(concurring in part and dissenting in part):
In concur in Part I but dissent from Part II of the majority opinion. My concern is two-fold: 1) the majority misreads, in my view, the import of the Supreme Court’s decision in Couch, and 2) I fail to understand the rationale by which the majority reaches its decision. The majority states:
“If ownership vel non were the test of the scope of the Fifth Amendment privilege, our analysis would be at an end. But such is not the test, for the Fifth Amendment privilege focuses on personal compulsion upon the person asserting it. Possession, not ownership, ‘bears the closest relationship to the personal compulsion forbidden by the Fifth Amendment.’ Couch v. United States, supra, 409 U.S. at 331.” Majority opinion at 689.
The majority also concedes that in certain situations a person may divest himself of actual possession, but yet retain constructive possession so as to leave the personal compulsions upon the accused substantially intact. Couch v. United States, 409 U.S. 333-335 (1973). Thus possession (either actual or constructive), not ownership, bears the most significant relationship to the fifth amendment protections.
The majority’s recognition of these governing principles is unassailable and to this extent I am in complete agreement. I would add further, however, that Couch noted that “[i]t is extortion of information from the accused himself that offends our sense of justice.” “As Mr. Justice Holmes put it: ‘A party is privileged from producing the evidence, but not from its production.’ ” 409 U.S. at 328 (emphasis added). The fifth amendment forbids “[ijnquistorial pressure or coercion against a potentially accused person, compelling her, against her will, to utter self-condemning words or produce incriminating documents.” 409 U.S. at 329 (emphasis added). The incrimination comes from being forced “to produce and authenticate any personal documents or effects that might incriminate” you. United States v. White, 322 U.S. 694, 698, 64 S.Ct. 1248, 1251, 88 L.Ed. 1542 (1944), cited with approval in Couch, supra, 409 U.S. at 330 (emphasis added).
Couch therefore reaffirmed the previously accepted view of the Supreme Court that “[w]hat is incriminating about the production of a document in response to an order is not its contents, as one might have thought, but the implicit authentication that the document *695is the one named in the order.” 409 U.S. at 348 (Marshall, J., dissenting).1
With these principles in mind, I turn to the following statement by the majority, with which I am in full agreement:
“To be successful with [their argument] appellants must convince us of the validity of two propositions: First, if the Goldsmiths had not given the ‘analyses’ to Fisher, they could have successfully resisted the summons because the documents sought would have been in a rightful personally privileged possession, and, second, the Goldsmiths should not be held to have lost their privilege solely because they surrendered actual possession to Fisher for the purpose of obtaining legal advice in connection with the investigation.”
I believe both questions should be answered in the affirmative.
I.
The Goldsmiths could have successfully asserted the fifth amendment while the “analyses” were in their actual possession.
In my view the majority obscures this issue by dismissing Goldsmiths’ actual possession as “fleeting and transitory, limited to the act of delivery from the accountant’s office to the office of their lawyer at his request.” Besides being misleading,2 this statement is irrelevant. Nowhere in Couch did the Supreme Court suggest that the personal compulsion prohibited by the fifth amendment is dissipated whenever actual possession has been brief. If a subpoena had been served on the Goldsmiths while they were in actual possession of the documents, I fail to understand how the personal compulsion, which is the essence of the fifth amendment prohibition, can vary depending on the length of time the documents had previously been in Goldsmiths’ actual possession.
Couch suggested that what is significant is not length of possession, but the quality of possession, i. e., rightful possession in a purely personal capacity. The majority’s concern that actual possession was allegedly “fleeting and transitory” is misplaced. These and other similar adjectives were referred to in Couch solely with respect to the constructive possession issue, i. e., whether, inter alia, the divestment of actual possession (not the length of actual possession) is so “fleeting” or so “temporary and insignificant” that constructive possession is retained.
More relevant is the recognition by Couch that while Boyd did not “address or contemplate the divergence of ownership and possession,” 409 U.S. at 330, the subsequent Supreme Court decision in United States v. White 3 did address this issue:
“[T]he papers and effects which the privilege protects must be the private property of the person claiming the *696privilege, or at least in his possession in a purely personal capacity.” 409 U.S. at 330 n. 10 (Couch’s emphasis).4
Thus, the Supreme Court in White clearly suggested that something short of “private property” may be sufficient to invoke the fifth amendment. Couch not only approved this language in White, but also, as if to make its position unquestionably clear, cited with approval the Ninth Circuit decision in United States v. Cohen:
“See also United States v. Cohen, 388 F.2d 464, 468 (CA9 1967), where the court, in upholding the right of a possessor, non-owner, to assert the privilege, noted that ‘it is possession of papers sought by the government, not ownership, which sets the stage for exercise of the governmental compulsion which it is the purpose of the privilege to prohibit.’ Though the instant case concerns the scope of the privilege for an owner, nonpossessor, the Ninth Circuit’s linkage of possession to the purposes served by the privilege was appropriate.
“We do not, of course, decide what qualifies as rightful possession enabling the possessor to assert the privilege.” 409 U.S. at 330 n. 12.
In Cohen, ownership and possession diverged just as in White. Only in Cohen, unlike White, there was possession in a “purely personal capacity.” 5 The taxpayers’ refusal to produce the records was upheld on fifth amendment grounds. Thus, except for the constructive possession issue, Cohen is virtually identical to the case before us.
In view of the Supreme Court’s emphasis on possession as most relevant to the question of compulsion, I cannot believe that we should disregard Couch’s discussion of White and Cohen. It is difficult for me to understand, in view of the many courts which have wrestled with this problem, why the Supreme Court would choose to single out this language in Cohen, if it disagreed so completely with the Ninth Circuit’s application of these principles. Yet, this is seemingly what the majority must conclude.6
It is my view therefore that the approach taken by the Ninth Circuit in Cohen has been explicitly approved by Couch and comports fully with the fifth amendment view embraced by the Supreme Court, i. e., that possession, not ownership, is the significant factor and that the privilege protects one from having to produce the evidence, though not from its production.
Since the majority apparently does not dispute that the Goldsmiths had actual possession in a purely personal capacity,7 the one question which remains unclear is whether the Goldsmiths had rightful possession. The lower court as I read its opinion, found it unnecessary to decide whether the Goldsmiths had been in “rightful” possession because of its reading of United States v. Egenberg, 443 F.2d 512 (3d Cir. 1971).7a While the question of “right*697ful” possession involves a conclusion of law, I would not deem it appropriate to draw a conclusion on this score from the limited factual findings below. As the Court in Cohen indicated, it is one thing for an accountant to relay to his client the existence of a demand of the government and to say, in effect, “I will produce them if you have no objection and will return them.” It is quite another thing for an accountant to say “I want to produce them and demand their return.” Yet either situation would be consistent with the facts found by the district court in this case. Accordingly, unless it can be said that the transference of the “analyses” to attorney Fisher bars the assertion of the Goldsmiths’ fifth amendment privilege, I would remand to permit the lower court, with or without additional testimony at its discretion, to make additional findings of fact and conclusions of law.
II.
Notwithstanding the transference of the “analyses” to their attorney, the Goldsmiths retained constructive possession enabling them to assert their fifth amendment privilege.
The Supreme Court in Couch expressly rejected a per se rule that only actual possession would enable one to successfully assert the fifth amendment privilege.8 The Court noted that “situations may arise where constructive possession is so clear or the relinquishment of possession is so temporary and insignificant as to leave the personal compulsions upon the accused substantially intact.” 409 U.S. at 333.
Neither in Couch nor in any other case called to our attention has the Supreme Court engaged in such an analysis in the context of an attorney-client relationship.9 The lower federal courts which have considered the matter have reached varying results.10 The government contends that the presence of such a relationship is irrelevant because appellants, at this stage, disavow any reliance on the attorney-client privilege. While this contention accurately reflects the position of appellants before us,11 I am compelled to disagree with the government’s conclusion. Where there was a temporary surrender of actual possession pursuant to an attorney-client relationship, the nature of that relationship is of crucial importance in determining what a citizen has relinquished by the transfer and what he has retained.12
In Couch the taxpayer had surrendered actual possession to her account*698ant in the regular course of a long term business relationship. The accountant was an independent contractor engaged to perform a service; the service was the preparation of tax returns. The purpose of the surrender involved public disclosure. Justice Powell’s opinion stresses this fact and its consequences:
“In Boyd, a pre-income tax case, the Court spoke of protection of privacy, 116 U.S. at 630, 6 S.Ct. [524], at 532, but there can be little expectation of privacy where records are handed to an accountant, knowing that mandatory disclosure of much of the information therein is required in an income tax return. What information is not disclosed is largely in the accountant’s discretion, not petitioner’s. Indeed, the accountant himself risks criminal prosecution if he knowingly assists in the preparation of a false return. 26 U.S.C. § 7602(2). His own need for self-protection would often require the right to disclose the information given him.”
In the case before us, the Goldsmiths surrendered the analyses to their attorney for the limited purpose of securing legal advice not with respect to the preparation of a tax return but rather with respect to a pending tax investigation.13 The Goldsmiths’ expectation was of cloistered scrutiny and consultation, not of public disclosure. They had the unqualified right to immediate possession ;14 they had the unqualified right to command and Fisher had the unqualified duty to obey.15 No disclosure or other disposition of the papers could properly be made by Fisher without the Goldsmiths’ approval.16 In short, the Goldsmiths did nothing more than temporarily relinquish physical possession.
The government conceded in oral argument that the Goldsmiths would have retained the right to assert the privilege if they had stayed in Fisher’s office while he examined the papers in their presence or if they had insisted on his reviewing the papers in their home. I believe such distinctions ignore the realities of the relationship between a client and his attorney and are wholly unrelated to the purposes of the fifth amendment privilege.
I think this is demonstrated by the logic of Judge Jertberg, writing for the majority in United States v. Judson, 322 F.2d 460 (9th Cir. 1963):
“Clearly, if the taxpayer in this case . had been subpoenaed and directed to produce the documents in question, he could have properly refused. The government concedes this. But instead of closeting himself with his myriad tax data drawn up around him, the taxpayer retained *699counsel. Quite predictably, in the course of the ensuing attorney-client relationship the pertinent records were turned over to the attorney. The government would have us hold that the taxpayer walked into his attorney’s office unquestionably shielded with the Amendment’s protection, and walked out with something less. * *
“The government states that the evil which the Fifth Amendment sought to prevent is not present when the prosecution seeks evidence of A’s guilt from B. But this argument ignores the realities of the relationship existing where B is A’s attorney. An attorney is his client’s advocate. His function is to raise all the just and meritorious defenses his client has. No other ‘third party,’ nor ‘agent,’ nor ‘representative’ stands in such a unique relationship between the accused and the judicial process as does his attorney. He is the only person besides the client himself who is permitted to prepare and conduct the defense of the matter under investigation. The attorney and his client are so identical with respect to the function of the evidence and to the proceedings which call for its production that any distinction is mere sophistry. *- * -x- * -x-
“The very nature of the tax laws requires taxpayers to rely upon attorneys, and requires attorneys to rely, in turn, upon documentary indicia of their clients’ financial affairs. In light of these realities a very real danger would be created if we were to sustain the government’s position. That danger was apparent to Judge Murphy in Application of House, supra, when he spoke of ‘heavy penalties.’
“The government has at its disposal inquisitorial powers and administrative procedures which it may invoke at its pleasure. If the government’s position were sustained here, those powers could be utilized to stimulate a taxpayer’s consultation with his attorney and the predictable transfer of his records. The government’s powers could then be utilized to compel disclosure of those matters by the attorney whenever the taxpayer were not available to utter the magic words. In our judgment, the inherent power thus to compel indirectly an individual’s self-incrimination is curbed by the Fifth Amendment as effectively as the power to compel the same result directly.16a
In my view, it is difficult to imagine a more compelling case of constructive possession where the personal compulsions upon the accused are left substantially intact.
III.
I am thus left with a perplexing question which continues to linger in my mind after reading the majority opinion: what is the rationale on which the majority bases its decision ? As far as I can discern, the majority opinion neither concludes that the Goldsmiths were without rightful possession in a purely personal capacity nor concludes that the Goldsmiths were stripped of personal compulsions by the transference of the “analyses” to their attorney. While *700these are essentially the propositions which the majority initially suggests the appellants must sustain, the majority appears never to reach them.
As I read the majority, whether there was rightful possession in a purely personal capacity or whether the personal compulsions on the Goldsmiths were left intact notwithstanding the transference of the “analyses” are questions which need not be reached, if the documents possessed by the Goldsmiths did not “have the capacity of coming within the penumbra of the Boyd [‘private book and papers’] rule.”17 The majority states:
“Thus, if the taxpayers are to success in their effort, they must prove that their brief experience of actual possession for a limited purpose coupled with turning their accountant’s records over to their attorney has the legal capacity to generate a subsequent right of constructive possession of sufficient intensity to elevate those records into the required category of their ‘private books and papers.’ We are unwilling to attribute a Fifth Amendment protection to the accountant’s work product based on such a limited possession by his client.” 18
If this is a statement of the holding in this case, I find it unhelpful and fear that it will create uncertainty for lower courts seeking to interpret it.
Does the majority hold that “personal compulsion,” which they recognize is forbidden by the fifth amendment, is dissipated when actual possession is only “limited” or “brief”? Does the majority hold that what is significant for fifth amendment analysis is the Goldsmiths’ intent, i. e., the “limited purpose” for which they obtained the “analyses”?
Does the majority hold that because the “analyses” were not protected by the privilege while in the possession of the accountant, the Goldsmiths’ fifth amendment privilege was forever lost because they could no longer have reasonable expectations of privacy in the “analyses”? These questions have immense ramifications for purposes of fifth amendment analysis.
Lastly, as a constitutional standard, I find the following statement by the majority vague and obscure, as well as at variance with Couch:
“possession of sufficient intensity to elevate those records into the required category of their ‘private books and papers.’ ”
The majority initially recognizes that the privilege may apply where “books and papers” are “possessed” rightfully in a purely personal capacity.19 White, Couch and Cohen, supra, support this view. Yet the majority appears unwilling to apply this rule, but rather reverts to the Boyd standard. Not only do I disagree with this approach, but I find it subject to various interpretations.
If by stating that possession must be of “sufficient intensity” to elevate the “analyses” to the level of “private books and papers,” the majority is requiring ownership of the “analyses,” 20 I believe Couch has now clearly decreed otherwise.
If by “private” the majority is referring to the origin of the papers at issue, I believe its reliance on Boyd is misplaced. In Boyd the privilege was sustained with respect to business invoices prepared and previously published by a party other than the citizen asserting the privilege.21
*701If by “private,” the majority is referring to “books and papers” in which a person has reasonable expectations of privacy, I again disagree. It is true Couch spoke of reasonable expectations of privacy, but it did so in a fourth amendment context solely with reference to whether the expectations of privacy accompanying the taxpayers’ transference of records to his accountant were such that the taxpayer retained constructive possession. Couch did not suggest that expectations of privacy were relevant to the issue of whether “books and papers” are possessed rightfully in a purely personal capacity. Moreover, Couch’s approval of Cohen supports this view. Lastly, it is far less evident to me, than the majority suggests without further elaborating, that there has been a “ 'shift in emphasis from property to privacy’ in the [Supreme] Court’s treatment of the Fifth Amendment in compelled production of documents.” 22 If anything, Couch represents, in my view, a shift from ownership to possession. In any event, in the absence of a fourth amendment claim, it simply is not readily apparent to me how prior Supreme Court cases justify this engrafting of privacy principles onto the fifth amendment privilege which is a protection against personal compulsion.23
I therefore respectfully dissent and conclude with this observation from Couch:
“The basic complaint of petitioner stems from the fact of divulgence of the possibly incriminating information, not from the manner in which or the person from whom it was extracted. Yet such divulgence, where it did not coerce the accused herself, is a necessary part of the process of law enforcement and tax investigation.” 24
The Supreme Court has recognized the need to obtain information in connection with tax investigations, but within limits. Once the government attempts to extract this information from the accused, the fifth amendment protections are violated. As the majority aptly warns, “[w]e must never wander from the principle that what the Fifth Amendment prohibits is compelled incrimination.” And in my view, this compulsion remains substantially intact even though Goldsmiths transferred the “analyses” to their attorney for purposes of legal representation.
I would therefore vacate and remand for a determination whether the Goldsmiths had rightful possession.
. In his dissent, Justice Marshall summarized the accepted view of self-incrimination as pertaining to the compulsory production of documents (although he did not agree with this view).
See also Douglas, J., dissenting, who stated,
“I can see no basis in the majority opinion, however, for stopping short of condemning only those intrusions resting on compulsory process against the author of the thoughts or documents.” 409 U.S. at 341.
. Goldsmith obtained the “analyses” from Berson in early August of 1971 (Berson estimated the date to be “[t]he 4th or 5th of August”) and turned them over to Fisher on August 17, 1971. The record therefore simply does not support the majority’s statement.
. 322 U.S. 694, 64 S.Ct. 1248, 88 L.Ed. 1542 (1944). White involved a union official who, having possession of union records, refused to honor a subpoena for their production on the grounds that they would incriminate him. The Supreme Court in White rejected this argument because the union official possessed the records in only an agency capacity.
. Couch’s emphasis of this language in White is particularly significant in view of several statements by the majority which suggests its reluctance to grapple with the italicized language in the text. See infra.
. In Cohen a subpoena had been served on the taxpayer who was in possession of workpapers owned by his accountant.
. Of course, if the majority finds there is no constructive possession, the Cohen question need not be reached. The majority, however, apparently professes to reach the Cohen question first:
“[t] he claim to the Fifth Amendment privilege must emanate from the taxpayer’s rights. And in the attorney-client relationship, the rights must flow upward from the taxpayer; not downward from the attorney.”
I agree with this statement only in the sense that constructive possession cannot be found unless there first is rightful actual possession in a purely personal capacity.
. It is undisputed that the “analyses” were prepared for the Goldsmiths from their own records and papers and pertained to their own business affairs. They obtained the “analyses” from Berson for their own personal purposes and did not hold them in furtherance of any objective of Berson, or as agent of Berson.
. See note 20 infra.
. 409 U.S. at 336 n. 20.
. While standing alone it would be a slender reed, indeed, the court did make a comment in Couch which can be read to indicate a material difference between the accountant-client and the attorney-client relationship:
“Technically the order to produce the records was directed to petitioner’s attorney since, after the summons was served upon the accountant, he ignored it and surrendered the records to the attorney. But constitutional rights obviously cannot be enlarged by this kind of action. The rights and obligations of the parties became fixed when the summons was served, and the transfer did not alter them.” 409 U.S. at 322, 329 n. 9, 93 S.Ct. 611, 616, 34 L.Ed.2d 548 (1973) (emphasis added).
. Compare United States v. Judson, 322 F.2d 460 (9th Cir. 1963), Colton v. United States, 306 F.2d at 633 (2nd Cir. 1962) and Application of House, 144 F.Supp. 95 (N.D.Cal.1956) with United States v. White, 477 F.2d 757 (5th Cir. 1973) (dissenting opinion). In re Fahey, 300 F.2d 383 (6th Cir. 1961) and United States v. Boccuto, 175 F.Supp. 886 (D.N.J.) appeal dismissed, 274 F.2d 860 (3d Cir. 1959).
. It is conceded by appellants that the lawyer-client privilege is inapplicable here. The authorities would seem to support the proposition that a document prepared by a tax accountant from bank records prior to the creation of an attorney-client relationship and given thereafter by the client to his attorney is not within the scope of the privilege. See, e. g., Bouschor v. United States, 316 F.2d 451 (8th Cir. 1963); Sale v. United States, 228 F.2d 682 (8th Cir. 1956); United States v. Kelly, 311 F.Supp. 1216 (E.D.Pa.1969).
. Justice Marshall, dissenting in Couch, stated:
“A transfer to a lawyer is protected, not simply because there is a recognized attorney-client privilege, but also because the *698ordinary expectation is that the lawyer will not further publicize what he has been given.” 409 U.S. at 350.
. Although the Goldsmiths have alleged on this appeal that they transferred the “analy-ses” for the limited purpose of securing legal advice with respect to the pending tax investigation (government does not dispute), the record is less specific. Although this presents no problem for me, at the very most a remand for such a finding would be necessary.
. The only qualification of this right of a client would appear to be an attorney’s lien. No suggestion is made that this. exception is applicable here.
. One could predict with some confidence that these realities of control would not be ignored in a case where a subpoena had been served on a taxpayer and he resisted production on the ground that he had given the records sought to his attorney. Cf. First National City Bank of N.Y. v. I.R.S., 271 F.2d 616 (2nd Cir. 1959); Hopson v. United States, 79 F.2d 302 (2nd Cir. 1935); United States v. Howard, 360 F.2d 373, 381 (3d Cir. 1966).
. As Judge Ainsworth noted in his dissenting opinion in United States v. White, 477 F.2d 757, 766 n. 6 (5th Cir. 1973) :
“The new ABA Code of Professional Responsibility in its Ethical Consideration 4-4 notes: ‘The attorney-client privilege is more limited than the ethical obligation of a lawyer to guard the confidences and secrets of his client. This ethical precept, unlike the evidentiary privilege, exists without regard to the nature or source of information or the fact that others share the knowledge.’ ”
. United States v. Judson, 322 F.2d 460, 466-468 (9th Cir. 1963). As Judge Jertberg also pointed out, the law has traditionally considered the client and his attorney as one in matters relating to the compulsory production of documents. As Professor Wig-more records:
“. . . [W] hen the client himself would he privileged from protection of the document, * .....' as exempt from self-incrimination, the attorney having possession of the document is nr '-, bound to produce. Such has invariably been the ruling. On the other hand, if the client would he compellable to produce * * * then the attorney is equally compellable, if the document is in his custody, to produce under the appropriate procedure.” (Original emphasis). 8 Wigmore, Evidence § 2307 (McNaughton Rev. 1961).
See also Colton v. United States, 306 F.2d 633, 639 (2nd Cir. 1962) and Brody v. United States, 243 F.2d 378, 387 (1st Cir. 1957).
. Majority opinion at 691.
. Majority opinion at 692 (emphasis added).
. Majority opinion at 689-691. ,
. I note particularly that the majority reaffirms United States v. Egenberg, 443 F.2d 512 (3d Cir. 1971), in a manner which may be read to suggest that Egenberg's “superior right” theory has survived Couch intact. If this is what the majority implies, I disagree. In my view Couch does not undercut the result in Egenberg but only because Egenberg involved possession in an agency rather. than personal capacity.
. As the Supreme Court noted in Wilson v. United States, 221 U.S. 221, 378, 31 S.Ct. 538, 543, 55 L.Ed. 771 (1911), “[w]here one’s private documents would tend to incrim*701inate him, the privilege exists although they were actually written by another person.” See note 1 supra (excerpt from Justice Douglas’ dissent in Couch).
. Majority opinion at 690.
. A careful reading of Couch reveals the distinction the Supreme Court maintained between the fourth and fifth amendment protections:
“We hold today that no [1] Fourth or [2] Fifth Amendment claim can prevail where, as in this case, there exists [1] no legitimate expectation of privacy and [2] no semblance of governmental compulsion against the person of the accused.” 409 U.S. 336.
. 409 U.S. at 329 (emphasis added).