Abstract
Criminal trials vary widely in the number of witnesses required to bring a single forensic result before a jury. A DNA comparison may require a procession of collectors, custodians, technicians, and analysts, while a set of cell phone records may enter on a written certification and a single expert. This paper proposes that such variation is not random. It follows from three features of the evidence: whether the evidence is a physical object or a record, whether the statements describing it are testimonial or nontestimonial under current confrontation doctrine, and whether the analysis that produces a result is performed by a single examiner or by a divided pipeline of personnel. The paper defines these three axes, sets out the classes they produce, and places the principal categories of forensic and documentary evidence within the resulting typology. It concludes that the testimonial burden of a piece of evidence, here called its foundation load, varies largely independently of the weight of the conclusion it supports, and that recognizing this independence is necessary for evaluating witness lists, stipulation proposals, and reform efforts on their merits.
1. Introduction
Observers of criminal trials frequently remark on the number of witnesses who testify briefly, say little of apparent importance, and leave the stand without cross-examination. A property room custodian confirms the date an item was logged. A courier confirms the date it was delivered to a laboratory. A technician confirms that a sample was extracted and passed to the next stage. The testimony that the jury will remember, such as an analyst’s statement that a profile matched the defendant, arrives only after this sequence is complete.
The same trial may present other evidence with no such procession. Telephone records may be admitted on the certification of a carrier’s records custodian who never appears. A bank’s account statements may enter the same way. An expert may then testify at length about what those records mean, and that testimony may prove more consequential than anything the procession produced.
The contrast raises a practical question that the literature on evidence has addressed mostly one doctrine at a time: what determines how many witnesses a piece of evidence requires? Authentication, chain of custody, the hearsay exceptions for business and public records, the rules governing expert testimony, and the Confrontation Clause each answer part of the question. What is lacking is a framework that combines their answers into a single account of why some evidence arrives with a crowd and other evidence arrives alone.
This paper offers such a framework. It introduces the term foundation load for the total testimonial burden a piece of evidence must bear before its result can be presented, and it identifies three axes along which that burden varies. The framework is descriptive. It does not argue that the present allocation of burdens is correct, a question taken up in White Papers 3, 4, and 6. It establishes the vocabulary on which those later papers depend.
2. The Concept of Foundation Load
Foundation is the showing required before evidence may be considered by the fact-finder. At minimum it requires authentication: evidence “sufficient to support a finding that the item is what the proponent claims it is.”[1] For physical items whose significance depends on their condition, it often requires a showing that the item has not been altered or substituted between collection and trial. For analytical results, it requires that the analysis be presented through a witness or record the law will accept, and in the case of expert opinion, that the opinion rest on a reliable method reliably applied.[2]
Foundation load, as used here, is the aggregate of these requirements expressed in terms of persons. It asks how many human beings must appear, or be accounted for by an accepted substitute, before the result can speak. The measure is deliberately practical. It corresponds to what attorneys prepare when they draft witness lists, to what laboratories lose when analysts are summoned to court, and to what trial observers see.
Two clarifications are necessary. First, foundation load is not a measure of reliability. A heavy foundation load may accompany highly reliable evidence, and a light one may accompany evidence of doubtful value. Second, foundation load is not fixed by the evidence alone. It depends on the governing law of the jurisdiction, on whether the parties stipulate, and on laboratory and agency practices that may add or remove links. The typology describes the default burden that evidence of a given kind carries before those variables are applied.
3. The Three Axes
3.1 Physical Object Versus Record
The first axis distinguishes evidence that is a tangible thing from evidence that is a record.
A physical object can be altered, contaminated, substituted, degraded, or consumed. Its evidentiary meaning depends on its condition, and its condition can change. The law therefore requires, in varying degrees, a showing that the item offered at trial is the item collected and that it has not been changed in any respect bearing on the question at issue. Where an item is unique and readily identifiable, such as a firearm with a serial number, a witness who recognizes it may suffice. Where an item is fungible or susceptible to tampering, such as a powder, a fluid, or a swab, courts generally require a chain of custody accounting for its handling.[3] The classic treatment of the subject identifies the chain’s purpose as establishing both identity and unchanged condition, with the stringency of the showing rising with the item’s susceptibility to alteration.[4]
A record carries its meaning in its content rather than its material condition. Its reliability is ordinarily grounded not in an account of its physical handling but in the regularity of the institution that made and kept it. The hearsay exception for records of a regularly conducted activity rests on this premise: a record made at or near the time of an event, by a person with knowledge, in the course of a regular practice, is trusted because the institution depends on its accuracy for its own purposes.[5] Official records rest on a similar premise with respect to public agencies.[6] Since 2000, the Federal Rules of Evidence have allowed business records to be authenticated by written certification without a live custodian,[7] and since 2017 they have allowed electronic records and data copied from devices to be authenticated by certification, including certification by hash value comparison.[8]
The distinction is not always clean. A cell phone is a physical object; the data extracted from it is a record; the extraction is a process performed on the object that produces the record. Such hybrids are treated in Section 5.
3.2 Testimonial Versus Nontestimonial
The second axis concerns the constitutional status of the statements through which evidence reaches the jury.
Since Crawford v. Washington, the Confrontation Clause of the Sixth Amendment has been read to bar the admission of testimonial statements by a witness who does not appear at trial unless the witness is unavailable and the defendant had a prior opportunity for cross-examination.[9] The Court declined in Crawford to define “testimonial” comprehensively, but its subsequent decisions have made the purpose for which a statement was made the governing consideration.[10]
For forensic evidence, the decisive cases are three. In Melendez-Diaz v. Massachusetts, the Court held that sworn certificates of analysis reporting that a seized substance was cocaine were testimonial and could not be admitted without the analysts’ live testimony.[11] In Bullcoming v. New Mexico, the Court held that a blood alcohol report could not be introduced through a “surrogate” analyst who had neither performed nor observed the test.[12] In Smith v. Arizona, the Court held that when a substitute expert conveys an absent analyst’s statements as the basis for the expert’s own opinion, and those statements support the opinion only if true, they are offered for their truth; whether they are testimonial remained a separate question, which the Court remanded.[13] Between these decisions, Williams v. Illinois produced a fractured result on a DNA profile generated by an outside laboratory, with no single rationale commanding a majority.[14]
The same line of cases distinguishes business and public records. The Melendez-Diaz majority observed that such records are generally admissible without confrontation because they are created for the administration of an entity’s affairs rather than for the purpose of establishing or proving some fact at trial, and it cautioned that records prepared specifically for use at trial do not acquire nontestimonial status merely by being kept regularly.[15]
This axis therefore turns on purpose of creation. A laboratory report prepared at the request of police for use in a prosecution is, under current doctrine, presumptively testimonial. A carrier’s call detail records, kept for billing and network management, are presumptively not. The distinction does not depend on which kind of statement is more accurate, a point White Paper 3 examines in detail.
3.3 Single Analyst Versus Pipeline
The third axis concerns how an analytical result is produced.
In a single-analyst workflow, one examiner performs the analysis from receipt of the item through the conclusion, or at least performs every step on which the conclusion depends. Traditional latent print comparison, firearms and toolmark examination, and questioned document examination commonly follow this pattern, as do many drug chemistry analyses.
In a pipeline workflow, the analysis is divided among personnel who perform successive stages. Forensic DNA typing is the principal example: extraction, quantitation, amplification, capillary electrophoresis, and interpretation may be performed by different technicians and analysts, each stage producing an intermediate record consumed by the next.[16] Toxicology in high-volume laboratories frequently follows a similar division between screening, confirmation, and review.
Where the resulting report is testimonial, the pipeline raises the question of which participants must testify. Bullcoming established that a surrogate who did not perform or observe the test is insufficient, but it did not require every participant to appear, and the Melendez-Diaz majority stated that not everyone whose testimony may be relevant to authenticity or accuracy must appear as part of the prosecution’s case, gaps in the chain going ordinarily to weight rather than admissibility.[17] The practical burden of a pipeline therefore depends on how a jurisdiction resolves which stages generate testimonial statements and which participants the defense may insist on confronting. The more stages that produce statements the prosecution relies on for their truth, the heavier the load.
4. The Classes Produced by the Axes
The three axes, taken as binary, produce eight combinations. Not all are equally populated, and one is nearly empty, but setting them out makes the structure of the typology visible. The notation used throughout the suite designates physical evidence P and record evidence R; testimonial status T and nontestimonial status N; and single-analyst work S and pipeline work L (for “line”).
| Class | Axes | Characteristic foundation load | Representative evidence |
|---|---|---|---|
| I | P–T–L | Heaviest: full custody chain plus multiple testifying analysts | DNA typing, particularly mixtures; laboratory toxicology |
| II | P–T–S | Heavy custody, one principal analyst | Drug chemistry; firearms and toolmarks; latent prints; handwriting |
| III | P–N–S | Custody or recognition only; no forensic report | Weapons, clothing, or items identified by a percipient witness |
| IV | P–N–L | Rare | Items processed by non-law-enforcement institutions for their own purposes, such as hospital clinical testing |
| V | R–N–S | Light record foundation, heavy expert analysis | Cell site analysis; forensic accounting; summary charts |
| VI | R–N–L | Light record foundation; divided institutional processing | Financial institution records produced through automated systems; carrier data compiled by several departments |
| VII | R–T–S | Author must testify | Certificates and affidavits prepared for litigation; investigative reports |
| VIII | R–T–L | Contested | Autopsy reports; sexual assault examination records, in jurisdictions treating them as testimonial |
Three features of the table deserve comment.
First, Classes I and II carry the procession that trial observers notice. Their load has two sources that compound: the physical axis generates custody witnesses, and the testimonial axis converts the analyst from an optional witness into a required one.
Second, Class V is the mirror image of Class I. Its foundation is light because the underlying data is a nontestimonial record authenticated by certification. Yet the interpretive work performed on that data by an expert may be extensive and may bear directly on the central disputed fact, such as the defendant’s location at the time of the offense.
Third, Class VIII is where doctrine remains unsettled. An autopsy report is both a medical record, prepared under a statutory duty and used for purposes beyond prosecution, and a document often prepared in anticipation that a death will be the subject of criminal proceedings. Courts have divided on its status.[18] Sexual assault examination records present a similar ambiguity between treatment and evidence collection. Because the classification of these items is itself in dispute, their foundation load is the least predictable of any category.
5. Placement of Common Evidence Types
5.1 DNA
DNA evidence occupies Class I and represents the heaviest foundation load in ordinary criminal practice. Every axis pushes in the same direction. The evidence is physical, often fungible in appearance, and highly sensitive to contamination, so custody from collection through laboratory receipt must be accounted for. It is frequently consumed in whole or in part during testing, which forecloses independent retesting and places greater weight on the documentation of the original analysis. The laboratory workflow is divided, and the final report is prepared for prosecution.
Mixture interpretation adds a further stage. Where a sample contains DNA from more than one contributor, the conclusion depends on judgments about the number of contributors, the separation of their profiles, and the statistical weight assigned to an inclusion, judgments that research has shown to be susceptible to contextual influence and to vary among analysts.[19] The use of probabilistic genotyping software introduces an additional analytical layer whose operation may itself require explanation.
Notably, the conclusion that emerges from this heavy foundation is narrow. A DNA result speaks to the source of biological material. It does not by itself establish when or how that material was deposited. The heaviest foundation load in the typology thus supports a conclusion of limited interpretive breadth.
5.2 Drug Chemistry and Toxicology
Controlled substance analysis and blood alcohol testing were the subjects of Melendez-Diaz and Bullcoming and remain the cases most directly governed by them. Drug chemistry is commonly a Class II item: a physical sample, a testimonial report, and a single analyst performing both screening and confirmatory testing. Toxicology frequently shifts toward Class I where laboratories divide sample preparation, instrumental analysis, and review among different personnel. Both involve fungible samples for which chain of custody carries particular weight. The interpretive breadth of the conclusion is ordinarily narrow: the identity and quantity of a substance, or the concentration of alcohol or a drug in a sample.
5.3 Pattern Evidence
Firearms and toolmark identification, latent print comparison, and handwriting examination occupy Class II. The items are physical and require custody or recognition foundation, though firearms and many documents are unique enough that identification rather than a full chain may suffice. The comparison is ordinarily performed by a single examiner, sometimes subject to verification by a second, and the report is prepared for prosecution.
Pattern evidence differs from DNA in the interpretive breadth of its conclusions and in the nature of the scrutiny they have drawn. The conclusions rest substantially on examiner judgment, and major reviews have questioned the empirical foundations of source attributions in several pattern disciplines.[20] The foundation load of pattern evidence is thus moderate, while the interpretive load it places on a single witness is considerable.
5.4 Records-Based Expertise
Cell site analysis and forensic accounting are Class V items. The underlying records, whether carrier call detail records and tower lists or bank statements and ledgers, are business records created for purposes unrelated to the prosecution and are commonly authenticated by certification.[21] No custody chain is required in the physical sense; the records are copies whose accuracy depends on institutional regularity.
The foundation is light and the analysis is heavy. A cell site analyst must interpret network architecture, sector orientation, and the factors that determine which tower serves a call, and the reliability of conclusions about location depends on the method used and on the precision claimed. A forensic accountant must reconstruct transactions, trace funds, and often present conclusions through summaries of voluminous records.[22] In both cases the expert witness may be the only person the jury sees in connection with the evidence, and that witness carries nearly the entire testimonial burden.
5.5 Digital Evidence
Evidence extracted from phones, computers, and storage media is a hybrid that does not fit a single cell. The device is a physical object, and its seizure and handling require custody foundation. The extracted data is a record, and the 2017 amendments to the Federal Rules permit its authentication by certification of a qualified person, including by hash value comparison demonstrating that the copy is identical to the original.[23] The extraction and analysis are often divided between an examiner who performs the extraction and an analyst who reviews and reports on its contents.
Digital evidence therefore reintroduces part of the procession that record evidence ordinarily avoids. Hash verification shortens the chain after extraction by allowing the integrity of the copy to be demonstrated mathematically rather than testimonially, but it cannot eliminate the physical custody of the device before extraction or the interpretive work of the analyst after it. Digital evidence is best described as physical at its origin and record-based thereafter, with a foundation load concentrated at the point of transition.
5.6 The Contested Middle
Autopsy reports and sexual assault examination records, discussed in Section 4, are the principal contested items. Their placement depends on how a jurisdiction resolves their testimonial status, and their foundation load varies accordingly from light, where they are treated as medical or official records, to heavy, where the examining physician or nurse must testify and may be unavailable years after the examination.
6. Observations
6.1 Foundation Load and Interpretive Breadth Vary Independently
The typology reveals that the evidence carrying the heaviest testimonial burden does not necessarily support the broadest conclusions. DNA, at the top of the load scale, speaks narrowly to the source of biological material. Cell site analysis, near the bottom, may speak broadly to a defendant’s movements over a period of hours. The number of witnesses a trial observer sees in connection with a piece of evidence is therefore a poor guide to how much that evidence contributes to the proof of guilt.
6.2 Purpose of Creation Does More Work Than Physical Nature
Of the three axes, the testimonial axis most sharply divides heavy from light loads. Physical evidence without a testimonial report (Class III) may require little more than a witness who recognizes the item. Record evidence prepared for prosecution (Class VII) requires its author. The constitutional line, drawn by reference to the purpose for which a statement was made, is the single largest determinant of whether an analyst must appear.
6.3 The Pipeline Multiplies Links, Not Witnesses to Each Link
Where a pipeline workflow produces a testimonial result, the additional witnesses it generates are witnesses to additional links, not additional witnesses to the same link. Each technician speaks to his or her own stage. This structure is developed in White Paper 2 under the heading of sequence, and it is the reason a long witness list in a Class I case should not be mistaken for corroboration.
6.4 Foundation Load Is Partly Chosen
Because foundation load depends on laboratory workflow, documentation practice, and the parties’ willingness to stipulate, it is not wholly determined by the nature of the evidence. A laboratory that assigns one analyst to perform or directly observe every stage of a DNA analysis reduces the load of its reports. An agency that documents custody transfers electronically, with verifiable timestamps, may shorten the chain that must be proved by live testimony. The typology describes the default; White Papers 4 and 6 examine the institutional choices that raise or lower it.
7. Limitations
The typology simplifies in several respects. It treats each axis as binary, although testimonial status is contested in some categories and many workflows combine single-analyst and pipeline features. It describes default burdens under federal rules and Supreme Court doctrine, while state rules on authentication, chain of custody, and notice-and-demand procedures vary considerably. It measures load in witnesses without weighting them by the length or difficulty of their testimony. And it does not account for the defense’s strategic choices, which may reduce load through stipulation or increase it by insisting on every available witness. The Foundation Load Index presented later in this volume refines the axes into graded dimensions to address the first and third of these limitations.
8. Conclusion
The number of witnesses required to bring a forensic or documentary result before a jury is governed by identifiable features of the evidence. Physical objects generate custody witnesses; records generally do not. Testimonial reports require their authors; nontestimonial records do not. Pipeline analysis multiplies the stages that may require accounting; single-analyst work concentrates them. The combination of these features determines the foundation load a piece of evidence carries, and that load bears no fixed relation to the reliability or importance of the conclusion it supports.
This framework gives courts, laboratories, attorneys, and observers a common vocabulary for describing what a witness list represents. The next paper applies that vocabulary to the question that a long witness list most often invites: whether many witnesses mean multiplied assurance, or only many links in a single chain.
Notes
[1] Fed. R. Evid. 901(a).
[2] Fed. R. Evid. 702, as amended in 2023 to clarify that the proponent must establish admissibility requirements by a preponderance of the evidence and that the expert’s opinion must reflect a reliable application of the principles and methods to the facts of the case.
[3] Mosteller (2020), § 213, discusses the distinction between items identifiable by unique characteristics and fungible items requiring a chain of custody.
[4] Giannelli (1983), pp. 530–535. The article remains the standard treatment of chain of custody as a component of authentication.
[5] Fed. R. Evid. 803(6). See also Wigmore (1940), on the historical rationale of regular entries.
[6] Fed. R. Evid. 803(8).
[7] Fed. R. Evid. 902(11) (domestic records) and 902(12) (foreign records in civil cases), added in 2000.
[8] Fed. R. Evid. 902(13) and 902(14), added in 2017. The advisory committee note to Rule 902(14) identifies hash value comparison as the typical method of certification.
[9] Crawford v. Washington (2004), pp. 53–54, 68.
[10] Davis v. Washington (2006), p. 822, articulated the “primary purpose” inquiry for statements to police, and later forensic cases have applied purpose-based reasoning to laboratory reports.
[11] Melendez-Diaz v. Massachusetts (2009), pp. 310–311.
[12] Bullcoming v. New Mexico (2011), p. 652.
[13] Smith v. Arizona (2024). The Court held that the absent analyst’s statements were offered for their truth and remanded for the state court to address whether they were testimonial.
[14] Williams v. Illinois (2012). A plurality concluded that the outside laboratory’s report was not offered for its truth and, alternatively, was not testimonial; Justice Thomas concurred in the judgment on the ground that the report lacked sufficient formality. Five justices rejected the plurality’s not-for-truth reasoning, which Smith later adopted as the Court’s holding.
[15] Melendez-Diaz v. Massachusetts (2009), pp. 321–324.
[16] Butler (2015) describes the stages of forensic DNA typing and the interpretive decisions at each.
[17] Melendez-Diaz v. Massachusetts (2009), p. 311 n.1.
[18] State appellate courts have divided on whether autopsy reports are testimonial, with results often turning on the circumstances of the particular examination and the degree of law enforcement involvement. The monograph’s Chapter 10 surveys the division.
[19] Dror and Hampikian (2011) reported that DNA analysts presented with the same mixture data, without contextual case information, reached differing conclusions. Butler (2015) discusses the interpretive challenges of mixtures generally.
[20] National Research Council (2009), pp. 7–8, 136–145; President’s Council of Advisors on Science and Technology (2016). See also Saks and Koehler (2005).
[21] Fed. R. Evid. 803(6), 902(11).
[22] Fed. R. Evid. 1006 permits summaries of voluminous records that cannot conveniently be examined in court.
[23] Fed. R. Evid. 902(14) and advisory committee note.
References
Bullcoming v. New Mexico, 564 U.S. 647 (2011).
Butler, J. M. (2015). Advanced topics in forensic DNA typing: Interpretation. Academic Press.
Crawford v. Washington, 541 U.S. 36 (2004).
Davis v. Washington, 547 U.S. 813 (2006).
Dror, I. E., & Hampikian, G. (2011). Subjectivity and bias in forensic DNA mixture interpretation. Science & Justice, 51(4), 204–208.
Fed. R. Evid. 702.
Fed. R. Evid. 803.
Fed. R. Evid. 901.
Fed. R. Evid. 902.
Fed. R. Evid. 1006.
Giannelli, P. C. (1983). Chain of custody and the handling of real evidence. American Criminal Law Review, 20(4), 527–568.
Melendez-Diaz v. Massachusetts, 557 U.S. 305 (2009).
Mosteller, R. P. (Ed.). (2020). McCormick on evidence (8th ed.). West Academic.
National Research Council. (2009). Strengthening forensic science in the United States: A path forward. National Academies Press.
President’s Council of Advisors on Science and Technology. (2016). Forensic science in criminal courts: Ensuring scientific validity of feature-comparison methods. Executive Office of the President.
Saks, M. J., & Koehler, J. J. (2005). The coming paradigm shift in forensic identification science. Science, 309(5736), 892–895.
Smith v. Arizona, 602 U.S. 779 (2024).
Wigmore, J. H. (1940). A treatise on the Anglo-American system of evidence in trials at common law (3rd ed.). Little, Brown.
Williams v. Illinois, 567 U.S. 50 (2012).
