Melendez-Diaz v. Massachusetts

with whom The Chief Justice, Justice Breyer, and Justice Alito join, dissenting.

The Court sweeps away an accepted rule governing the admission of scientific evidence. Until today, scientific analysis could be introduced into evidence without testimony from the “analyst” who produced it. This rule has been established for at least 90 years. It extends across at least 35 States and six Federal Courts of Appeals. Yet the Court undoes it based on two recent opinions that say nothing about forensic analysts: Crawford v. Washington, 541 U. S. 36 (2004), and Davis v. Washington, 547 U. S. 813 (2006).

It is remarkable that the Court so confidently disregards a century of jurisprudence. We learn now that we have misinterpreted the Confrontation Clause — hardly an arcane or seldom-used provision of the Constitution — for the first 218 years of its existence. The immediate systemic concern is that the Court makes no attempt to acknowledge the real differences between laboratory analysts who perform scientific tests and other, more conventional witnesses— “witnesses” being the word the Framers used in the Confrontation Clause.

Crawford and Davis dealt with ordinary witnesses— women who had seen, and in two cases been the victim of, the crime in question. Those cases stand for the proposition that formal statements made by a conventional witness— one who has personal knowledge of some aspect of the defendant’s guilt — may not be admitted without the witness appearing at trial to meet the accused face to face. But *331Crawford and Davis do not say — indeed, could not have said, because the facts were not before the Court — that anyone who makes a testimonial statement is a witness for purposes of the Confrontation Clause, even when that person has, in fact, witnessed nothing to give them personal knowledge of the defendant’s guilt.

Because Crawford and Davis concerned typical witnesses, the Court should have done the sensible thing and limited its holding to witnesses as so defined. Indeed, as Justice Thomas warned in his opinion in Davis, the Court’s approach has become “disconnected from history and unnecessary to prevent abuse.” 547 U. S., at 838 (opinion concurring in judgment in part and dissenting in part). The Court’s reliance on the word “testimonial” is of little help, of course, for that word does not appear in the text of the Clause.

The Court dictates to the States, as a matter of constitutional law, an as-yet-undefined set of rules governing what kinds of evidence may be admitted without in-court testimony. Indeed, under today’s opinion the States bear an even more onerous burden than they did before Crawford. Then, the States at least had the guidance of the hearsay rule and could rest assured that “where the evidence f[ell] within a firmly rooted hearsay exception,” the Confrontation Clause did not bar its admission. Ohio v. Roberts, 448 U. S. 56, 66 (1980) (overruled by Crawford). Now, without guidance from any established body of law, the States can only guess what future rules this Court will distill from the sparse constitutional text. See, e. g., Méndez, Crawford v. Washington: A Critique, 57 Stan. L. Rev. 569, 586-593 (2004) (discussing unanswered questions regarding testimonial statements).

The Court’s opinion suggests this will be a body of formalistic and wooden rules, divorced from precedent, common sense, and the underlying purpose of the Clause. Its ruling has vast potential to disrupt criminal procedures that already give ample protections against the misuse of scientific *332evidence. For these reasons, as more fully explained below, the Court’s opinion elicits my respectful dissent.

I

A

1

The Court says that, before the results of a scientific test may be introduced into evidence, the defendant has the right to confront the “analysts.” Ante, at 310-311. One must assume that this term, though it appears nowhere in the Confrontation Clause, nevertheless has some constitutional substance that now must be elaborated in future cases. There is no accepted definition of analyst, and there is no established precedent to define that term..

Consider how many people play a role in a routine test for the presence of illegal drugs. One person prepares a sample of the drug, places it in a testing machine, and retrieves the machine’s printout — often, a graph showing the frequencies of radiation absorbed by the sample or the masses of the sample’s molecular fragments. See 2 P. Giannelli & E. Imwinkelried, Scientific Evidence § 23.03 (4th ed. 2007) (describing common methods of identifying drugs, including infrared spectrophotometry, nuclear magnetic resonance, gas chromatography, and mass spectrometry). A second person interprets the graph the machine prints out — perhaps by comparing that printout with published, standardized graphs of known drugs. Ibid. Meanwhile, a third person — perhaps an independent contractor — has calibrated the machine and, having done so, has certified that the machine is in good working order. Finally, a fourth person — perhaps the laboratory’s director — certifies that his subordinates followed established procedures.

It is not at all evident which of these four persons is the analyst to be confronted under the rule the Court announces today. If all are witnesses who must appear for in-court confrontation, then the Court has, for all practical purposes, *333forbidden the use of scientific tests in criminal trials. As discussed further below, requiring even one of these individuals to testify threatens to disrupt if not end many prosecutions where guilt is clear but a newly found formalism now holds sway. See Part I-C, infra.

It is possible to read the Court’s opinion, however, to say that all four must testify. Each one has contributed to the test’s result and has, at least in some respects, made a representation about the test. Person One represents that a pure sample, properly drawn, entered the machine and produced a particular printout. Person Two represents that the printout corresponds to a known drug. Person Three represents that the machine was properly calibrated at the time. Person Four represents that all the others performed their jobs in accord with established procedures.

And each of the four has power to introduce error. A laboratory technician might adulterate the sample. The independent contractor might botch the machine’s calibration. And so forth. The reasons for these errors may range from animus against the particular suspect or all criminal suspects to unintentional oversight; from gross negligence to good-faith mistake. It is no surprise that a plausible case can be made for deeming each person in the testing process an analyst under the Court’s opinion.

Consider the independent contractor who has calibrated the testing machine. At least in a routine case, where the machine’s result appears unmistakable, that result’s accuracy depends entirely on the machine’s calibration. The calibration, in turn, can be proved only by the contractor’s certification that he or she did the job properly. That certification appears to be a testimonial statement under the Court’s definition: It is a formal, out-of-court statement, offered for the truth of the matter asserted, and made for the purpose of later prosecution. See ante, at 309-311. It is not clear, under the Court’s ruling, why the independent contractor is not also an analyst.

*334Consider the person who interprets the machine’s printout. His or her interpretation may call for the exercise of professional judgment in close cases. See Giannelli & Imwinkelried, supra. If we assume no person deliberately introduces error, this interpretive step is the one most likely to permit human error to affect the test’s result. This exercise of judgment might make this participant an analyst. The Court implies as much. See ante, at 318-320.

And we must yet consider the laboratory director who certifies the ultimate results. The director is arguably the most effective person to confront for revealing any ambiguity in findings, variations in procedures, or problems in the office, as he or she is most familiar with the standard procedures, the office’s variations, and problems in prior cases or with particular analysts. The prosecution may seek to introduce his or her certification into evidence. The Court implies that only those statements that are actually entered into evidence require confrontation. See ante, at 309. This could mean that the director is also an analyst, even if his or her certification relies upon or restates work performed by subordinates.

The Court offers no principles or historical precedent to determine which of these persons is the analyst. All contribute to the test result. And each is equally remote from the scene, has no personal stake in the outcome, does not even know the accused, and is concerned only with the performance of his or her role in conducting the test.

It could be argued that the only analyst who must testify is the person who signed the certificate. Under this view, a laboratory could have one employee sign certificates and appear in court, which would spare all the other analysts this burden. But the Court has already rejected this arrangement. The Court made clear in Davis that it will not permit the testimonial statement of one witness to enter into evidence through the in-eourt testimony of a second:

*335“[W]e do not think it conceivable that the protections of the Confrontation Clause can readily be evaded by having a note-taking policeman [here, the laboratory employee who signs the certificate] recite the unsworn hearsay testimony of the declarant [here, the analyst who performs the actual test], instead of having the declarant sign a deposition. Indeed, if there is one point for which no case — English or early American, state or federal — can be cited, that is it.” 547 U. S., at 826.

Under this logic, the Court’s holding cannot be cabined to the person who signs the certificates. If the signatory is restating the testimonial statements of the true analysts— whoever they might be — then those analysts, too, must testify in person.

Today’s decision demonstrates that even in the narrow category of scientific tests that identify a drug, the Court cannot define with any clarity who the analyst is. Outside this narrow category, the range of other scientific tests that may be affected by the Court’s new confrontation right is staggering. See, e. g., Comment, Toward a Definition of “Testimonial”: How Autopsy Reports Do Not Embody the Qualities of a Testimonial Statement, 96 Cal. L. Rev. 1093, 1094, 1115 (2008) (noting that every court post -Crawford has held that autopsy reports are not testimonial, and warning that a contrary rule would “effectively functio[n] as a statute of limitations for murder”).

2

It is difficult to confine at this point the damage the Court’s holding will do in other contexts. Consider just two — establishing the chain of custody and authenticating a copy of a document.

It is the obligation of the prosecution to establish the chain of custody for evidence sent to testing laboratories — that is, to establish “the identity and integrity of physical evidence *336by tracing its continuous whereabouts.” 23 C. J. S., Criminal Law §1142, p. 66 (2006). Meeting this obligation requires representations — that one officer retrieved the evidence from the crime scene, that a second officer checked it into an evidence locker, that a third officer verified the locker’s seal was intact, and so forth. The iron logic of which the Court is so enamored would seem to require in-court testimony from each human link in the chain of custody. That, of course, has never been the law. See, e. g., United States v. Lott, 854 F. 2d 244, 250 (CA7 1988) (“[G]aps in the chain [of custody] normally go to the weight of the evidence rather than its admissibility”); 29A Am. Jur. 2d, Evidence § 962, p. 269 (2008) (“The fact that one of the persons in control of a fungible substance does not testify at trial does not, without more, make the substance or testimony relating to it inadmissible”); 23 C. J. S., supra, § 1142, at 67 (“It is generally not necessary that every witness who handled the evidence testify”).

It is no answer for the Court to say that “[i]t is up to the prosecution to decide what steps in the chain of custody are so crucial as to require evidence.” Ante, at 311, n. 1. The case itself determines which links in the chain are crucial— not the prosecution. In any number of cases, the crucial link in the chain will not be available to testify, and so the evidence will be excluded for lack of a proper foundation.

Consider another context in which the Court’s holding may cause disruption: The long-accepted practice of authenticating copies of documents by means of a certificate from the document’s custodian stating that the copy is accurate. See, e.g., Fed. Rule Evid. 902(4) (in order to be self-authenticating, a copy of a public record must be “certified as correct by the custodian”); Rule 902(11) (business record must be “accompanied by a written declaration of its custodian”). Under one possible reading of the Court’s opinion, recordkeepers will be required to testify. So far, courts have not read Crawford and Davis to impose this largely *337meaningless requirement. See, e. g., United States v. Adefehinti, 510 F. 3d 319, 327-328 (CADC 2008) (certificates authenticating bank records may be admitted without confrontation); United States v. Ellis, 460 F. 3d 920, 927 (CA7 2006) (certificate authenticating hospital records). But the breadth of the Court’s ruling today, and its undefined scope, may well be such that these courts now must be deemed to have erred. The risk of that consequence ought to tell us that something is very wrong with the Court’s analysis.

Because the Court is driven by nothing more than a wooden application of the Crawford and Davis definition of “testimonial,” divorced from any guidance from history, precedent, or common sense, there is no way to predict the future applications of today’s holding. Surely part of the justification for the Court’s formalism must lie in its predictability. There is nothing predictable here, however, other than the uncertainty and disruption that now must ensue.

B

With no precedent to guide us, let us assume that the Court’s analyst is the person who interprets the machine’s printout. This result makes no sense. The Confrontation Clause is not designed, and does not serve, to detect errors in scientific tests. That should instead be done by conducting a new test. Or, if a new test is impossible, the defendant may call his own expert to explain to the jury the test’s flaws and the dangers of relying on it. And if, in an extraordinary case, the particular analyst’s testimony is necessary to the defense, then, of course, the defendant may subpoena the analyst. The Court frets that the defendant may be unable to do so “when the [analyst] is unavailable or simply refuses to appear.” Ante, at 324. But laboratory analysts are not difficult to locate or to compel. As discussed below, analysts already devote considerable time to appearing in court when subpoenaed to do so. See Part I-C, infra; see also Brief for State of Alabama et al. as Amici Curiae 26-28. Neither the *338Court, petitioner, nor amici offer any reason to believe that defendants have trouble subpoenaing analysts in cases where the analysts’ in-court testimony is necessary.

The facts of this case illustrate the formalistic and point-' less nature of the Court’s reading of the Clause. Petitioner knew, well in advance of trial, that the Commonwealth would introduce the tests against him. The bags of cocaine were in court, available for him to test, and entered into evidence. Yet petitioner made no effort, before or during trial, to mount a defense against the analysts’ results. Petitioner could have challenged the tests’ reliability by seeking discovery concerning the testing methods used or the qualifications of the laboratory analysts. See Mass. Rule Crim. Proc. 14(a)(2) (2009). He did not do so. Petitioner could have sought to conduct his own test. See Rule 41. Again, he did not seek a test; indeed, he did not argue that the drug was not cocaine. Rather than dispute the authenticity of the samples tested or the accuracy of the tests performed, petitioner argued to the jury that the prosecution had not shown that he had possessed or dealt in the drugs.

Despite not having prepared a defense to the analysts’ results, petitioner’s counsel made what can only be described as a pro forma objection to admitting the results without in-court testimony, presumably from one particular analyst. Today the Court, by deciding that this objection should have been sustained, transforms the Confrontation Clause from a sensible procedural protection into a distortion of the criminal justice system.

It is difficult to perceive how the Court’s holding will advance the purposes of the Confrontation Clause. One purpose of confrontation is to impress upon witnesses the gravity of their conduct. See Coy v. Iowa, 487 U. S. 1012, 1019-1020 (1988). A witness, when brought to face the person his or her words condemn, might refine, reformulate, reconsider, or even recant earlier statements. See ibid. A further purpose is to alleviate the danger of one-sided inter*339rogations by adversarial government officials who might distort a witness’ testimony. The Clause guards against this danger by bringing the interrogation into the more neutral and public forum of the courtroom. See Maryland v. Craig, 497 U. S. 836, 869-870 (1990) (Scalia, J., dissenting) (discussing the “value of the confrontation right in guarding against a child’s distorted or coerced recollections”); see also Comment, 96 Cal. L. Rev., at 1120-1122 (“During private law-enforcement questioning, police officers or prosecutors can exert pressure on the witness without a high risk of being discovered. Courtroom questioning, in contrast, is public and performed in front of the jury, judge and defendant. Pressure is therefore harder to exert in court”).

But neither purpose is served by the rule the Court announces today. It is not plausible that a laboratory analyst will retract his or her prior conclusion upon catching sight of the defendant the result condemns. After all, the analyst is far removed from the particular defendant and, indeed, claims no personal knowledge of the defendant’s guilt. And an analyst performs hundreds if not thousands of tests each year and will not remember a particular test or the link it had to the defendant.

This is not to say that analysts are infallible. They are not. It may well be that if the State does not introduce the machine printout or the raw results of a laboratory analysis; if it does not call an expert to interpret a test, particularly if that test is complex or little known; if it does not establish the chain of custody and the reliability of the laboratory; then the State will have failed to meet its burden of proof. That result follows because the State must prove its case beyond a reasonable doubt, without relying on presumptions, unreliable hearsay, and the like. See United States v. United States Gypsum Co., 438 U. S. 422, 446 (1978) (refusing to permit a “ ‘conclusive presumption [of intent],’ ” which “ ‘would effectively eliminate intent as an ingredient of the offense’ ” (quoting Morissette v. United States, 342 U. S. 246, 275 *340(1952))). The State must permit the defendant to challenge the analyst’s result. See Holmes v. South Carolina, 547 U. S. 319, 331 (2006) (affirming the defendant’s right to “have a meaningful opportunity to present a complete defense” (internal quotation marks omitted)). The rules of evidence, including those governing reliability under hearsay principles and the latitude to be given expert witnesses; the rules against irrebutable presumptions; and the overriding principle that the prosecution must make its case beyond a reasonable doubt — all these are part of the protections for the accused. The States, however, have some latitude in determining how these rules should be defined.

The Confrontation Clause addresses who must testify. It simply does not follow, however, that this clause, in lieu of the other rules set forth above, controls who the prosecution must call on every issue. Suppose, for instance, that the defense challenges the procedures for a secure chain of custody for evidence sent to a laboratory and then returned to the police. The defense has the right to call its own witnesses to show that the chain of custody is not secure. But that does not mean it can demand that, in the prosecution’s case in chief, each person who is in the chain of custody — and who had an undoubted opportunity to taint or tamper with the evidence — must be called by the prosecution under the Confrontation Clause. And the same is true with laboratory technicians.

The Confrontation Clause is simply not needed for these matters. Where, as here, the defendant does not even dispute the accuracy of the analyst’s work, confrontation adds nothing.

C

For the sake of these negligible benefits, the Court threatens to disrupt forensic investigations across the country and to put prosecutions nationwide at risk of dismissal based on erratic, all-too-frequent instances when a particular laboratory technician, now invested by the Court’s new constitu*341tional designation as the analyst, simply does not or cannot appear.

Consider first the costs today’s decision imposes on criminal trials. Our own Court enjoys weeks, often months, of notice before cases are argued. We receive briefs well in advance. The argument itself is ordered. A busy trial court, by contrast, must consider not only attorneys’ schedules but also those of witnesses and juries. Trial courts have huge caseloads to be processed within strict time limits. Some cases may unexpectedly plead out at the last minute; others, just as unexpectedly, may not. Some juries stay out longer than predicted; others must be reconstituted. An analyst cannot hope to be the trial court’s top priority in scheduling. The analyst must instead face the prospect of waiting for days in a hallway outside the courtroom before being called to offer testimony that will consist of little more than a rote recital of the written report. See Part I-B, supra.

As matters stood before today’s opinion, analysts already spent considerable time appearing as witnesses in those few cases where the defendant, unlike petitioner in this case, contested the analyst’s result and subpoenaed the analyst. See Brief for State of Alabama et al. as Amici Curiae 26-28 (testifying takes time); ante, at 328 (before today’s opinion, it was “‘almost always the case that [analysts’ certificates] [we]re admitted without objection’ ” in Massachusetts courts). By requiring analysts also to appear in the far greater number of cases where defendants do not dispute the analyst’s result, the Court imposes enormous costs on the administration of justice.

Setting aside, for a moment, all the other crimes for which scientific evidence is required, consider the costs the Court’s ruling will impose on state drug prosecutions alone. In 2004, the most recent year for which data are available, drug possession and trafficking resulted in 362,850 felony convictions in state courts across the country. See Dept, of Justice, Bureau of Justice Statistics, M. Durose & P. Langan, *342Felony Sentences in State Courts, 2004, p. 2 (July 2007). Roughly 95% of those convictions were products of plea bargains, see id., at 1, which means that state courts saw more than 18,000 drug trials in a single year.

The analysts responsible for testing the drugs at issue in those cases now bear a crushing burden. For example, the district attorney in Philadelphia prosecuted 25,000 drug crimes in 2007. Brief for National Dist. Attorneys Association et al. as Amici Curiae 12-13. Assuming that number remains the same, and assuming that 95% of the cases end in a plea bargain, each of the city’s 18 drug analysts, ibid., will be required to testify in more than 69 trials next year. Cleveland’s district attorney prosecuted 14,000 drug crimes in 2007. Ibid. Assuming that number holds, and that 95% of the eases end in a plea bargain, each of the city’s six drug analysts (two of whom work only part time) must testify in 117 drug cases next year. Id., at 13.

The Federal Government may face even graver difficulties than the States because its operations are so widespread. For example, the Federal Bureau of Investigation (FBI) laboratory at Quantico, Virginia, supports federal, state, and local investigations across the country. Its 500 employees conduct over 1 million scientific tests each year. Dept, of Justice, FBI Laboratory 2007, Message from the FBI Laboratory Director, http://www.fbi.gov/hq/lab/lab2007/ labannual07.pdf (as visited June 22, 2009, and available in Clerk of Court’s ease file). The Court’s decision means that before any of those million tests reaches a jury, at least one of the laboratory’s analysts must board a plane, find his or her way to an unfamiliar courthouse, and sit there waiting to read aloud notes made months ago.

The Court purchases its meddling with the Confrontation Clause at a dear price, a price not measured in taxpayer dollars alone. Guilty defendants will go free, on the most technical grounds, as a direct result of today’s decision, adding nothing to the truth-finding process. The analyst will *343not always make it to the courthouse in time. He or she may be ill; may be out of the country; may be unable to travel because of inclement weather; or may at that very moment be waiting outside some other courtroom for another defendant to exercise the right the Court invents today. If for any reason the analyst cannot make it to the courthouse in time, then, the Court holds, the jury cannot learn of the analyst’s findings (unless, by some unlikely turn of events, the defendant previously cross:examined the analyst). Ante, at 309. The result, in many cases, will be that the prosecution cannot meet its burden of proof, and the guilty defendant goes free on a technicality that, because it results in an acquittal, cannot be reviewed on appeal.

The Court’s holding is a windfall to defendants, one that is unjustified by any demonstrated deficiency in trials, any well-understood historical requirement, or any established constitutional precedent.

II

All of the problems with today’s decision — the imprecise definition of “analyst,” the lack of any perceptible benefit, the heavy societal costs — would be of no moment if the Constitution did, in fact, require the Court to rule as it does today. But the Constitution does not.

The Court’s fundamental mistake is to read the Confrontation Clause as referring to a kind of out-of-court statement— namely, a testimonial statement — that must be excluded from evidence. The Clause does not refer to kinds of statements. Nor does the Clause contain the word “testimonial.” The text, instead, refers to kinds of persons, namely, to “witnesses against” the defendant. Laboratory analysts are not “witnesses against” the defendant as those words would have been understood at the framing. There is simply no authority for this proposition.

Instead, the Clause refers to a conventional “witness”— meaning one who witnesses (that is, perceives) an event that gives him or her personal knowledge of some aspect of the *344defendant’s guilt. Both Crawford and Davis concerned just this kind of ordinary witness — and nothing in the Confrontation Clause’s text, history, or precedent justifies the Court’s decision to expand those cases.

A

The Clause states: “In all criminal prosecutions, the accused shall enjoy the right ... to be confronted with the witnesses against him.” U. S. Const., Amdt. 6. Though there is “virtually no evidence of what the drafters of the Confrontation Clause intended it to mean,” White v. Illinois, 502 U. S. 346, 359 (1992) (Thomas, J., concurring in part and concurring in judgment), it is certain the Framers did not contemplate that an analyst who conducts a scientific test far removed from the crime would be considered a “witnes[s] against” the defendant.

The Framers were concerned with a typical witness — one who perceived an event that gave rise to a personal belief in some aspect of the defendant’s guilt. There is no evidence that the Framers understood the Clause to extend to unconventional witnesses. As discussed below, there is significant evidence to the contrary. See Part II-B, infra. In these circumstances, the historical evidence in support of the Court’s position is “ Too meager ... to form, a solid basis in history, preceding and contemporaneous with the framing of the Constitution.’” Boumediene v. Bush, 553 U. S. 723, 752 (2008) (quoting Reid v. Covert, 354 U. S. 1, 64 (1957) (Frankfurter, J., concurring in result)). The Court goes dangerously wrong when it bases its constitutional interpretation upon historical guesswork.

The infamous treason trial of Sir Walter Raleigh provides excellent examples of the kinds of witnesses to whom the Confrontation Clause refers. Raleigh’s Case, 2 How. St. Tr. 1 (1603); see Crawford, 541 U. S., at 44-45 (Raleigh’s trial informs our understanding of the Clause because it was, at the time of the framing, one of the “most notorious in*345stances” of the abuse of witnesses’ out-of-court statements); ante, at 315 (same). Raleigh’s accusers claimed to have heard Raleigh speak treason, so they were witnesses in the conventional sense. We should limit the Confrontation Clause to witnesses like those in Raleigh’s trial.

The Court today expands the Clause to include laboratory analysts, but analysts differ from ordinary witnesses in at least three significant ways. First, a conventional witness recalls events observed in the past, while an analyst’s report contains near-contemporaneous observations of the test. An observation recorded at the time it is made is unlike the usual act of testifying. A typical witness must recall a previous event that he or she perceived just once, and thus may have misperceived or misremembered. But an analyst making a contemporaneous observation need not rely on memory; he or she instead reports the observations at the time they are made. We gave this consideration substantial weight in Davis. There, the “primary purpose” of the victim’s 911 call was “to enable police assistance to meet an ongoing emergency,” rather than “to establish or prove past events potentially relevant to later criminal prosecution.” 547 U. S., at 822, 827. See also People v. Geier, 41 Cal. 4th 555, 605-609, 161 P. 3d 104, 139-141 (2007). The Court cites no authority for its holding that an observation recorded at the time it is made is an act of “witness[ing]” for purposes of the Confrontation Clause.

Second, an analyst observes neither the crime nor any human action related to it. Often, the analyst does not know the defendant’s identity, much less have personal knowledge of an aspect of the defendant’s guilt. The analyst’s distance from the crime and the defendant, in both space and time, suggests the analyst is not a witness against the defendant in the conventional sense.

Third, a conventional witness responds to questions under interrogation. See, e.g., Raleigh’s Case, supra, at 15-20. But laboratory tests are conducted according to scientific *346protocols; they are not dependent upon or controlled by interrogation of any sort. Put differently, out-of-court statements should only “require confrontation if they are produced by, or with the involvement of, adversarial government officials responsible for investigating and prosecuting crime.” Comment, 96 Cal. L. Rev., at 1118. There is no indication that the analysts here — who work for the State Laboratory Institute, a division of the Massachusetts Department of Public Health — were adversarial to petitioner. Nor is there any evidence that adversarial officials played a role in formulating the analysts’ certificates.

Rather than acknowledge that it expands the Confrontation Clause beyond conventional witnesses, the Court relies on our recent opinions in Crawford and Davis. Ante, at 309-311. The Court assumes, with little analysis, that Crawford and Davis extended the Clause to any person who makes a “testimonial” statement. But the Court’s confident tone cannot disguise the thinness of these two reeds. Neither Crawford nor Davis considered whether the Clause extends to persons far removed from the crime who have no connection to the defendant. Instead, those cases concerned conventional witnesses. Davis, supra, at 826-830 (witnesses were victims of defendants’ assaults); Crawford, supra, at 38 (witness saw defendant stab victim).

It is true that Crawford and Davis employed the term “testimonial,” and thereby suggested that any testimonial statement, by any person, no matter how distant from the defendant and the crime, is subject to the Confrontation Clause.. But that suggestion was not part of the holding of Crawford or Davis. Those opinions used the adjective “testimonial” to avoid the awkward phrasing required by reusing the noun “witness.” The Court today transforms that turn of phrase into a new and sweeping legal rule, by holding that anyone who makes a formal statement for the purpose of later prosecution — no matter how removed from the crime— must be considered a “witnes[s] against” the defendant. *347Ante, at 309-311. The Court cites no authority to justify this expansive new interpretation.

B

No historical evidence supports the Court’s conclusion that the Confrontation Clause was understood to extend beyond conventional witnesses to include analysts who conduct scientific tests far removed from the crime and the defendant. Indeed, what little evidence there is contradicts this interpretation.

Though the Framers had no forensic scientists, they did use another kind of unconventional witness — the copyist. A copyist’s work may be as essential to a criminal prosecution as the forensic analyst’s. To convict a man of bigamy, for example, the State often requires his marriage records. See, e. g., Williams v. State, 54 Ala. 131, 134, 135 (1875); State v. Potter, 52 Vt. 33, 38 (1879). But if the original records cannot be taken from the archive, the prosecution must rely on copies of those records, made for the purpose of introducing the copies into evidence at trial. See ibid. In that case, the copyist’s honesty and diligence are just as important as the analyst’s here. If the copyist falsifies a copy, or even misspells a name or transposes a date, those flaws could lead the jury to convict. Because so much depends on his or her honesty and diligence, the copyist often prepares an affidavit certifying that the copy is true and accurate.

Such a certificate is beyond question a testimonial statement under the Court’s definition: It is a formal out-of-court statement offered for the truth of two matters (the copyist’s honesty and the copy’s accuracy), and it is prepared for a criminal prosecution.

During the Framers’ era copyists’ affidavits were accepted without hesitation by American courts. See, e. g., United States v. Percheman, 7 Pet. 51, 85 (1833) (opinion for the Court by Marshall, C. J.); see also Advisory Committee’s Note on Fed. Rule Evid. 902(4), 28 U. S. C. App., p. 390 (“The *348common law . . . recognized the procedure of authenticating copies of public records by certificate”); 5 J. Wigmore, Evidence §§1677, 1678 (J. Chadbourn rev. 1974). And courts admitted copyists’ affidavits in criminal as well as civil trials. See Williams, supra; Potter, supra. This demonstrates that the framing generation, in contrast to the Court today, did not consider the Confrontation Clause to require in-court confrontation of unconventional authors of testimonial statements.

The Court attempts to explain away this historical exception to its rule by noting that a copyist’s authority is “narrowly circumscribed.” Ante, at 322. But the Court does not explain why that matters, nor, if it does matter, why laboratory analysts’ authority should not also be deemed “narrowly circumscribed” so that they, too, may be excused from testifying. And drawing these fine distinctions cannot be squared with the Court’s avowed allegiance to formalism. Determining whether a witness’ authority is “narrowly circumscribed” has nothing to do with Crawford’s testimonial framework. It instead appears much closer to the pre-Crawford rule of Ohio v. Roberts, under which a statement could be admitted without testimony if it “bears adequate indicia of reliability.” 448 U. S., at 66 (internal quotation marks omitted).

In keeping with the traditional understanding of the Confrontation Clause, this Court in Dowdell v. United States, 221 U. S. 325 (1911), rejected a challenge to the use of certificates, sworn out by a clerk of court, a trial judge, and a court reporter, stating that defendants had been present at trial. Those certificates, like a copyist’s certificate, met every requi