Describing the Withheld Material

Rule 26(b)(5)(A)(ii) adds a second duty to the express claim. A party that withholds otherwise discoverable information must “describe the nature of the documents, communications, or tangible things not produced or disclosed,” and must do so in a manner that, “without revealing information itself privileged or protected, will enable other parties to assess the claim.”¹ The duty is mandatory. One court described the obligation to log privileged documents as “mandatory under the specific terms of Rule 26(b)(5).”²

The express claim announces that something is being held back and names the protection invoked. It gives the requester nothing to test, because the label “attorney-client privilege” fits a protected memorandum and a routine business e-mail equally well. The description supplies the missing facts about each withheld item, so that the requester, and eventually a judge, can decide whether the label fits. The usual vehicle is the privilege log, which lists the withheld items and describes each one.

Three subsections follow. The first explains how the rule balances disclosing too little against disclosing too much. The second lists what a log entry should contain. The third takes up e-mail chains and attachments, where the choice of what to list as one item decides whether the claim can be assessed at all.

Assessing Without Revealing

The rule’s single sentence gives two instructions that pull against each other. The description must enable other parties to assess the claim, which sets a floor. It must not reveal the protected information, which sets a ceiling. Every log entry has to clear both.

The ceiling follows from what a privilege protects. The attorney-client privilege shields the content of a confidential communication between lawyer and client, and work-product protection shields a lawyer’s preparation of the case; neither means much once the opponent has read the material. A rule that made a party prove its claim by displaying the document would destroy the privilege in the act of claiming it. The rule asks instead for the nature of the item. A proper entry reports facts about the document, such as its kind, its date, who was involved, and its general subject, and it leaves the advice, the strategy, and the mental impressions inside undisclosed.

The floor follows from who holds the knowledge. Only the withholding party has seen the documents, and a privilege claim rests on facts about each one: who wrote it, to whom, when, and about what. A bare conclusion such as “privileged” gives the requester nothing to test and gives a court nothing to review; it would make the party the unreviewable judge of its own claim. Courts accordingly decline to credit thin descriptions. A local rule quoted in one opinion directs a party asserting privilege to disclose the type of document, its general subject matter, its date, and information “sufficient to identify the document for a subpoena duces tecum,” including the author, the addressees, and their relationship to one another. The court added that failing to include sufficiently descriptive information may result in a finding that the privilege has not been established.³ Each item on that list describes the document from the outside, and none requires disclosing the advice within it. Under Massachusetts Rule of Civil Procedure 26(b)(5), a trial judge reached the same kind of result. The defendants had not carried their burden, the judge found, because they neither produced the privilege log the rule requires nor otherwise described the documents “so as to permit a determination of whether or not they are privileged.”⁴

The ordinary way to meet both instructions is a log, and a party that serves none leaves the court holding only its say-so. In Williams v. Sprint/United Management Co. the defendant claimed that the metadata in its electronic spreadsheets would reveal privileged communications or attorney work product, yet it provided no privilege log for those electronic documents. The court held that the defendant had waived attorney-client and work-product protection for the metadata, except for metadata directly corresponding to the adverse impact analyses and the social security number information that the court had permitted the defendant to remove.⁵ The waiver consequence has its own section below. The point here is that a bare assertion of risk, with no description at all, falls below the floor.

Court-ordered procedures put the log in the same place. One protocol for reviewing data extracted from a computer required defense counsel, within twenty days after delivery of the extracted files, to hand over all nonprivileged material together with a privilege log that identified each document withheld and described its nature “without revealing information which is itself privileged,” so that the plaintiff could assess the privilege.⁶ Another court, noting that neither side’s privilege objections came with the required log, described it as a list that describes the pertinent documents without disclosing the allegedly privileged communications they contain.⁷ Both treat the log as the single device that meets the floor and the ceiling together.

Hypothetical Illustration

A company withholds one e-mail, dated March 3, from its general counsel to its chief engineer. Its log could describe the item in any of three ways. Entry one reads “Document 14: privileged.” Entry two reads “Document 14: e-mail dated March 3 from the general counsel to the chief engineer, no other recipients, concerning the legal consequences of the licensing agreement; attorney-client privilege.” Entry three reads “Document 14: e-mail dated March 3 in which the general counsel tells the chief engineer that the licensing agreement is void and should be terminated before the deadline; attorney-client privilege.”

Entry one fails the floor, because it gives the requester nothing to test. Entry three fails the ceiling, because it hands over the advice itself and damages the privilege it claims. Entry two names the kind of document, the date, the participants, and the general subject. The requester can see a lawyer writing to a company employee about a legal matter, and can press the claim if something looks wrong, without learning what the lawyer said. Only entry two satisfies both clauses.

A description need not prove the privilege; it need only make the claim testable. Had entry two listed an outside vendor among the recipients, the requester would have a concrete question to ask, and entry one could never supply one. Neither brevity nor detail is safe by itself, since the short entry says too little and the long entry risks saying too much. The entry that works identifies the document by its external features and stops before the advice.

The description must also be accurate. An assessment built on a false picture of a document’s status is worthless, however carefully the entry avoids revealing content, and truthfulness never conflicts with secrecy, because an entry can report a document’s real status and still leave its substance undisclosed. In one DuPont dispute, a circuit court found that an affidavit had failed to note what DuPont’s counsel had done in another court regarding the non-privileged status of the Alta documents, and it called the omission misleading and an intentional concealment of potentially relevant information.⁸ The accuracy requirement guards against exactly that: a submission that hides a document’s real status.

Contents of a Log Entry

Rule 26(b)(5)(A)(ii) sets a standard and prescribes no form, so the contents of a log entry follow from what assessing a claim requires. Each fact belongs in the entry because it lets the requester test some part of the claim, and an entry that omits the fact leaves that part resting on the holder’s say-so.¹⁰

Courts and parties that have specified the contents converge on nearly the same list. A local rule quoted in one opinion calls for the type of document, its general subject matter, its date, and information sufficient to identify it for a subpoena, including the author, the addressees, and the relationship of the author, addressees, and recipients to each other.¹¹ A procedure quoted in United States v. Triumph Capital Group, Inc. required a log setting out the author, the recipients, the date, and the general nature of the documents, and it gave the government a chance to brief the court before the court decided whether the attorney-client privilege applied.¹² In Wells Fargo Bank v. Superior Court, Wells Fargo and its counsel provided a log that set out, for each withheld document, the privilege asserted and the document’s sequential number, general nature, date, author, and recipients.¹³ Another court ordered a party to serve a log “providing information as to each such document sufficient to sustain the burden of asserting the privilege(s),” and pointed to a decision that lists ten items which together suffice for the attorney-client privilege.¹⁴ The lists differ in wording but fit a single checklist.

Exam Analysis Steps

A complete entry answers six questions about the item, and each answer lets the requester test part of the claim.

What is it? Give the type of document and, where the log uses them, a sequential number, so that each entry matches one specific item.
When was it made? State the date, which fixes the item in time against the events in the case.
Who wrote it, and who received it? Name the author and every recipient, so that confidentiality can be examined person by person.
How do those people relate to the client and to one another? Say so, especially for anyone who does not obviously belong to the privileged relationship.
What is it generally about? Describe the subject in terms that show the purpose of the communication without disclosing the advice or strategy inside it.
What protection is claimed? Name the attorney-client privilege, work product, or both.

The first two questions rest on the fact that privilege is established document by document. In SmithKline Beecham Corp. v. Apotex Corp. the plaintiffs’ first log ran 370 pages and still gave the court too little to rule on. The court ordered a fuller log of 386 pages, and in the process the plaintiffs discovered that nearly 100 documents they had treated as privileged were discoverable.¹⁵ The party asserting the privilege bears the burden of establishing it, so the burden of that itemizing falls on the party, and the exercise forces it to look at each item.¹⁶

The author and recipient questions matter most for the attorney-client privilege, because a name tells the reader nothing until the reader knows who the person is. In one partnership dispute, plaintiffs moved to compel more than 1,500 documents identified in the defendants’ privilege logs, alleging that witnesses had testified that each defendant’s individual counsel attended negotiating sessions and discussed legal advice with nonclients.¹⁷ A requester can test a charge of that kind only if the log shows who received what. One court observed that it may be desirable to establish mandatory guidelines for a detailed log that identifies the individuals who author and receive each document and explains their relationship to the document and to the party asserting the privilege.¹⁸ The original SmithKline log failed in many instances to identify recipients and authors at all.¹⁹

The subject-matter entry sits under the rule’s ceiling. It should show the purpose of the communication, for example that it sought legal advice on a regulatory question, and stop before saying what the advice was. The last question prevents a different failure. In United States v. ChevronTexaco Corp. the log described many documents as a “recording of attorney legal analysis,” sometimes followed by “and work product.” The court could not tell what the phrase invoked, so for its in camera review it assumed that the attorney-client privilege was meant where “and work product” followed, and that both protections were meant where it did not.²⁰ The court’s assumptions supplied what the log should have said. Because the two protections have different elements, an entry that does not say which one is invoked leaves the requester guessing which to test.

Naming the protection is necessary but not enough. An entry that rests on a label and the word “self-explanatory,” with no affidavit behind it, was held insufficient in United States v. Construction Products Research, Inc. The word may be true for whoever holds the document, but the requester and the court do not hold it, and the word says nothing about who communicated with whom or why.

Hypothetical Illustration

A pharmaceutical distributor withholds a letter and logs it this way: “Letter dated September 12 from the distributor’s regulatory-affairs director to its outside regulatory counsel; no other recipients; the director is the employee responsible for regulatory compliance, and counsel has advised the distributor on regulatory matters for several years; request for legal advice concerning import-labeling requirements; attorney-client privilege.” The entry gives the type of document, the date, the author, the only recipient, each person’s connection to the client, the general subject, and the single protection invoked. A requester reading it sees a communication from a client employee to the client’s lawyer, with no outside recipient, about legal advice, and can test the claim by asking, for instance, whether anyone else received a copy. Nothing in the entry discloses what advice was sought or given.

Reduce that entry to “Letter re: labeling, self-explanatory” and the same letter becomes an entry no requester can test. The consequence is stated in SmithKline: the description of each document and its contents must be detailed enough for the court to decide whether the elements of the privilege, or of work product, are established, and failing this the documents must be produced.²¹ Any entry can be checked the same way. If the requester cannot answer one of the six questions, the entry is too thin; if the requester can read the advice, it is too full.

E-mail Threads and Attachments

Rule 26(b)(5)(A)(ii) requires a party to describe the nature of the “documents, communications, or tangible things” it withholds, so the unit of description is each communication.²³ An e-mail thread is a stack of communications, each with its own author, its own recipients, and its own date.

Electronic review shows why those identifiers matter. In one case, the person running a party’s review identified 210,635 responsive e-mails and then tried to filter out the privileged ones by searching the address line of every e-mail for the producing party’s name together with any of three other terms.²⁴ A screen of that kind works from addresses, message by message. A log that collapses a chain into one entry shows the requester a single address line and hides the rest.

The failure appears in the review of Merck’s privilege log. The reviewer reported that e-mail threads in which attorneys were ultimately involved were usually listed as a single message. Some of those threads ran ten to fourteen messages before the direct or limited exchanges with the attorneys began, and the reviewer concluded that each message should have received its own Bates number and its own identification on the log.²⁵ Technology now makes it possible to link physically what would once have been separate memoranda, but that does not justify treating the linked messages as one communication and denying the demanding party a fair opportunity to evaluate the privilege claims.²⁶

Merck had claimed privilege on the “entire document,” so the claim reached the earlier messages too. A later report from Merck did not assure the reviewer that all non-privileged threads had been produced, and the reviewer therefore noted the over-assertion document by document.²⁷ The defect lies in the logging, and it exists even when the thread proves protected: the same report found entire threads privileged where they were later integrated into communications made solely to attorneys for legal advice.²⁸ Whether a thread is privileged is a merits question. The logging rule asks only that its messages be visible.

An attachment travels as its own item. The reviewer denied the claim for e-mails to or from attorneys that revealed neither what the client was communicating (an attached study, report, or article, for example) nor what the attorney was advising (because the comments appeared on the attachment), and it did so whatever the ruling on the attachments. The e-mails may have been the means of obtaining legal advice, but they did not themselves reveal confidential information about the attorney-client relationship.²⁹ Because a cover e-mail and its attachment can come out differently, a log that merges them hides the difference.

A log that lists a long chain on a single line, or merges a cover e-mail with its attachment, fails in the same way: it asks the requester to accept the claim for items it cannot see.