Concept
Methods Under Rule 901(b)
Authentication and identification
Methods Under Rule 901(b)
Rule 901(a) sets one standard for every exhibit: the proponent must produce evidence sufficient to support a finding that the item is what the proponent claims it is.1 Rule 901(b) turns that standard into working practice with ten illustrations, which the rule itself calls “examples only” and “not a complete list.”2 Each illustration describes a connecting fact that litigants routinely use: a witness who saw the item made, a voice the listener knows, a call placed to an assigned number, a record found in the office where such records are kept.3
Because the list illustrates and does not limit, any proof that could support the required finding is allowed, whether or not a numbered entry describes it.4 Rule 901(b)(10) states the point outright by accepting any method of authentication or identification allowed by a federal statute or by a rule the Supreme Court prescribed.5 A route created elsewhere in federal law therefore counts as a Rule 901 method without any change to the list.
The entries cut both ways. A proponent who recites a listed method without facts that fit it has not met the burden, and a proponent whose facts fit no entry may still succeed if a reasonable juror could find the connection. The working question is therefore the same each time: what does the proponent claim the item is, what link does that claim require, and which evidence in the record supplies the link?
The sections below group the illustrations by the kind of link they supply. Testimony from a witness with knowledge comes first because the others build on it. Chain of custody follows, although no numbered entry names it, because courts treat it as the same circumstantial reasoning applied to items with no marks to recognize. Two ways of identifying handwriting come next, then the item's own features, then voices and telephone calls, then records, and finally processes and systems.
Witness With Knowledge
Rule 901(b)(1) states the first and simplest illustration: “Testimony that an item is what it is claimed to be.”7 Read with Rule 901(a), the entry means that a witness who knows the item is what it is claimed to be supplies the needed evidence by saying so.8 Courts treat it as the workhorse of the list. An Ohio appellate court, applying its counterpart to Rule 901(B)(1), called it the most commonly used method and noted that the threshold for authentication is low,9 and another court lists testimony of a witness with knowledge first among its examples of sufficient evidence.10
The method works because the fact that links an item to a case usually lives in someone's experience. A person who watched the item being made, made it, handled it, or knows it from long dealings can say how it connects to the dispute. That person testifies under oath and faces cross-examination, so the jury can judge credibility, and the judge asks only whether the testimony, taken as true, would let a reasonable juror find the connection. The same mechanism explains the limit: a witness who cannot say what the item is hands the jury one more unsupported assertion, no more reliable than the label on the item itself.
Because the foundation comes from the witness's own knowledge, the method fits any kind of item. A witness who knows a signature can authenticate it, the author of an e-mail can authenticate it, someone who knows a voice and its speaker can identify the voice, and the person who printed a document from a computer can authenticate it as a printout. A pay stub can be authenticated by the bookkeeper who generated it or the employee who received it.11 The testimony may be as simple as a witness explaining that the person on the other end of a conversation was the defendant because the witness has known the defendant for many years, or as elaborate as a series of witnesses who each describe how, when, and from whom they obtained an object.12 The second form is chain of custody, taken up in the next section. This section concerns the single witness whose own knowledge ties the item to its claimed identity.
The need for a person with knowledge shows most clearly when no one stands behind a writing. In Hernandez v. Apple Auto Wholesalers of Waterbury LLC, a car buyer swore that only $500 in cash was requested and paid, while the dealer's purchase order and contract recited a $1,000 cash payment. The court found no record evidence that the dealer received $1,000 and no testimony attesting to the recital, so the writing was unauthenticated and its assertion of payment was inadmissible hearsay.13 The business-records exception offered no way around the gap, because it too requires testimony from a custodian or another qualified witness.14 A paper that says a payment occurred is only a claim. It becomes evidence of the payment when someone with knowledge vouches for it, and in Hernandez no one did.
United States v. Brewer shows the opposite pole. The government authenticated bank surveillance footage through a teller who testified that it accurately depicted the events of the day in question. On cross-examination she admitted that she had no independent recollection of that day. The Seventh Circuit held that her clear initial testimony met the threshold, which left the jury to decide “the evidence's true authenticity and probative value.”15 What separates Brewer from Hernandez is not the strength of the witness but whether anyone with knowledge vouched for the item at all. The teller's later vagueness reduced the weight of the footage without erasing the foundation she had already given.
A witness need not have been present when the events happened, so long as the witness knows the item is what it is claimed to be. In Alderson v. Bonner the original videotape had been destroyed, and the plaintiffs offered a copy. Officer Skoglund, who had viewed the original, testified that there was no material difference between the two tapes. The Idaho court held that his testimony showed he knew the videotape was what the plaintiffs claimed, which satisfied the authentication requirement even though the plaintiffs had not shown where the copy came from.16 His knowledge covered the fact that mattered: the match between the copy and the original he had seen.
Photographs and maps follow the same pattern, because the witness's knowledge fixes what the picture shows. In Lucero v. Stewart, the Ninth Circuit found the foundation adequate for photographs of a lounge where the witness testified that they accurately represented the lounge as of his visit, though he could not give the precise date they were taken. In Zuk v. Zuk, a Pennsylvania court held that an owner's testimony that two exhibits were maps of the entire property as it originally existed supported a finding that they were what he described, because the owners were familiar enough with the property to endorse the maps as accurate depictions.17 Ownership and an actual visit were each a source of knowledge about the thing in the picture.
Chain of Custody
Recognition works only when an item has something to recognize, such as a serial number, a distinctive scratch, or initials the witness placed on it. A bag of white powder, a vial of blood, or an ordinary light bulb offers nothing of the kind, because every specimen of its type looks like every other. The claim about such an item has two parts: that it is the very thing seized or collected, and that it has not been swapped or altered since.19 A glance at the exhibit can establish neither, so the proponent substitutes continuity of possession for distinctive features and calls the people who handled the item. In United States v. Anderson, the First Circuit stated the premise: because the drugs there were not readily identifiable in any way, a testimonial tracing of the chain of custody was required.20
A typical drug prosecution shows how the tracing is built. A police officer takes the substance from the defendant, seals it in a container, marks the container, and testifies to what he did with it next. A second witness describes receiving the marked container and says what he did with it, and so on down the line, until a chemist's report becomes testimony about the substance taken from the defendant.21 The mark does quiet but important work. It turns an unmarked substance into an identifiable package that each later witness can recognize.
Anderson followed that pattern. Besore made a controlled purchase from the defendant and then met agent Thibodeau, who put the drugs in an evidence envelope, marked it with an identifying number that he recited at trial, took it to the DEA office, and locked it in the safe. Thibodeau then mailed the drugs in a bag bearing his initials to the DEA laboratory in New York, where the forensic chemist, O'Rourke, identified the bag by those initials, by the date he received it, and by the date he sealed it again. O'Rourke also established that the bag stayed in the laboratory's custody until it returned to Maine for trial.22 The court found the chain adequate. Besore and his vehicle were searched extensively before the purchase, and afterward he remained in Thibodeau's sight during the drive, so he could not have contaminated or exchanged the evidence. Chain-of-custody evidence, the court reiterated, must be “adequate -- not infallible.”23 That standard follows from the question the court is asking. The judge does not need certainty that no handoff went wrong, only enough evidence for a reasonable juror to find that the item is the one the proponent claims.
The Federal Rules never mention chain of custody by name, yet courts treat the technique as one more use of circumstantial proof to establish genuineness.24 That classification explains most of what follows. The court asks the ordinary Rule 901(a) question, and each handoff is one fact bearing on the answer, so the court does not enforce a free-standing requirement of unbroken possession. The Idaho court said so in Alderson: proof of a chain of custody is not a separate requirement for admissibility. The plaintiffs there had not shown where an offered copy of a videotape came from, but the officer's testimony showed no break up to the moment he first viewed the original and no material difference between the two tapes.25 A rule demanding an unbroken chain would exclude genuine evidence whenever a single custodian was forgotten, even when every other fact pointed to identity.
Because a gap is one weak fact among many, courts usually send it to the jury as a matter of weight. In Bruther v. General Electric Co., the plaintiffs could not account for a light bulb after the accident. The court answered that Rule 901 regulates admissibility, and, relying on Seventh Circuit precedent, that “any discrepancies in the chain of custody go to the weight of the evidence, not its admissibility.”26 The jury, not the court, must evaluate what the gap signifies.27 United States v. Ortiz reached the same result. Agents and a Miami chemist traced the cocaine, but no one accounted for how it was sent from Puerto Rico to Miami, and the First Circuit held that the lapse affected only weight, especially because the defendant offered no evidence that the drugs had been switched, mishandled, or adulterated.
United States v. Lepanto shows how the Tenth Circuit sized up a break. A postal inspector recovered a roll of carpet the next day from the dumpster behind the defendant's residence, initialed it in two places, and handed it to a second inspector, who sealed it in a bag and sent it to the Postal Service crime laboratory. Technicians there removed a piece of gray insulation that an expert connected to one of the bombs. The defendant listed four breaks in the chain. The court found only one significant, the overnight stay in the dumpster, and saw little concern in the failure to initial all three pieces of carpet, since the roll stayed bundled until it reached the laboratory. A judge considers the nature of the article, the circumstances of its preservation and custody, and the likelihood that intermeddlers tampered with it, and the ruling is reviewed only for clear abuse of discretion. A bundled roll of carpet is harder to tamper with undetected than a vial of blood, so the more fragile and fungible the item, the more custody detail a judge will want. Other evidence also filled the gap: an expert matched paint chips to items from the defendant's apartment, and a co-resident identified part of the carpet as coming from that apartment. Absent some showing of tampering, courts do not presume it, and the court held that any remaining uncertainty about the source of the exhibit went to weight.28
At the other end of the scale, a single link can carry the whole chain. In United States v. Sutton, the government offered a box simply as the box that William Lambert gave Agent Blount, and Blount identified it. The prosecution asked the jury to infer that this was the same box of records that another witness had compiled and that the defendant had directed a third person to destroy, and the court held the inference reasonable in light of all the evidence.29 The chain needed to be only as long as the circumstances made it.
The weight rule has limits, and the hypothetical shows the first. A jury can assign weight only when some evidence of custody exists to be weighed, so a record with no custody evidence at all leaves nothing from which a reasonable juror could find identity. The second limit concerns affirmative evidence that the item was switched, mishandled, or altered. A bare absence of proof about a handoff is a gap, while evidence that something actually happened to the item can erode the foundation, and the judge may then conclude that no reasonable juror could find the item unchanged. The usual exam error is treating any break in custody as grounds for exclusion. The better answer asks whether any evidence of custody exists and whether anything specific points to alteration.
Lay Handwriting Opinion
Rule 901(b)(2) names as one way to satisfy Rule 901(a) “[a] nonexpert's opinion that handwriting is genuine, based on a familiarity with it that was not acquired for the current litigation.”31 When the claim is that a particular person wrote or signed a document, the link to be shown is the writer's identity.32 A person who has watched someone write, or has read enough of that person's genuine writing, comes to know the hand much as one knows a familiar face. A lay witness who knows the signature on a letter can therefore testify that the letter was signed by its claimed author, after which the letter goes to the jury.33 The same logic covers the familiar case of someone identifying a spouse's handwriting, where years of ordinary dealings supply the knowledge.34
The rule attaches two limits to the method. The first is that the witness must be a nonexpert. The opinion rests on recognition built from ordinary experience, which a juror can evaluate by asking how the witness came to know the hand. Skilled comparison has a different foundation. An expert who compares the questioned writing with authenticated specimens, and a jury that makes the same comparison itself, proceed under the method taken up in the next section.35
The second limit is that the familiarity must not have been acquired for the current litigation, and the reason follows from what such familiarity would be. A witness who first sees a signature after suit is filed, and studies it beside samples gathered by one side's lawyer, is forming a view for trial, and that process can be steered toward the answer the proponent needs. Familiarity that came from the witness's own dealings before the dispute carries no such risk, because nothing about the case shaped it. The rule therefore preserves genuine lay recognition and leaves litigation-driven comparison to experts and jurors working from authenticated specimens. Exam facts often signal the limit by describing a witness who examined a signature carefully at counsel's request. The inquiry looks first to when and why the familiarity arose and only then to how much of it there is.
Familiarity that satisfies the rule arises in two ways: the witness has seen the person write, or has seen writings purporting to be that person's under circumstances indicating their genuineness.36 Seeing the person write is the more direct basis, because the witness observed the writer and the writing together. The second route is a step removed. The witness did not see the writing made but met writings whose genuineness the circumstances secured, such as letters sent over years of correspondence or a statement signed with others present. What the two routes share is that something other than the witness's own assurance vouches for the source of the familiarity.
In United States v. Standing Soldier, Captain Hill testified that a note attributed to the defendant bore the same signature as a statement the defendant had signed earlier, and the defendant objected that Hill was not familiar with his signature. The Eighth Circuit rejected the objection because the earlier signature was clearly made in circumstances indicating its genuineness, which gave Hill adequate familiarity to authenticate the note, and the weight of his testimony was for the jury.37 The court's reasoning rested on the reliability of the exemplar, and its last clause carries the usual division of labor: once the foundation exists, any dispute about how well Hill knew the signature is for the jury.
In United States v. Sliker, the Second Circuit added that, to the extent papers had to be treated as assertions by Marks, Griffith's testimony could have qualified under Rule 901(b)(2) to authenticate Marks's handwriting, and it pointed to an earlier decision in which a witness had seen three checks signed.38 A witness who knows the writer's hand therefore offers an independent way to reach an authorship finding, and a proponent who has such a witness is not confined to the contents of the document.
The paralegal's failure does not end the supplier's effort to authenticate the guaranty. The list in Rule 901(b) is illustrative, so the supplier may still turn to comparison with authenticated specimens, to distinctive characteristics of the guaranty itself, or to any other proof that would let a reasonable juror find that the owner signed it.39 What the failed opinion cannot do is count as the lay handwriting identification the rule describes.
Specimen Comparison
Rule 901(b)(3) gives a third illustration: “A comparison with an authenticated specimen by an expert witness or the trier of fact.”41 A specimen is a known sample, such as a signature the alleged signer undeniably made, and the method sets the questioned item beside it to ask whether both came from the same hand. The trier of fact is whoever decides the disputed facts, the jury or the judge when no jury sits.42 This method fills the gap left by lay opinion. When no witness knows the writer's hand from dealings outside the lawsuit, recognition is unavailable, and the proponent can substitute a side-by-side examination of samples whose genuineness has been shown.
The rule has two moving parts, and each has its own reason. The first concerns who may compare. An expert qualifies because training lets the expert see features of a signature that an untrained eye misses. The trier of fact qualifies for a different reason: examining two writings takes no special skill, and the jury can see both in the courtroom, so no witness has to vouch for a resemblance the jury can inspect for itself.43 In the standard illustration of a letter offered as signed by a man named Pearce, a qualified handwriting expert could compare Pearce's signature on the letter with authenticated specimens of his signature or handwriting, and the jury could make the same comparison.44 The rule names these two comparers and no others, so a witness who is neither an expert nor the jury has no role under this entry.
The second part concerns the baseline. A comparison is only as reliable as the sample it rests on, because a match between a questioned signature and an unproven sample shows only that two papers resemble each other. A forger could supply a convincing specimen as easily as a convincing questioned item, so the specimen must itself be authenticated, whether by circumstantial evidence or by a witness who can say the specimen is what it is claimed to be.45 Whatever the route, the authentication must come from a source other than the expert who uses the specimen for the comparison.46 That requirement prevents circularity, since an expert who vouched for the sample and then reasoned from it would be proving the sample's genuineness by assuming it. Once a reasonable juror could find the specimen genuine and a proper comparer has made the comparison, the question returns to the ordinary Rule 901(a) inquiry, and disputes about how closely the writings match go to the jury.
The method is not limited to signatures. One court illustrated the entry with Dawson, where the employment applications at issue were on the same form as applications whose authenticity was conceded, so the trier of fact could compare the questioned documents with authenticated ones.47
The common mistake is to credit any side-by-side look without asking who is doing the looking, since a specimen can be perfectly authenticated while the comparer is still the wrong person.
Only the comparer varied across those routes, because the authenticated card stayed the same. The reverse failure is equally instructive. A qualified expert handed a signature of unknown origin is the right comparer with no baseline, and the comparison cannot authenticate the questioned item, because nothing shows the sample is the signer's. The lay-recognition entry covers a different situation: a witness whose familiarity grew from dealings with the signer before the dispute may give an opinion without any comparison to samples. Because Rule 901(b) is illustrative, a proponent whose comparison fails may still rely on any other proof from which a reasonable juror could find genuineness.48
Distinctive Characteristics
Rule 901(b)(4) lets an item help prove its own origin, within limits, by listing as a sufficient method “[t]he appearance, contents, substance, internal patterns, or other distinctive characteristics of the item, taken together with all the circumstances.”50 Like every Rule 901(b) illustration, it serves Rule 901(a), so the target stays the same: evidence from which a reasonable juror could find that the item is what the proponent claims.51 No one has to have seen the item made. A letter's addressee, its postmark, what it discusses, and how it is phrased can each point toward a particular source, and the jury weighs the pointers together. The method is therefore the natural choice for letters, notes, e-mails, and records that reach the proponent with no one able to say where they came from.
The two qualifiers in the rule carry its reasoning. A characteristic must be distinctive, because a detail that many people could supply says nothing about which of them produced the item. It must also be taken together with all the circumstances, because a feature of the page takes its meaning from what surrounds it, such as where the item turned up, what other evidence confirms its details, and what communication came before it. The court in Zenith illustrated the point with a “chop,” a seal used like a signature. It gave the chop the weight of a signature while acknowledging that a chop may be forged and that many people with the same surname may share one, so each chop had to be judged on the document where it appeared.52
The clearest example of features working together is United States v. Bagaric. Agents found a letter during a consent search of Logarusic's home, and he challenged its authentication. The Second Circuit explained that authentication may rest entirely on circumstantial evidence, including the appearance, contents, substance, and other distinctive characteristics of the writing. The letter was addressed to Logarusic and postmarked in Asuncion, Paraguay, where the claimed author, Baresic, lived. It was signed with names that matched an alias Baresic had used to enter the United States, it referred to people in Chicago, where four of the defendants lived, and it asked Logarusic to contact someone by a nickname that the proof tied to a codefendant. It also complained of efforts to expel the writer as a dangerous terrorist, which fit testimony that Baresic was a fugitive sought in Sweden for the murder of the Yugoslavian ambassador. The trial judge found ample demonstration that the letter was in fact what the government claimed, a letter from Baresic to Logarusic, and the appellate court agreed.53
No single detail settled the matter. The postmark placed the sender where Baresic lived, the names and nickname matched an alias and a known sobriquet, the references fit people around Logarusic, and the letter turned up in the home of its addressee. Much of the proof also came from outside the letter. Testimony supplied the alias, the nickname, and the fugitive status, so the contents persuaded because independent evidence confirmed them, and that is why the method works only when the circumstances back up what the page claims.
A proponent with no such details is tempted to rely on what the page says about itself, and that is the usual error on this topic. Does a letterhead with a signature block beneath it authenticate a letter? Courts were traditionally unwilling to treat letterhead as enough, and more recent decisions accept it as at least a partial basis for authentication, though letterhead is accessible to outsiders and easy to produce. It therefore serves as one strand of proof and rarely as the whole of it.54
A web page shows the same reasoning at work. The Second Circuit accepted in Zhyltsou that contents or distinctive characteristics can sometimes supply circumstantial proof by themselves, and that a writing may be authenticated by evidence that its contents were not a matter of common knowledge. The information on the page offered to tie it to the defendant, however, was also known to the cooperating witness and likely others, some of whom might have had reasons to create a false profile in the defendant's name, and nothing apart from the page showed that he even had such a profile.55 The court required some basis beyond the cooperating witness's own testimony from which a reasonable juror could conclude that the page was in fact the defendant's profile, and no such showing was made.56 A state appellate court reached a parallel result for a social-networking printout. A photograph, a birth date, and a location were not sufficient distinctive characteristics, because someone other than the purported creator could have made the site and posted the content, and a printout from such a site required a greater degree of authentication than those details supplied.57 The decisions differ from Bagaric in who held the details. The alias, the nickname, and the fugitive status belonged to a small circle and were confirmed by independent testimony, while a name, a photograph, and facts about a life belong to anyone who knows the person. Contents authenticate when they are specific to the claimed source, and they fail when anyone could have copied them.
Where an item is found is one of the circumstances the rule counts, and it often does work that contents cannot do alone. Zenith treated a party's production of documents as circumstantial evidence of authenticity, and it drew on United States v. Natale, where the defendants' presence at the office where a notebook was discovered, their meetings there with a witness, and one defendant's admission that the office was his supported authenticity together with the notebook's contents.58 People tend to keep their own papers in their own places, so the place raises the likelihood that the item belongs to the claimed source. Many people may have had access to the same place, though, so courts pair location with contents instead of resting on location alone.
Internal patterns work through habits of expression that are hard to imitate and easy to recognize. A writer who misspells a word the same way every time leaves that mark on every document. In United States v. Larson the court admitted evidence that the defendant misspelled approach as “approuch” three times in one ransom note and had done the same in an earlier letter to the Pardon Board, and in United States v. Clifford letters were authenticated in part by unusual misspellings.59 A turn of phrase the claimed author always used, or a relatively unknown nickname for the recipient, can serve the same way.60 A pattern authenticates only when something shows that the claimed author shares it. A misspelling in a questioned note shows that the writer spelled badly, and the earlier letter known to come from the defendant is what turned the error into a connection.
The reply doctrine is the most common circumstantial technique under this entry. It applies when evidence shows that a communication was sent to a person or business and a later communication purporting to answer it arrives, and the later communication is then treated as authenticated as coming from the recipient of the first.61 The Pearce illustration shows the logic: if the reply attributed to Pearce is responsive to an earlier letter from Carter, Carter can authenticate his own letter and testify that the reply answers it, and a reasonable jury could find that Pearce wrote it.62 The mechanism is informational. A stranger who never saw Carter's letter would not know what it asked, so a response that addresses its particular points suggests that its writer received it. The preconditions come from the same reasoning: someone must establish that the first communication was sent to the claimed author, and the reply must actually respond to it, because a generic answer that would fit any inquiry says little about who received the question.
Once the proponent has laid this foundation, the judge and jury divide the work as before. The district court need only assess whether the proponent has offered a satisfactory foundation from which the jury could reasonably find the evidence authentic, and the factual determination of whether it is what the proponent claims is ultimately reserved for the jury.63 Arguments that someone else could have known those details or written the item therefore go to weight.
Voice Identification
Rule 901(b)(5) lists as a sufficient method an opinion identifying a person's voice, whether heard firsthand or through mechanical or electronic transmission or recording, based on hearing the voice at any time under circumstances that connect it with the alleged speaker.65 The entry applies Rule 901(a) to a recurring claim: that the voice on a call, a voicemail, or a recording belongs to a particular person.66 A voice carries no name tag, so the jury needs a witness who can match the sound to a speaker. The rule supplies that witness in a layperson who recognizes a voice the way anyone recognizes a familiar face, and the testimony can be as simple as a witness explaining that he knows the person he spoke with because he has known that person for many years and always recognizes him.67
Two phrases in the text do most of the work, and the first is “at any time.” The witness's task is a comparison: the witness holds a memory of how the alleged speaker sounds and asks whether the voice on the item matches it. That comparison works equally well whether the memory formed a year before the recorded call or a week after it, so a rule requiring the familiarity to predate the event would shut out capable witnesses without making any identification more reliable.
The second phrase, “under circumstances that connect it with the alleged speaker,” keeps the opinion from resting on a guess. Hearing a voice is not enough by itself, because the witness must also know whose voice it was, and that knowledge comes from an occasion that tied the sound to the person: a conversation in which the witness met the speaker, a call with a speaker the witness could identify, or an interview after an arrest. Without such an occasion the witness has nothing stored to compare against, and the opinion adds only the proponent's bare assertion. The timing language is generous, and the connection language keeps that generosity from licensing speculation.
United States v. Trent gives a clean example of the rule satisfied. Sergeant Rieck met Trent in person to discuss a possible heroin transaction and afterward spoke with him by telephone, and at trial the sergeant identified Trent's voice on the recorded calls. The Seventh Circuit explained that its cases require only “minimal familiarity” with the voice and held that the in-person meeting supplied it. The government laid the foundation by asking whether the sergeant believed the man he had met in person was the man he had spoken with on the phone, and he said yes. The meeting both gave the sergeant a sound to remember and tied that sound to Trent, which is the connection the rule requires. The court described the bar as low, noting that hearing a voice a single time during a court proceeding has been held sufficient, and corroboration such as phone records and saved contacts reinforced the result without carrying it.
Challenges to a voice identification often add demands the rule does not contain. In United States v. Zepeda-Lopez the defendant argued that an agent's identification of his voice on audio tapes was inadmissible because the agent did not speak Spanish and so could not compare idiom, usage, or inflection, and because the tapes were never sent to the FBI for voiceprint identification. The Tenth Circuit noted that he cited no authority for those asserted foundational deficiencies.68 The text of the rule explains why: it asks whether the witness's exposure connected the voice with the alleged speaker and says nothing about fluency in the speaker's language or about scientific comparison. A thin acquaintance remains a fair subject for cross-examination, and the jury decides how much the identification is worth.
An exam stem will often build the problem around timing or training, asking whether a witness who first met the speaker after the recording was made may identify the voice, or whether a voiceprint expert is required. The usual error is to demand either. The rule's text imposes neither requirement, and the only question is whether the witness's exposure connected the voice to the alleged speaker. A separate method for telephone calls, which authenticates a call by the number dialed and the circumstances of the conversation, comes next.
Telephone Calls
Rule 901(b)(6) gives the proponent of a telephone conversation a method that starts from the number dialed instead of from anyone's recognition of a voice. It accepts evidence that the call was made to the number assigned at the time to a particular person, if circumstances, including self-identification, show that the person answering was the one called, or to a particular business, if the call related to business reasonably transacted over the telephone.70 Like every illustration in Rule 901(b), the method serves Rule 901(a), so the question remains whether a reasonable juror could find that the conversation was with the person or business the proponent claims.71 Voice identification needs a witness who has heard the speaker before, while this method builds the link from facts about the call itself, and courts accordingly accept that a witness who recognizes the speaker's voice is not a precondition to admitting evidence of a call.72
A telephone number works like an address. The telephone company assigns it to a subscriber, and a call placed to it reaches that subscriber's line, so a proponent who dialed already holds a fact that points toward the claimed recipient before anyone answers. A directory listing can show whose number it was, and the proof must reach the time of the call, because numbers change hands and a listing from another year says nothing about who held the line on the day in question. Assignment still leaves a gap, since the line may be answered by a household member, a coworker, or a stranger standing nearby. The two prongs of the rule close that gap in different ways.
When the claim is that the proponent spoke with a particular person, the rule adds circumstances, including self-identification, showing that the person answering was the one called. Self-identification carries weight here because the proponent chose the destination. The answerer's announcement of a name responds to a call the answerer did not initiate, and the separately provable assignment of the number corroborates it, so an impostor would have had to anticipate this particular call. Reaching the person called suffices when the caller dialed the number in the directory assigned to that person and the person reached identified herself or provided other clues that confirmed her identity.73
The voicemail is not necessarily lost to the tenant, because contents that fit only the landlord or a witness who knows his voice could still supply a foundation through other methods. The dialed-number method simply has no number to work from.
The business prong asks a different question and gets by without any self-identification. A call to the number assigned to a particular business authenticates the conversation as one with that business when the call related to business reasonably transacted over the telephone.74 The reason follows from the claim being made. The proponent asserts that the business was reached, and a business that lists a number for taking orders, scheduling work, or confirming reservations expects calls on those subjects and puts someone on the line to handle them. The subject of the call therefore does the work that self-identification does for a person. The second condition also marks the prong's limit: a call to a business number about a matter outside what the business reasonably does by phone has no such subject-matter link, and the proponent must turn to another method.
The Federal Circuit's decision in Phigenix, Inc. v. Immunogen, Inc. shows what happens when none of these facts is supplied. A declarant testified that his company and Genentech had multiple discussions through telephone conversations about the company's technology and patent portfolio. The court observed that the declarant had not established that a particular number was dialed, that the number dialed was assigned at the time to Genentech, or that the call related to business reasonably transacted over the telephone.75 Each omission matches a link in the rule. Without a number, nothing points toward Genentech; without the assignment, the number points nowhere; and without the subject matter, the call has no tie to the business even if it reached it. A statement that discussions occurred through telephone conversations is a conclusion about the calls, and the rule asks for the facts that let a reasonable juror reach that conclusion independently.
Incoming calls lack the feature that makes outgoing calls manageable, because the recipient neither chose the number nor controlled who was dialing. In United States v. Dhinsa, Uberoi, the girlfriend of Satinderjit, testified about threatening telephone calls from a man who identified himself as “Gurmeet Singh.” The first caller told her to pass a message to her boyfriend or both of them would be shot, and a second caller using the same name later asked whether she had delivered it. The Second Circuit found a sufficient basis to conclude that the calls were adequately authenticated.76 Its governing statement was that a mere assertion of identity by a person talking on the telephone does not in itself authenticate that person's identity, but additional evidence, which “need not fall in[to] any set pattern,” may provide the necessary foundation.77 The court's examples were similarity between what the speakers discussed and what each later did, and a conversation that discloses knowledge of facts known peculiarly to the claimed speaker.78
The limit exists because a bare announcement of a name costs a caller nothing and is exactly what an impostor would say. Outgoing calls overcome that weakness through the dialed number, which the proponent chose and can prove. An incoming call offers no such anchor, so something other than the caller's own words must tie the call to the claimed speaker. In Dhinsa the second call asked about the first message and repeated its words, which tied the two calls together and gave the court content to weigh beyond the name. The decisions differ in what the witness could describe: Uberoi recounted what the callers said and asked, while the Phigenix declarant described conversations only in general terms, leaving the court nothing to evaluate. A frequent exam error is applying the outgoing-call method to a call the witness received. A witness who only answered the telephone cannot rely on an assigned number, and the foundation must come from the content of the call, from later conduct that matches it, or from some other circumstance that connects it to the claimed speaker.
Public Records
Rule 901(b)(7) authenticates an item through custody. It accepts evidence that a document was recorded or filed in a public office as authorized by law, or that a purported public record or statement is from the office where items of this kind are kept.80 Like every illustration in Rule 901(b), it is one way of meeting Rule 901(a), so the question remains whether the proponent's evidence would let a reasonable juror find that the item is what the proponent claims.81 The court in Smart & Associates, LLC v. Independent Liquor (NZ) Ltd. listed the method among the Rule 901(b) examples of sufficient proof, beside testimony from a witness with knowledge and the item's distinctive characteristics.82 What sets this method apart is that the connecting fact is a place. The proponent shows that the paper sits in, or came from, the office where the law puts papers of that kind, and that custody does the work a witness to the writing would otherwise do.
Official custody proves something because a public office handles its files under legal duties and fixed procedures, and a paper found in the proper office's file arrived there through those procedures. A person who wanted to pass off a false deed or a false agency report would have to get it into the office's records, and the office's control over its own files makes that hard to do unnoticed. The place the item came from therefore points toward the source the proponent claims. The Zenith court drew the contrast with private custody, noting that a document's emergence from public custody makes a prima facie showing of authenticity, while private custody is too varied in its circumstances to carry that weight alone and becomes strong evidence only together with other circumstances.83
The rule's two clauses ask for slightly different facts. The first, for a document recorded or filed in a public office as authorized by law, most naturally fits papers that private parties create and then deposit in an official file, such as a deed placed in a county's land records. The words “as authorized by law” tie the deposit to a legal permission or duty, so the paper entered the file through the office's rule-governed intake and not because a stranger left it on a counter. The second clause, for a purported public record or statement from the office where items of this kind are kept, fits what the office itself produces or holds, and its limiting words are “where items of this kind are kept.” A paper that really is a county record still lacks the link if it came from some other place, because only the proper repository carries the assurance of official handling.
Because the connecting fact is custody and source, the proof usually comes from a witness who can describe the office's practice and say where the exhibit came from. Testimony about where an item came from is the same kind of proof that chain-of-custody testimony supplies, only with a very short chain.84 The witness need not be the file's official keeper. In United States v. Morales, the Ninth Circuit considered immigration forms that the government offered from the aliens' files. An immigration agent testified that the forms were kept in the file of each person who has contact with immigration and that a copy of each form is sent to that file in the normal course of business. The court held that this testimony supported a reasonable inference that the forms were accurate copies of documents from the A-files, which were recorded and filed in a public office as required by law.85
The two witnesses differ in what they know about the office, and that knowledge accounts for the whole difference in result. The neighbor could describe only the paper itself, which the jury could already see, while the deputy could describe the office's practice and the paper's place in its files, which is the link a reasonable juror needs. An official-looking heading or the words “public record” on the page cannot substitute for that knowledge. The usual error on this topic is to treat the label as the proof when the label is only the claim.
The proof here establishes genuineness and nothing more. A deed shown to be the one on file has not thereby been shown to contain true statements. In Morales the court treated the Confrontation Clause and Rule 803(8) objections to the forms as a claim distinct from the authentication objection.86 Later in this chapter, self-authentication takes up documents that need no sponsoring witness at all. Without such a route, the proponent under Rule 901(b)(7) must supply the custody evidence through testimony.
Ancient Documents
Rule 901(b)(8) lets the proponent authenticate an old document or data compilation, which is a body of stored information such as a set of archived records, without calling anyone who watched it being made. The proponent produces evidence that the item meets three conditions, all of which must be met:88
- The item is in a condition that creates no suspicion about its authenticity.89
- The item was in a place where, if authentic, it would likely be.90
- The item is at least 20 years old when offered.91
Like every illustration in Rule 901(b), this one is a way of satisfying Rule 901(a), so the proponent's task remains to produce evidence from which a reasonable juror could find that the item is what the proponent claims it is.92 The method exists because authenticating evidence for an old document may be hard to find, since the people who wrote, signed, or handled a paper two decades ago may be gone or unable to remember it, and most methods taught above depend on such people.93 The rule replaces their testimony with an argument from probability.
The three conditions work as a set because each answers a different doubt, and none can carry the showing alone. Age by itself proves nothing, since an old forgery is still a forgery.94 The condition requirement asks whether the face of the document gives a reason to doubt it, as an overwritten date or a patched signature would. The place requirement tests custody, because papers tend to rest where the people with an interest in them keep them, so a paper found in its natural repository probably arrived there through ordinary handling. Together the conditions rest on the assumption that a document which appears genuine, has gone twenty or more years without encountering suspicion, and has been stored in a normal place deserves at least a jury's look at its authenticity.95
A 1997 deportation case shows all three conditions operating together. Government lawyers sought to prove that a Russian native had joined the Nazis in persecuting and murdering Lithuanian Jews during the Second World War, and they offered records taken from the archives of the former Soviet Union. The records showed no signs of tampering or fraud, and the repository was a likely home for them, because it also housed many other war documents that trial testimony showed to be authentic.96 Wartime papers offered in 1997 were decades past the twenty-year line, nothing about their condition invited suspicion, and an archive that held genuine wartime records was where authentic ones would be found. Each condition supplied a different piece of the inference that no one had fabricated the records for the lawsuit.
The age condition is the only one measured by a number, and the rule measures it when the item is offered. A document that turns twenty before the proponent tenders it qualifies, and a paper that falls short cannot use this method by itself. It may still be authenticated by any other proof that supports the Rule 901(a) finding.97
Satisfying the three conditions makes only a prima facie case of genuineness. The judge asks whether a reasonable jury could find the document genuine, and if the answer is yes the document is admitted. The jury then decides whether it is in fact genuine, and the opponent remains free to offer counter-evidence that the document is not authentic or is otherwise inaccurate.98 An authenticating process conducted in conformity with one of the Rule 901(b) illustrations satisfies the requirement of authenticating or identifying an item of evidence.99
The showing reaches authenticity and stops there. An ancient letter that passes Rule 901(b)(8) is established as the letter it claims to be, but whether the factual assertions inside it may be considered for their truth is a different question, one of hearsay.100 A ledger can therefore be proven genuine and still be excluded when offered to prove what its entries say. The usual errors on this topic are to treat age alone as authentication and to treat authentication as proof that the contents are true.
Process or System
Rule 901(b)(9) gives the proponent of machine-generated output a method that focuses on the machine: “Evidence describing a process or system and showing that it produces an accurate result.”102 Like every illustration in Rule 901(b), it is a route to the general standard of Rule 901(a), so the question remains whether the evidence would let a reasonable juror find that the item is what the proponent claims it is.103 The method exists because no person perceived the result. A program that averages five years of repair bills did not watch the repairs, and the employee who ran the program may not know what the figures will be until they print. Since no witness can say from observation that the numbers are right, the proponent connects the output to its claim by showing that the process which produced it can be trusted. The rule's two parts, describing the process and showing that it is accurate, follow from that need.
A trucking company that wants to offer a printout of the average monthly cost of service, maintenance, and repairs for its fleet of 200 trucks over five years supplies the standard illustration. Evidence showing how the basic data were collected, how the information was arranged for computer processing, and the basic features of the company's software program would give an adequate foundation to authenticate the printout.104 Each of those three subjects marks a place where the figures could go wrong. Repair costs recorded carelessly at the garage would corrupt every average built on them, costs arranged for processing under the wrong month or the wrong truck would land in the wrong place, and a program that divided by the wrong number would misstate the result however good the input. A witness who walks through the path the numbers took from the garage to the page therefore gives a reasonable juror what is needed to find that the printout is a correct statement of the company's average costs.
How much of that path the witness must describe depends on what the computer did to the data. In Pegasystems Inc. v. Appian Corp., a trade-secret dispute between two software companies, the Virginia Court of Appeals restated the line: electronic records that are merely stored in a computer raise no computer-specific authentication issues, while authentication issues may arise when a computer processes data instead of simply storing it. A stored e-mail or scanned invoice resembles a paper in a file cabinet, because the machine holds the content without creating any of it, and the ordinary methods taught above, such as a witness with knowledge or distinctive characteristics, connect it to its claimed source. Processed output differs because the machine supplies the content, so the process becomes the thing the proponent must vouch for. The court added that the information required about data input and processing depends on the nature and completeness of the data, the complexity of the manipulation, and the routineness of the operation. That sliding scale follows from the same mechanism as the rule itself. Every added step is another place for error, and an unfamiliar operation gives a juror less everyday experience to lean on. Totaling a column of entered figures needs little explanation, while a custom model that weights, filters, and projects needs a great deal more.
A foundation that skips the process fails even when the output looks professional.
The usual error is to treat the computer's involvement as proof of accuracy, as though a number printed by a machine vouched for itself. The rule asks for evidence about the process, and the fact that a particular machine printed the page says nothing about whether the process behind it works.
Opposing parties often argue that electronic data are easy to alter and should therefore face a stricter test. In Pegasystems the trial court had excluded a party's software evidence and denied it the chance to authenticate it, and the Court of Appeals held both rulings erroneous. The opponent had cited decisions acknowledging that electronic evidence may be manipulated, but the court read them to apply traditional authentication principles, and it noted that introducing the authenticated evidence would not have deprived the opponent of the chance to challenge it, since the opponent remained free to argue to the jury that the software was corrupted.105 The reason fits the division of labor between judge and jury taught earlier. A worry about tampering is a doubt about whether the output is accurate, and once the foundation clears the ordinary threshold, that doubt belongs to the jury as a matter of weight.
Because the foundation is testimony about a system, calling a live witness for every printout of routine system output can be burdensome. A later section considers when a written certification can stand in for that live testimony.