indictment detailing an elabo- rate tax evasion scheme allegedly undertaken by Hubbell and his wife, along with their lawyer and accountant.

We now face the following question: Under sections 593(b)(3) and 594(e) of the Ethics in Government Act, does this indictment "arise out of" or "relate[ ] to" the original scope of the Independent Counsel's prosecutorial jurisdiction? To answer this question, we must evaluate the indictment not against the two referrals by the Special Division--those refer- rals could not have expanded the original scope of the Inde- pendent Counsel's jurisdiction, see Morrison, 487 U.S. at 680 n.18--but against the Special Division's original August 5, 1994 grant of authority. Given Morrison's constitutional limitation on the Special Division's power to define indepen- dent counsel jurisdiction, see id. at 679, the precise question before us is this: Is the indictment of Hubbell for tax evasion "demonstrably related to the factual circumstances that gave rise to the Attorney General's investigation and request for the appointment of the independent counsel"?

Before addressing this issue, this court holds that the Special Division's January 1998 referral is entitled to defer- ence--"substantial" deference, according to Judge Williams; at least "due" deference, according to Judge Wald. Either way, I disagree.

At the outset, I think it important to clarify that, as a procedural matter, we are not directly reviewing the Special Division's referral. The statute nowhere authorizes appeals from Special Division referrals. Courts of appeals and the Special Division play different institutional roles under the independent counsel statute. They act at different stages of the investigation, and they have different documents before them. Although the Special Division made an implicit deter- mination of "relatedness" in its referral, it did so in the context of a request for investigative and prosecutorial au- thority. That investigation has now progressed beyond mere allegations and has evolved into an indictment. Our task, like the district court's before us, is to resolve a specific dispute whose parameters have crystallized: We must decide whether the April 30, 1998 indictment, in all its concreteness and particularity, is "related to" the original scope of the Indepen- dent Counsel's prosecutorial jurisdiction. The Special Divi- sion did not resolve this question, nor could it.

In support of giving the referral deference, Judge Williams accepts the view that the Special Division "act[ed] quite like [an agency] glossing its own regulation" when it exercised its referral power. Maj. Op. at 8. However, the Special Divi- sion possesses none of the characteristics of agencies that entitle their legal judgments to judicial deference. See Chev- ron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 865 (1984). In making "related to" determina- tions, the Special Division exercises no special or technical expertise that circuit and district courts lack. The Special Division has no political accountability through either execu- tive or congressional oversight, and its members are neither appointed by the President nor confirmed by the Senate for their role. No provision of law allows for direct judicial review of its decisions. Unlike administrative agencies--and unlike even the U.S. Sentencing Commission, see 28 U.S.C. ss 991(b), 994-995--the Special Division exercises no congressionally-delegated policymaking responsibilities. This last fact alone undermines the agency analogy, for "the power authoritatively to interpret its own regulations is a component of the agency's delegated lawmaking powers." Martin v. Occupational Safety & Health Review Comm'n, 499 U.S. 144, 151 (1991) (emphasis added), quoted in Maj. Op. at 9.

Morrison left no doubt that in both form and function, the Special Division is a court, not an agency. The opinion consistently refers to the Special Division as a "court": "The court consists of three circuit court judges or justices appoint- ed by the Chief Justice of the United States," 487 U.S. at 661 n.3; "we do not think it impermissible for Congress to vest the power to appoint independent counsel in a specially created federal court," id. at 676; "there is no risk of partisan or biased adjudication of claims regarding the independent counsel by that court," id. at 683; "once the court has appointed a counsel and defined his or her jurisdiction, it has no power to supervise or control the activities of the counsel," id. at 695. Were this language not clear enough, Morrison explicitly held that the Special Division's functions fall within the range of powers assigned to federal courts by the Ap- pointments Clause, see id. at 673-77, and by Article III, see id. at 677-85.

Morrison also acknowledged that the Special Division func- tions in a judicial capacity when exercising its referral power. Observing that "in order to decide whether to refer a matter to the counsel, the court must be able to determine whether the matter falls within the scope of the original grant," the Supreme Court said that the referral power involves "the power to 'reinterpret' or clarify the original grant." Id. at 685 n.22 (emphasis added); see also In re Espy, 80 F.3d 501, 507 (Spec. Div. D.C. Cir. 1996) ("In referring a related matter, this court is interpreting, but not expanding, the independent counsel's original prosecutorial jurisdic- tion...."). In other words, the referral function requires the Special Division to decide a legal question--i.e., whether the matters contained in an independent counsel's request for referral are "related to" the original jurisdictional grant. See id. at 507-09; In re Olson, 818 F.2d 34, 47-48 (Spec. Div. D.C. Cir. 1987). Answering this question calls on the Special Division--whether explicitly (as in Espy) or implicitly (as in this case)--to develop and apply a theory of "relatedness" consistent with the independent counsel statute and the Con- stitution.

Because the Special Division's January 1998 "related to" determination amounts to a legal judgment, we owe it no deference. We typically apply de novo review to decisions of other courts (except the Supreme Court) on questions of federal law, see Elder v. Holloway, 510 U.S. 510, 516 (1994) (questions of law must be resolved de novo on appeal), and "[w]hen de novo review is compelled, no form of appellate deference is acceptable," Salve Regina College v. Russell, 499 U.S. 225, 238 (1991). No one contends that the Special Division, as part of the D.C. Circuit, creates binding circuit precedent through its decisions. In relation to this court, the Special Division's legal determinations resemble those of our sister circuits, whose conclusions of law we review neither directly nor deferentially.

Acknowledging that referrals could be characterized as judicial acts, Judge Williams says that referrals most closely resemble the issuance of search warrants by federal magis- trates or state judges, whose determinations of probable cause, "[a]lthough made ex parte and resolving constitutional questions," are accorded " 'great deference' " under settled law. Maj. Op. at 11 (citing Illinois v. Gates, 462 U.S. 213, 236 (1983)). In Gates, however, the Supreme Court made clear that the deferential standard of appellate review applicable to search warrants stems directly from the intensely fact-bound nature of an issuing magistrate's probable-cause determina- tion. See 462 U.S. at 232 ("[P]robable cause is a fluid concept--turning on the assessment of probabilities in partic- ular factual contexts--not readily, or even usefully, reduced to a neat set of legal rules."); id. at 241 ("[P]robable cause deals 'with probabilities. These are not technical; they are the factual and practical considerations of everyday life on which reasonable and prudent men, not legal technicians, act.' ") (citation omitted). Unlike a magistrate, the Special Division does not "make a practical, common-sense decision" about probabilities when it considers requests for referrals. Id. at 238. Rather, it compares two legal documents--the independent counsel's referral request and the original grant of prosecutorial jurisdiction--and then determines whether the former is "related to" the latter within the meaning of the statute. Like a court evaluating a complaint on a motion to dismiss, the Special Division finds no facts, weighs no evi- dence, and makes no credibility determinations. It simply decides a question of law. Given the interpretive nature of this task, deference cannot be justified on the grounds "that the [Special Division] is 'better positioned' than the appellate court to decide the issue in question or that probing appellate scrutiny will not contribute to the clarity of legal doctrine." Salve Regina College, 499 U.S. at 233.

It is true that a referral under the independent counsel statute presents a situation "[w]here ... the relevant legal principle can be given meaning only through its application to the particular circumstances of a case." Miller v. Fenton, 474 U.S. 104, 114 (1985). But the Supreme Court has direct- ed that in such situations a federal appellate court must retain "its primary function as an expositor of law." Id. De novo appellate review is particularly important where, as here, the relevant legal principle has constitutional dimen- sions. See, e.g., Thompson v. Keohane, 516 U.S. 99, 107 (1995) (requiring federal habeas court to review de novo whether suspect was "in custody" at time of interrogation for purposes of Miranda); Miller, 474 U.S. at 112 (requiring "independent federal determination" of the voluntariness of a confession).

I disagree that referrals will have no real function under the statute unless we defer to the Special Division. See Maj. Op. at 11. For one thing, the Special Division's referral authority provides independent counsel, sensitive to both the limitations on their office and the ethic of self-restraint hon- ored by federal prosecutors, with an avenue for seeking independent and impartial confirmation of their authority. Immune from direct judicial review, a referral gives legitima- cy to an independent counsel's investigation even if a federal court later determines that a specific indictment exceeds the prosecutor's jurisdiction. This legitimacy provides a level of protection against any attempt by the Attorney General to remove or by Congress to impeach an independent counsel on the grounds that he has overstepped his authority. See 28 U.S.C. s 596(a)(1). The referral process also empowers the Special Division to enforce the boundaries of independent counsel jurisdiction. Where the Special Division rejects a request for referral, only the most zealous and imprudent prosecutor would pursue matters covered by the rejected application. Far from existing "simply to relieve the solitude of the Independent Counsel's office," Maj. Op. at 11, the referral power thus serves important functions under the statutory scheme, constraining independent counsel when they near the limit of their authority and safeguarding their political independence if their investigative authority is chal- lenged.

For all of these reasons, I would hold that this court owes no deference to the Special Division's "related to" determina- tion. In assessing the legality of this tax indictment, our review should be de novo.

III This court's conclusion that the Independent Counsel has authority to proceed with this indictment stems from its view, flawed in my judgment, that we would "undercut [Morrison]" were we to construe the independent counsel statute in ways "that prevent this Independent Counsel from performing his duty in a manner reasonably approximating that of an ordi- nary prosecutor." Maj. Op. at 20. Insisting on affording the Independent Counsel the same "leeway," id. at 15, given to "every other prosecutor," id. at 19, the court ignores the basic premise critical to the constitutionality of the statute: Inde- pendent counsel do not and cannot have the powers of ordinary prosecutors. If they did, the office would " 'imper- missibly undermine[ ]' the powers of the Executive Branch." Morrison, 487 U.S. at 695 (citation omitted). Among other constraints, it is the limited jurisdiction of independent coun- sel that "give[s] the Executive Branch sufficient control over the independent counsel to ensure that the President is able to perform his constitutionally assigned duties." Id. at 696. That an ordinary federal prosecutor might have authority to indict Hubbell for tax evasion therefore tells us nothing about whether this Independent Counsel--constrained by the stat- ute and the separation of powers principle--may bring the same indictment. Nor do we learn anything from the author- ity possessed by the Watergate Special Prosecutor, see Maj. Op. at 17, for the full scope of his authority flowed directly from the Attorney General, thus presenting no separation of powers concerns.

Reviewing the record de novo, I would find that the indict- ment of Hubbell for tax evasion was not "demonstrably related to the factual circumstances that gave rise to the Attorney General's investigation and request for the appoint- ment of the independent counsel." Morrison, 487 U.S. at 679. Hubbell's alleged failure to pay taxes on fees for work from 1994 to 1997 in Washington, D.C., is unrelated to whether, almost a decade earlier in Little Rock, Arkansas, "any individuals or entities have committed a violation of any federal criminal law ... relating in any way to James B. McDougal's, President William Jefferson Clinton's, or Mrs. Hillary Rodham Clinton's relationship with Madison Guaranty Savings and Loan Association, Whitewater Development Cor- poration, or Capital Management Services, Inc." Application for Appointment of Independent Counsel. The Attorney General who prescribed the original subject matter for inves- tigation agrees. According to her amicus brief:

[A]n offense cannot be 'related' within the meaning of [the statute] solely because an independent counsel dis- covers it during a legitimate phase of his investigation. That interpretation of the statute would allow an inde- pendent counsel to prosecute offenses that bear no rela- tionship to his original grant of jurisdiction.

Amicus Br. for United States at 32. "The current prosecu- tion," the Attorney General concludes, "does not fall within [the Independent Counsel's] authority directly to investigate the Whitewater/Madison Guaranty matter." Id. at 34.

In reaching the opposite conclusion, this court rests its finding of "relatedness" on two assumptions: that the fees Hubbell received were payments for his silence, and that his failure to pay taxes on them had an obstructive effect on the Whitewater investigation. In my view, both assumptions are flawed, and each independently undermines the statutory and constitutional constraints on independent counsel jurisdiction.

Assumption of obstructive effect I begin with the court's second assumption, i.e., that Hub- bell's failure to pay taxes on alleged hush money had an obstructive effect on the Whitewater investigation. Not limit- ed to Hubbell's failure to pay taxes on alleged hush money, the indictment's sheer breadth belies the court's obstruction theory. The indictment charges Hubbell with non-payment of taxes on income from a book contract with HarpersCollins Publishers, early withdrawals from an IRA account and pen- sion plan, and sales of his home and a Little Rock warehouse. In addition, by charging Hubbell's lawyer and accountant for aiding and abetting tax evasion, the Independent Counsel did not limit the indictment to individuals who could have ob- structed the Whitewater investigation.

According to my colleagues, "any criminal conduct that could hide the hush money ... tends to impede investigation and prosecution of the matter being hushed up." Maj. Op. at 16. "The less disclosure of the payments," they say, "the less chance that they and their nature will come to light...." Id. But even assuming that Hubbell's "consulting" fees were hush money (an assumption not supported by this record, see infra at 16-17), the facts of this case provide no support for the court's concealment theory. In his 1994 tax return, Hubbell actually disclosed $376,075 in "consulting" income that the Independent Counsel suspects is hush money. This amount included $100,000 from Hong Kong China Limited and $62,775 from Revlon, the two sources the court high- lights, see Maj. Op. at 4-5 & n.1, en route to finding that "[t]he timing, sources, and extent of the payments make the belief that they were hush money reasonable," id. at 20. Acknowledging these disclosures, the court rests its finding of concealment on the fact that Hubbell did not report an additional $77,000 in "consulting" fees in 1994. See Maj. Op. at 16. But since Hubbell disclosed the lion's share of the alleged hush money on his 1994 tax return--including the very payments that the Independent Counsel and my col- leagues believe have the strongest whiff of obstruction--it seems odd to think that Hubbell chose not to disclose the remaining fraction in order to conceal hush money payments. In any case, nothing in the record (beyond the non-disclosure itself) suggests that the $77,000 was hush money. See infra at 16-17. We thus lack any basis for suspecting that Hub- bell's non-disclosure reflected a concerted effort to hide hush money instead of a tendency (generally apparent from the indictment) to ignore the internal revenue laws.

The court's concealment theory makes much more sense in a case like United States v. Haldeman, 559 F.2d 31 (D.C. Cir. 1976), cited in Maj. Op. 17. There the defendants directly obstructed the Watergate investigation by lying under oath to the Special Prosecutor, a grand jury, and a Senate committee about hush money payments they had made to the Watergate burglars. See 559 F.2d at 59; see also United States v. Blackley, No. 98-3036, slip op. at 9 (D.C. Cir. Jan. 26, 1999) (upholding conviction of high-ranking Department of Agricul- ture official who concealed payments he received from indi- viduals regulated by the Department by failing to disclose those payments on an official financial disclosure form whose very purpose was "to bring suspicious influences to the surface").

The court next adopts the Independent Counsel's argument that Hubbell's failure to pay taxes "enhanc[ed] the financial or economic effect of the hush money payments.... And that contributes to the obstruction of the investigation." Oral Arg. Tr. at 16; see Maj. Op. at 16 ("[T]he more value Hubbell can squeeze from hush money ... , the more chance it will succeed in preventing his cooperation."). But even if we again assume the payments to be hush money, their obstruc- tive effect ended upon Hubbell's receipt; his subsequent non- payment of taxes bought his benefactors no further silence. The Independent Counsel never alleged that Hubbell's bene- factors somehow discounted the value of the alleged hush money by the probability that he would not pay taxes. In- deed, who could have foreseen the bizarre nature of the tax evasion scheme detailed in the indictment? Who would have expected that, having reported $376,075 of his $450,010 in "consulting" income on his 1994 tax return, Hubbell would then, as with his other income, not pay taxes on it?

The court's economic enhancement theory permits virtually unlimited expansion of the Independent Counsel's jurisdic- tion. Had Hubbell used the alleged hush money for profit- able but illegal gambling or insider trading, for example, this theory would allow the Independent Counsel to indict him for these crimes simply because they increased the value of the money. Surely this result stretches the concept of "related- ness" beyond its statutory and constitutional breaking point.

Together, the concealment and economic enhancement the- ories enable the Independent Counsel to comb through all of Hubbell's investments and expenditures--including, as the indictment reveals, his purchase of clothes, groceries, and laundry services--until discovering some illegality on which to indict him. By adopting these theories, the court converts the Office of the Independent Counsel from a device for investigating a specific "subject matter," 28 U.S.C. s 593(b)(3), into a tool for prosecuting a specific individual. As the district court observed, "[t]he Madison-Whitewater matters that were the subject of the Original Grant and the tax matters that are the subject of this case have nothing in common--nothing, at least, that appears on this record-- except Webster Hubbell." United States v. Hubbell, 11 F. Supp. 2d 25, 32 (D.D.C. 1998). This is precisely the result Justice Scalia, dissenting in Morrison, most feared:

[T]he most dangerous power of the prosecutor [is] that he will pick people that he thinks he should get, rather than cases that need to be prosecuted. With the law books filled with a great assortment of crimes, a prosecu- tor stands a fair chance of finding at least a technical violation of some act on the part of almost anyone. In such a case, it is not a question of discovering the commission of a crime and then looking for the man who has committed it, it is a question of picking the man and then searching the law books, or putting investigators to work, to pin some offense on him.

487 U.S. at 728 (Scalia, J., dissenting) (quoting then-Attorney General Robert Jackson, The Federal Prosecutor, Address Delivered at the Second Annual Conference of United States Attorneys (April 1, 1940)).

Deeply corrosive to the statutory and constitutional limits on independent counsel jurisdiction, the court's concealment and economic enhancement theories cannot justify a finding of "relatedness" in this case. In my view, this conclusion in and of itself is enough to sustain the district court's quashing of the indictment. But because the court's contrary holding also rests on the assumption that Hubbell's fees were hush money, I set forth my views on this issue as well.

Assumption that Hubbell's fees were hush money Even if the court's concealment and economic enhancement theories had merit, they have no applicability to this case for one simple reason: Although both depend entirely upon Hubbell's involvement in an underlying crime of obstruction, this record contains no credible evidence of such a crime.

In his brief, the Independent Counsel refers to the "possi- bility" that Hubbell "might have accepted money as an in- ducement to obstruct the Madison investigation." OIC Br. at 20 (emphasis added); see also id. (stating that "large pay- ments" on which Hubbell did not pay taxes "may be related to his non-cooperation with respect to Whitewater and Madison-related matters") (emphasis added). Yet the Inde- pendent Counsel has chosen not to indict Hubbell for accept- ing hush money payments or anyone else for making them. Although I agree with my colleagues that the Independent Counsel need not formally charge Hubbell for accepting hush money in order to indict him for tax evasion, I disagree that we may sustain the tax indictment simply because the Inde- pendent Counsel is "diligently pursuing his investigation of the main allegation that [hush] money was channeled to Hubbell," Wald Op. at 1. Without credible evidence that Hubbell accepted hush money, there can be no "demon- strabl[e] relat[ionship]" between Hubbell's tax crimes and the Independent Counsel's original grant of jurisdiction. Morri- son, 487 U.S. at 679 (emphasis added).

At oral argument, the Independent Counsel conceded that "we right now, as a matter of public record, don't know" that Hubbell's "consulting" income was hush money. Oral Arg. Tr. at 13. Notwithstanding this uncertainty, my colleagues have seen no need either to accept the Independent Counsel's offer to submit evidence in camera--evidence that he claims shows that the commission of that crime was more likely than not, see id.--or to remand to the district court to consider such evidence. Instead, they accede to his request that we not "indulge in the assumption that there is no evidence of obstruction," id., shoring up the Independent Counsel's innu- endo with non-record evidence plus a little innuendo of their own, see Maj. Op. at 4-5 n.1.

The record in this case is quite unlike the record in United States v. Haldeman, where this court upheld the convictions of H.R. Haldeman, John Ehrlichman, and John Mitchell for "conceal[ing] a cover-up" in the Watergate affair. Maj. Op. at 17 (emphasis omitted). In that case, the allegations that the defendants paid hush money to the Watergate burglars and then lied about it under oath were supported by vast amounts of credible evidence, primarily consisting of direct testimony and tape recordings of key conversations among the co-conspirators. See Haldeman, 559 F.2d at 55-59 (de- scribing defendants' "commitments" to pay the burglars hush money and detailing defendants' extensive conspiracy to raise, transfer, deliver, and conceal the hush money payments).

My colleagues say that independent counsel should enjoy the same discretion that U.S. Attorneys have to charge defendants with "cover-up" crimes without charging them for the underlying crimes. See Maj. Op. at 18 (citing Depart- ment of Justice guidelines). But as I have pointed out, Hubbell's tax offenses cannot plausibly be labeled "cover-up" crimes. See supra at 13-14. Moreover, the court's analogy is constitutionally flawed. It ignores the critical fact that U.S. Attorneys, unlike independent counsel, never need to prove that a particular crime falls within the jurisdiction ceded to them. They have authority to follow a trail of criminality and (subject to Justice Department procedures described below) to convert an investigation of real estate fraud into a prosecution of tax crimes. If the constitutional separation of powers precludes anything, it precludes the extension of such vast discretion to independent prosecutors who lack "the unifying influence of the Justice Department" and "the perspective that multiple responsibilities provide." 487 U.S. at 732 (Scalia, J., dissenting).

IV Finally, in addition to adopting virtually limitless theories of "relatedness" and assuming that Hubbell accepted hush money, my colleagues discount the constitutionally significant requirement that independent counsel "comply with the writ- ten or other established policies of the Department of Justice respecting enforcement of the criminal laws" unless "to do so would be inconsistent with the purposes of the [Act]." 28 U.S.C. s 594(f)(1); see Morrison, 487 U.S. at 696. Under Department of Justice Tax Division Directive 86-59, all feder- al prosecutors, expressly including independent counsel, must follow certain procedures before "seeking to expand nontax grand jury investigations to include inquiry into possible federal criminal tax violations." Authority to Approve Grand Jury Expansion Requests to Include Federal Criminal Tax Violations, 5 Dep't of Justice Manual 6-227, 6-227 (1995-1 Supp.). This directive prevents prosecutors from targeting an individual for tax crimes when their investigative authority extends only to a non-tax criminal investigation. Specifically, it requires all prosecutors to submit written requests to officials at the Internal Revenue Service and Tax Division "containing pertinent information relating to the alleged fed- eral tax offenses." Id. at 6-228. Approval requires an IRS referral certifying that "there is reason to believe that federal criminal tax violations have been committed." Id. at 6-229. In addition, prosecutors must conduct grand jury proceedings "in sufficient time to allow the results of the tax segment of the grand jury proceedings to be evaluated by the Internal Revenue Service and the Tax Division before undertaking to initiate criminal proceedings." Id.

In this case, I see no reason why the Independent Counsel could not have complied with Tax Directive 86-59. My colleagues are willing to assume that requiring such compli- ance would be "inconsistent with the purposes" of the inde- pendent counsel statute, 28 U.S.C. s 594(f)(1), attributing to the IRS--merely because it is an executive branch agency--a conflict of interest so severe that it cannot make a reasoned judgment as to whether Hubbell should be prosecuted for tax crimes. They apparently view compliance with Justice De- partment procedures as categorically "inconsistent with the purposes" of the statute whenever such procedures submit federal prosecutors to oversight by executive branch agencies. Surely section 594(f)(1) is not so toothless. This provision, which at the time of Morrison required compliance "except where not possible," 28 U.S.C. s 594(f) (Supp. V 1983) (amended 1994), was viewed by the Supreme Court as one of the statutory safeguards ensuring that the Executive Branch has enough control over independent counsel to accomplish its constitutionally assigned functions. See Morrison, 487 U.S. at 696.

V In sum, I believe that the court's finding of "relatedness"-- premised on deference to the Special Division, unbounded theories of obstruction of justice, and unsubstantiated suspi- cions of illegality--undermines the constitutional constraints on independent counsel jurisdiction. Morrison endowed the statute's jurisdictional controls with constitutional significance because it recognized that the absence of such controls would enable independent counsel to usurp executive power. This case provides reason to worry that the Office of the Indepen- dent Counsel indeed functions as a "mini-Executive ... oper- ating in an area where so little is law and so much is discretion." Morrison, 487 U.S. at 732 (Scalia, J., dissenting). Because the constitutional separation of powers demands greater vigilance, I believe that this Independent Counsel has exceeded his jurisdiction.

My conclusion would not impair the Ethics in Government Act's "central purpose ... [of] permit[ting] the effective investigation and prosecution of high level government and campaign officials," United States v. Wilson, 26 F.3d 142, 148 (D.C. Cir. 1994), quoted in Maj. Op. at 15. This Independent Counsel's original grant gives him all the authority he needs to prosecute Hubbell and others for obstruction of justice if he has evidence that Hubbell received hush money. See 28 U.S.C. s 593(b)(3). If he wants to prosecute Hubbell for self- standing tax offenses discovered in the course of his investi- gation, he can ask the Attorney General to expand his investi- gative jurisdiction, see id. s 593(c), and then set in motion Justice Department procedures for prosecuting criminal tax violations, see supra at 18. As the subsequent indictment of Hubbell illustrates, see Neil A. Lewis, Starr Indicts Hubbell a 3d Time, N.Y. Times, Nov. 14, 1998, at A1, the Independent Counsel, short of pursuing this tax indictment, retains vast investigative and prosecutorial authority under the Special Division's January 1998 referral.

I respectfully dissent.

Williams, Circuit Judge, dissenting from Part II: The commentator who predicted that Fisher and Doe would "inev- itably lead" to "metaphysical speculation" was apparently all too prescient. See Samuel A. Alito, Jr., "Documents and the Privilege Against Self-Incrimination," 48 U. Pitt. L. Rev. 27, 59 (1986). The majority opinion supplies some such specula- tion and demands more from the district court on remand. I would limit the district court's inquiry about the subpoenaed documents to verifying that the Independent Counsel, in securing Hubbell's indictment, has only used information that he would have had if the documents had appeared in his office, unsolicited and without explanation.

* * * It is clear that a prosecutor who has obtained personal documents by subpoena may not, without violating either the Fifth Amendment or a use immunity of equivalent scope granted under 28 U.S.C. s 6002, use against the subpoenaed person any testimonial, incriminating information that is com- municated by that person's "act of production" of the docu- ments. See Doe v. United States, 487 U.S. 201, 209 (1988) ("Doe II"). The Supreme Court identified three species of information possibly communicated by such an act of produc- tion--possession, authentication and existence. See id.

Here the only interesting issue is the "existence" theory; possession and authentication seem properly outside the case. Hubbell's prior possession is irrelevant if, as appears to be the case, the Independent Counsel relied on the documents only for the information that they contain, and thus had no occasion to rely on Hubbell's act of production, or anything else, for evidence that Hubbell at one time had "possessed" them. Nor does it appear that he used the production for authentication; he never sought to show the grand jury that Hubbell, by delivering the documents in response to the subpoena, had identified them as being ones that matched the descriptive language of the subpoena.

Thus we are left with "existence." From the truism that the Independent Counsel could not use the contents of the documents unless they (at some time) existed, and unless he learned of that existence, the majority leaps to the proposi- tion that the Independent Counsel's awareness of their exis- tence stems from a testimonial aspect of Hubbell's act of production. Accordingly, it says, the Independent Counsel may use the information in the documents if but only if he can show that he possessed, before securing the subpoena, a knowledge of the documents' existence sufficiently detailed that his later knowledge, after their delivery, was a "foregone conclusion."

But not all aspects of the act of production are testimonial. Where an item of information that the prosecutor receives from a document delivery flows from a non-testimonial as- pect, he does not depend on any testimonial aspect. Informa- tion as to the existence of the pieces of paper turned over by a subpoenaed party can always be traced to non-testimonial information. I elaborate below.

* * * "Testimonial." Before the Fisher Court introduced the "foregone conclusion" discussion on which the majority is so focused--and which I discuss below--it observed that the whole issue of whether something is "testimonial" depends on the facts and circumstances. Fisher v. United States, 425 U.S. 391, 410 (1976). But what facts and circumstances are relevant?

One possibility might be that all actions from which we can glean information are considered testimonial communications for purposes of Fifth Amendment analysis. But the prece- dents upon which Fisher relied in more or less rejecting the view of Boyd v. United States, 116 U.S. 616 (1886), that the Fifth Amendment protects the contents of subpoenaed docu- ments, appear to rule this out. Those cases involve the government forcing a person to try on a blouse worn by the perpetrator to establish whether it fit the defendant, Holt v. United States, 218 U.S. 245 (1910), or to give blood samples, Schmerber v. California, 384 U.S. 757, 764-65 (1966), voice samples, United States v. Wade, 388 U.S. 218, 222-23 (1967), or handwriting samples, Gilbert v. California, 388 U.S. 263, 266-67 (1967). They do not rely on anything like the "fore- gone conclusion" rationale; instead, they find that such acts are not testimonial because they fit into the category of "compulsion which makes a suspect or accused the source of 'real or physical evidence.' " Schmerber, 384 U.S. at 764.

Nor can these cases be recharacterized as ones where the prosecutor's grasp of the information obtained was a foregone conclusion. Of course it is true that, for example, it is typically not much to admit that one can speak. But in giving a voice sample, one also admits that one's voice has various characteristic idiosyncrasies--a non-obvious and incriminating fact that the law allows the prosecutor to secure by compul- sion. The prosecutor's and jury's access to that information is as dependent on the speaker's compelled implicit admission of ability to speak as their access to the information on documents is dependent on the subpoenaed party's implicit admission of the documents' existence.

One can, of course, discern a communicative element in the giving of a voice sample: a person commanded to speak implicitly says, "This is the way I sound when I speak." But that information adds nothing to what a jury learns from its own ears (or from a properly authenticated tape, if that is the way it is done, see, e.g., United States v. Dionisio, 410 U.S. 1 (1973) (grand jury subpoena requiring suspects to read tran- scripts into a recording device is consistent with Fifth Amendment)). Similarly, a person giving a blood sample implicitly says, "This is my blood." Though there is implicit communication, the prosecutor need not rely on it, so long as he has the blood and a witness to the blood-giving itself.

"Foregone Conclusion." The most confusing part of Fisher is the language that the courts have taken to tie "foregone conclusion" closely to the "testimonial" analysis and vice versa. 425 U.S. at 411. The Court said: "Surely the Govern- ment is in no way relying on the 'truthtelling' of the taxpayer to prove the existence of or his access to the documents. The existence and location of the papers are a foregone conclu- sion...." Id. (citation omitted). In my view, the latter sentence should be read in light of the former. That is, "foregone conclusion" is only a subset of the broader set: instances where sources independent of testimonial aspects of the compulsion fully account for the prosecutor's evidence.

This relationship is illustrated in an example used in Fish- er: "When an accused is required to submit a handwriting exemplar he admits his ability to write and impliedly asserts that the exemplar is his writing. But in common experience, the first would be a near truism and the latter self-evident." Id. The Court here is implicitly referring to Gilbert v. California, 388 U.S. at 265-67, one of the cases it had just relied upon in more or less overruling Boyd. In Gilbert, the police got the suspect to write out some handwriting exem- plars while he was in custody and being questioned. When the Court calls the implicit admission of ability to write a "near-truism," the "near" is critical. Consider a kidnapping, in which a ransom note is a major piece of evidence, but the suspect claims illiteracy. Suppose police, posing as terrorists, frightened him into writing something (the text of which had no bearing on the kidnapping). There might be some sort of due process argument, but in using the handwriting sample the prosecutor would not need to rely on any implicit testimo- nial aspect of the scenario ("Yes, I do know how to write."); accordingly he could use the sample without violation of the defendant's privilege against self-incrimination.

More important is the defendant's implicit admission that the exemplars are his. This is not, strictly speaking, "self- evident"; rather, it is supported by evidence (the testimony of the witnessing police officers) of a non-testimonial aspect of the act of production, here the act of writing. But as the Court says, the government is "in no way relying on the 'truthtelling' of [defendant]" to prove anything: it is relying only on the immediate personal observations of the policemen and on nontestimonial aspects of the defendant's act to link the handwriting to the defendant. Everything else of eviden- tiary value, namely the idiosyncrasies of the writing, depends only on the writing itself. As we'll see more explicitly below, this matches the relation of a prosecutor to documents deliv- ered pursuant to subpoena. The information on the docu- ments stands or falls on its own value, even though (by definition) produced by the defendant's act of production. Thus: for handwriting, the link to the defendant is estab- lished by police witnesses and the testimonial value of the defendant's act of production is redundant; for documentary information, so long as the prosecutor depends as here only on information in the documents themselves for the link to defendant, the communicative aspect of the act of delivery is equally redundant.

Existence. In light of the above, the only sense of "exis- tence" that is covered by the Fifth Amendment is that which refers back to the subpoena. The responsiveness of the documents to the subpoena gives knowledge of "the existence of the papers demanded," 425 U.S. at 410 (emphasis added). "Yes, these are the records you described in the subpoena." If the government could refer back to the subpoena to identify documents and to clarify relationships that were not clear on their face or by other independent means, then it would be using a testimonial component of the transaction-- the witness's implicit statement that the documents match the subpoena's description. Hubbell's claim for blanket exclusion of the contents, by contrast, relies on existence in a quite different sense--the fact that these particular pieces of paper are in being. But this is quite easily confirmed by these papers' own physical presence, which is "self-evident" at the time and place of production and so long thereafter as the government maintains proper custody. Existence in that sense is as "self-evident" as the blood and its characteristics in Schmerber, the voice samples and their characteristics in Wade, and the handwriting and its characteristics in Gilbert.

Some of the language in Fisher and Doe, to be sure, suggests a more sweeping view of "existence." Fisher I have discussed above. Doe upholds a decision quashing certain subpoenas, based on trial court findings (endorsed by the court of appeals) that delivery of the documents gave the prosecutor previously absent knowledge of their existence, possession, and authenticity. See United States v. Doe, 465 U.S. 605, 613-14 & nn.11-13 (1984). But the implications are quite unclear. The Court relied explicitly and entirely on the "two courts" rule. Id. While the majority argues that the findings below were structurally applications of law to fact, see Maj. at 34-35 n.24, the Court's treatment of them was as simple fact. Second, to the extent that its rehearsing of the arguments embraced in the courts below may suggest the sort of "existence" theory employed by the majority, the inference is drawn in question by the Court's reliance on the anticipated use of the act of production for authentication of the documents, i.e., use of an indisputably testimonial aspect of subpoena compliance. See 465 U.S. at 614 n.13.

Accordingly, the logic of Fisher and Doe, if not every phrase, clearly supports the prosecutor's right to use informa- tion from subpoenaed documents regardless of whether he was previously able to describe them. The particular docu- ments' existence speaks for itself once they have been deliv- ered; so long as his use of them is independent of any testimonial aspects of the witness's act of production, that use is consistent with the witness's Fifth Amendment privi- lege.

* * * The majority confuses the issue with a rather odd distinc- tion: if compulsion "acts upon, and requires the exercise of an individual's mental faculties for communication," it is testimo- nial; if it "merely utilizes the body of the accused as a form of evidence," it is not. Maj. at 39.

To the extent that the majority here acknowledges that the bare physical aspects of a production--the meanings that are directly apparent to the senses--are unprotected by the Fifth Amendment, it is correct. But there is no reason to restrict this to the "body of the accused."

In fact, the majority fails to explain the cases that fall outside this apparent restriction. Gilbert v. California, ap- provingly cited by the majority, concerned a handwriting sample that the suspect had to write out and then turn over to the police. But more telling is Baltimore Dept. of Social Servs. v. Bouknight, 493 U.S. 549, 554-55 (1990). In that case, a woman was ordered to turn over a child whom she was believed to have abused and who was last seen in her custody. The Court held, among other things, that the woman "cannot claim the privilege based upon anything that examination of [the child] might reveal."

Thus in Bouknight it was not the body of the accused that was used as a form of evidence, but the body of another. Documents are exactly analogous: physical objects the exami- nation of which yields evidentiary value or clues. Documents do, of course, represent the concrete embodiment of mental activity, but that is a false lead: these thoughts (the contents of the documents) were, we assume, put to paper quite voluntarily--if they were not, they would unquestionably be protected. See Doe, 465 U.S. at 610-11. In Bouknight the Court was obviously indifferent to the necessity that the suspect find and turn over a child whose location was com- pletely unknown to the government; the court here should be equally indifferent to the necessity that Hubbell do the same for documents.

Delivery of the child in Bouknight clearly depended on the suspect's exercise of her mental faculties; in fact, her intellec- tual efforts turning up the child were no less then they would have been had the government known its whereabouts in advance. The case thus flatly contradicts the majority idea that self-incrimination occurs whenever the subpoena "re- quires the exercise of an individual's mental faculties." The majority might say, with internal consistency though not with conformity to the cases, that the government may use the product of forced mental exercise so long the mental exercise is no more than an automaton's execution of intellection already carried out by the government. But that would take it to the position that a document subpoena must itself set forth whatever descriptive detail is necessary (under the majority's murky test) about the documents' character and location. Even the majority evidently recoils at this absurdi- ty.

There are of course non-physical aspects to the production in Bouknight. In another part of that opinion, the Court used a different analysis for the suspect's "implicit communi- cation of control over [the child] at the moment of produc- tion," 493 U.S. at 455, saying that although this was arguably an incriminating "testimonial assertion," id., some uses of it might be permissible under the doctrine that the Fifth Amendment may not be invoked to resist compliance with certain types of regulatory regimes, id. at 555-62. Here, of course, the government has no interest in Hubbell's control of the documents at the moment of production, and seeks to draw no inferences from that control. But the majority's concern here with information about the documents before their delivery is utterly different from the Bouknight Court's focus on "the moment of production."

* * * The majority's confusion is further evident in its attempt to draw some distinction between whether something is "testi- monial" and whether it has "testimonial value." See Maj. at 42 n.27, 45-46 n.31. This dissent, the majority argues, wrongly frames the former rather than the latter issue as the key. Putting aside such issues as whether this terminological difference makes sense--if something is not "testimonial," its "testimonial value" is obviously zero--or is to be found in the cases--Fisher, for example, simply refers to "the more diffi- cult issue[ ]" of "whether the tacit averments ... are [ ] testimonial," 425 U.S. 410--the actual analysis endorsed by the majority is not much different from the analysis in the "Foregone Conclusion" section above. The majority writes:

Where the government need not rely upon the truthtell- ing of the witness, because it has prior knowledge of the information that will be communicated through the act of production, 'no constitutional rights are touched.'

Maj. at 46 n.31 (emphasis added) (quoting Fisher). The only disagreement here is the italicized portion: for some rea- son, the majority believes that possession of "prior knowl- edge" is the only circumstance in which the government "need not rely upon the truthtelling of the witness." But the majority never explains how, under its theory, there is no testimonial self-incrimination if the government need not rely because it already knows, while there is testimonial self- incrimination if the government has another reason for dis- pensing with reliance on communicative aspects of the wit- ness's acts. Nor could any such explanation be consistent with precedent. In Bouknight, which concerned a subpoena to turn over a missing child, the Court found that the target could not "assert the privilege upon the theory that compli- ance would assert that the child produced is in fact [the child sought]" because that fact was one "the State could readily establish," 493 U.S. at 555, despite the fact that the govern- ment could not have made such a finding until after the production.1 Here, too, the government need not rely on any communicative or testimonial aspect of Hubbell's act of pro- duction; once it acquired the documents, their intrinsic value evidently served its purposes quite adequately. The majority imposes a wholly artificial and impermissible limitation on the reasons for which the government "need not rely" on testimo- nial implications of the act of production.

* * * For a district judge, the challenge of the majority's view is to determine the "quantum" of relative prosecutorial igno- rance that triggers a self-incrimination violation. Prosecutors know that businessmen keep business records (just as they know that living humans have blood and literate persons have handwriting); this is plainly too little information for the majority. But evidently the prosecutor need not have ad- vance knowledge of the details that he is interested in. See Maj. at 54. Somewhere in that range is an imaginary line

* * * Sensibly construed, the act of production doctrine shields the witness from the use of any information (resulting from his subpoena response) beyond what the prosecutor would receive if the documents appeared in the grand jury room or in his office unsolicited and unmarked, like manna from heaven. See DOJ Amicus Br. at 42; Alito, supra, at 59-60. The prosecutor would in such a case not be able to identify, verify someone's control over, or authenticate the documents except to the extent their own contents--or other sources-- did so. He would thus make no use of any testimonial aspect of the act of production. Yet, like DNA and handwriting idiosyncrasies, the contents would themselves be unprotected, except to the extent that deciphering might depend on the context of the subpoena--the information conveyed by the suspect's implicit matching of them with the subpoena de- scription.

This distinction between contents and production is appar- ently missed by the majority. Its first hypothetical of the murder weapon claims that this "manna from heaven" theory would allow the government to compel a suspect "to incrimi- nate himself verbally" by revealing the location of the murder weapon. But in the majority's hypo, the weapon is obviously the fruit of poisoned testimony: a revelation under compul- sion. A more apt instance would be if a suspect had previous- ly--without compulsion--written down the location in his day planner, and the government subpoenaed the planner. The production of the day planner, like the production of a missing child, is compulsory but non-testimonial. The much more harmful contents are obviously testimonial, but they are not the fruit of any unlawful compulsion (so long as the government's use is independent of the context of the subpoe- na).2