concurring and dissenting:

The majority is on solid ground in ruling that the so-called “summaries” do not constitute privileged “work-product,” and that appellant may not invoke vicariously a Fifth Amendment right of her clients. Hickman v. Taylor, 329 U.S. 495 , 67 S.Ct. 385 , 91 L.Ed. 451 (1947), and Fisher v. United States, 425 U.S. 391 , 96 S.Ct. 1569 , 48 L.Ed.2d 39 (1976), are dispositive of those assignments.

I do entertain some doubt concerning this court’s refusal on a supposed jurisdictional ground to review the district court’s order refusing to strike the two inadvertently produced summaries. The nature of this proceeding, collateral as it is to the trial of Margolis, would seem to me to remove that order from the category of a step “towards final judgment in which [it] will merge” *214 (Cohen v. Beneficial Industrial Loan Corp., 337 U.S. 541, 546 , 69 S.Ct. 1221, 1225 , 93 L.Ed. 1528 (1949)), and to bring it within the purview of 28 U.S.C. § 1291 .

However, that small difference aside, I wholly disagree with the majority’s conclusion that on this record the attorney-client privilege does not operate to shield the summaries from production and disclosure.

The attorney-client privilege, as I understand it, is that of the client and is intended to protect from public disclosure confidential communications he makes to his attorney in connection with legal matters. Necessarily, the privilege is not narrowly applied, for it would be a poor thing indeed if an attorney could be compelled to disclose the advice given his client in response to such a communication, for the essence of the communication would necessarily inhere in the response and be fairly deducible from it.

Here we are not concerned with a recalcitrant attorney-witness but with documents prepared by the attorney. If the issue were as the majority postulates, simply “whether summaries of a client’s business transactions with third persons compiled by an attorney from unprivileged facts are entitled to the protection of the privilege,” I would certainly join in voting to affirm. But the record does not provide such an easy springboard. The summaries are constituted of a melange of facts. Appellant testified before the district court that the information from which the documents were compiled was derived from manifold sources: from communications by the client, from co-counsel, from public records, and in part from calculations made by counsel on the basis of such composite information and not reflected in any other document. The documents sought by the government are thus amalgams as far as the source of their content is concerned: in part derived from communications from the client intended to be confidential, in part from public, non-confidential sources.

Of course, as the majority points out, it is not enough that an experienced tax planner might be able to “glean from the summaries sufficient information to enable him to understand the general tenor of the tax advice given by appellant.” If that were the extent of the deduction, then the privilege, being that of the client, would not be violated. But I suggest the “experienced tax planner” could most likely deduce much more. Advice in the abstract in tax matters is not the norm. It most generally is based upon specific information.

In my view, a document need not, as the majority declares, “irresistibly” disclose a confidential communication in order to be protected by the privilege; if it is such as to reasonably permit such a communication to be deduced, I would honor the privilege and protect the client.

I would reverse.