United States v. Mitchell

Opinions of the United

2004 Decisions States Court of Appeals

for the Third Circuit

4-29-2004

USA v. Mitchell

Precedential or Non-Precedential: Precedential

Docket No. 02-2859

Follow this and additional works at: http://digitalcommons.law.villanova.edu/thirdcircuit_2004

Recommended Citation

"USA v. Mitchell" (2004). 2004 Decisions. Paper 734.

http://digitalcommons.law.villanova.edu/thirdcircuit_2004/734

This decision is brought to you for free and open access by the Opinions of the United States Court of Appeals for the Third Circuit at Villanova

University School of Law Digital Repository. It has been accepted for inclusion in 2004 Decisions by an authorized administrator of Villanova

University School of Law Digital Repository. For more information, please contact Benjamin.Carlson@law.villanova.edu.

PRECEDENTIAL Suite 540 West - Curtis Center

Independence Square West

IN THE UNITED STATES COURT OF Philadelphia, Pennsylvania 19106

APPEALS FOR THE THIRD CIRCUIT

____________________ Counsel for Appellant

NO. 02-2859

___________________ PATRICK L. MEEHAN

United States Attorney

UNITED STATES OF AMERICA

LAURIE MAGID

v. Deputy United States Attorney

for Policy and Appeals

BYRON MITCHELL MICHAEL L. LEVY

Assistant United States Attorney

Appellant ROBERT A. ZAUZM ER (Argued)

________________ Assistant United States Attorney

PAUL A. SARMOUSAKI

Assistant United States Attorney

On Appeal from the United States Senior Appellate Counsel

District Court for Eastern District of Pennsylvania

the Eastern District of Pennsylvania Suite 1250

(D.C. No. 96-cr-407-1) 615 Chestnut Street

District Judge: Philadelphia, Pennsylvania 19106

Honorable J. Curtis Joyner

__________________________ Counsel for Appellee

_______________________

Argued September 9, 2003

OPINION

Before: BARRY, BECKER and _______________________

GREENBERG, Circuit Judges.

TABLE OF CONTENTS

(Filed April 29, 2004)

I. Introduction . . . . . . . . . . . . . . . . . . . 3

MAUREEN KEARNEY ROWLEY

Chief Federal Defender

DAVID L. McCOLGIN II. Facts and Procedural History . . . . . 4

Supervising Appellate Attorney

ROBERT EPSTEIN (Argued) A. The Offense and Mitchell’s

Assistant Federal Defender First Trial and Appeal . . . . . . . . 4

Federal Court Division B. Latent Fingerprint Identification

Defender Association of Philadelphia

and the Daubert Hearing . . . . . . 5 Expert Testimony . . . . . . . . . . 24

1. The Field of Latent 1. Testability . . . . . . . . . . . . . 24

Fingerprint Identification . . 5

2. Peer Review . . . . . . . . . . . . 28

2. The Daubert Hearing . . . . . . 7

3. Error Rate . . . . . . . . . . . . . . 29

a. The Government’s

4. Maintenance of

Experts . . . . . . . . . . . . . . 7

Standards . . . . . . . . . . . . . . 32

b. Mitchell’s Experts . . . . 12

5. General Acceptance . . . . . . 32

c. Mitchell’s Exhibits . . . . 15

6. Relationship to

d. The Government’s Established Reliable

Rebuttal Witness . . . . . 15 Techniques . . . . . . . . . . . . 32

3. The District Court’s 7. Degree to Which the

Daubert and Judicial Expert Testifying Is

Notice Rulings . . . . . . . . . . 16 Qualified . . . . . . . . . . . . . . 33

C. Mitchell’s Second Trial . . . . . 17 8. Non-Judicial Uses . . . . . . . 34

1. The Government’s Case . . . 17 D. Application to the Record of

Core Daubert Principles . . . . . 35

2. Mitchell’s Case and

Cross-Examination of the E. Conclusion on the

Government’s Experts . . . . 19 Admissibility of the

Government’s Evidence . . . . . 38

D. Withholding of the NIJ

Solicitation and Mitchell’s

Post-Trial Motion . . . . . . . . . . 20

IV. Admissibility of Mitchell’s

E. This Appeal . . . . . . . . . . . . . . . 21 Expert Testimony . . . . . . . . . . . . . 39

A. Introduction . . . . . . . . . . . . . . 39

III. Admissibility of the B. Velasquez . . . . . . . . . . . . . . . . 39

Government’s Expert

C. The Parties’ Interpretations of

Testimony . . . . . . . . . . . . . . . . . . . 21

the District Court’s Rulings . . 40

A. Standard of Review . . . . . . . . 21

D. Discussion . . . . . . . . . . . . . . . . 43

B. Standard for Admissibility

under Rule 702 . . . . . . . . . . . . 22

V. The District Court’s

C. Application of Daubert

Declaration of Judicial Notice . . . 45

Factors to Government’s

2

A. Appropriateness of Judicial the application of the various Daubert

Notice . . . . . . . . . . . . . . . . . . . 45 factors to the prosecution’s expert

testimony. We conclude that the testimony

B. Harmless Error Analysis . . . . . 47

passes Daubert muster, and that there are

“good grounds,” id. at 590, for its

admission. In a related matter, we must

VI. Withholding of the NIJ

decide whether the District Court properly

Solicitation . . . . . . . . . . . . . . . . . . 48

took judicial notice that “human friction

A. Standard of Review ridges are unique and permanent

and Applicable Law . . . . . . . . 49 throughout the area of the friction ridge

skin, including small friction ridge areas,

B. Discussion . . . . . . . . . . . . . . . . 51

and that . . . human friction ridge skin

arrangements are unique and permanent.”

App. 1472a. We conclude that the District

VII. Admission of Alleged

Court erred in taking judicial notice, but

Prior Consistent Statements . . . . . 53

that the error was harmless.

We also consider Mitchell’s contention

VIII. Conclusion . . . . . . . . . . . . . . . . 55 that the District Court erroneously

excluded from trial significant portions of

his proffered expert testimony on the

APPENDIX: Colloquies with the unre liability of late nt fin gerp rint

District Court Regarding identification. Portions of the colloquies

Admissibility of Mitchell’s between the Court and counsel are less

Proposed Experts. . . . . . . . . . . . . . 55 than pellucid, but we are satisfied that

what the Court really did was to operate on

a three-tier theory of what expert

BECKER, Circuit Judge. testimony was admissible: allowing (1)

specific criticisms and (2) general

I. Introduction

reliability criticisms, but excluding (3)

This appeal by Byron Mitchell from a testimony about whether latent fingerprint

judgment in a criminal case raises identification is a “science.” Within that

important questions concerning the framework, the exclusion of evidence that

admissib ility of latent fingerprint latent fingerprint identification is a science

identification evidence under Fed. R. Evid. was proper under Kumho Tire Co. v.

702. We adjudicate on the basis of a Carmichael, 526 U.S. 137 (1999).

voluminous record developed at a Daubert

The final fingerprint-related issue

hearing, see Daubert v. Merrell Dow

concerns the putative withholding by the

Pharmaceuticals, Inc., 509 U.S. 579

government of a Department of Justice

(1993), and explore in considerable detail

solicitation for research proposals directed

3

at validating the reliability of latent First Trial and Appeal

f i n ge r p r int i d e n t if i c a ti o n . T h is

This case began in 1991 when two

solicitation, Mitchell contends, was not

men with handguns robbed an armored

only improperly and intentionally withheld

car employee of approximately $20,000

by the government in violation of its

as he entered a check cashing agency at

obligations under Brady v. Maryland, 373

29th Street and Girard Avenue in North

U.S. 83 (1963), but would have been

Philadelphia. The robbers then got into a

powerful evidence, not only substantively

beige car driven by a third person,

but also to impeach the government’s

engaging in gunfire with the armored car

expert witnesses who testified that latent

employees as they fled. The beige car,

fingerprint identification was a well-

which had been stolen about an hour

established discipline with a strong and

beforehand, was abandoned by the

well-verified foundation. The District

robbers roughly a mile from the agency.

Court concluded that the solicitation was

The government sought to prove at trial

not material under the “reasonable

that the robbers were William Robinson

probability of a different outcome”

(a/k/a “Bookie”) and Terrence Stewart

standard of Brady and its progeny. We

(a/k/a “T”), and that the getaway driver

agree.

was Mitchell. According to the

The remaining issue on appeal is government, the robbery had a fourth

whether plain error was committed by the participant, Kim Chester, who knew of

admission of testimony that a key the plans, helped case the robbery site,

government witness gave a statement to and assisted the others in spending the

the FBI and testified at a prior proceeding. proceeds of the robbery. Chester

Mitchell characterizes the admission of testified for the prosecution at Mitchell’s

this evidence as improper under the trial as an uncharged accomplice. Both

hearsay rules, Fed. R. Evid. 801, 802. We Robinson and Stewart died before trial,

conclude that testimony about the and thus Mitchell was the sole defendant.

existence of a statement is not itself a

Mitchell was charged with conspiracy

“statement”; that the testimony was not

to commit and commission of Hobbs Act

“offered . . . to prove the truth of the

robbery, 18 U.S.C. § 1951, and use of

matter asserted,” Fed. R. Evid. 801(c), and

and carrying a firearm during a crime of

thus not inadmissible under Fed. R. Evid.

violence, 18 U.S.C. § 924(c). In the first

802; and that, at all events, the plain error

trial, at which Mitchell was convicted of

standard is not met. We will therefore

all counts, the government introduced

affirm the judgment.

into evidence an anonymous note that

had been left in the front seat of the

abandoned beige car, apparently written

II. Facts and Procedural History

by someone who observed the robbers

A. The Offense and Mitchell’s exiting the beige car and getting into a

4

different car. The note read, “Light 1. The Field of Latent

green ZPJ-254. They changed cars; this Fingerprint Identification

is the other car.” On appeal, we held the

Criminals generally do not leave

note to be inadmissible hearsay not

behind full fingerprints on clean, flat

subject to any exception in Fed. R. Evid.

surfaces. Rather, they leave fragments

803. United States v. Mitchell, 145 F.3d

that are often distorted or marred by

572 (3d Cir. 1998). In view of the

artifacts, terms we explain in the

limited other evidence connecting

margin.1 These “latent” prints—from the

Mitchell to the robbery—Chester’s

Latin lateo, “to lie hidden,” because they

testimony was questionable, no robbery

are often not visible to the naked eye

proceeds were ever linked to Mitchell,

until dusted or otherwise revealed— are

and the fingerprints recovered from the

the typical grist for the fingerprint

beige getaway car were identified as

identification expert’s mill. Testimony at

Mitchell’s but in poor condition—we

the Daubert hearing suggested that the

concluded that admission of the

typical latent print is a fraction—perhaps

anonymous note was not harmless error.

1/5th— of the size of a full fingerprint.

Id. at 579-80. Accordingly, we vacated

App. 435a-436a. A “full” fingerprint is

Mitchell’s conviction and remanded for a

familiar to anyone who has been

new trial. Id.

fingerprinted for identification or law

B. Latent Fingerprint Identification enforcement reasons: It is the print made

and the Daubert Hearing by rolling the full surface of the fingertip

onto a fingerprint card or electronic

Prior to the retrial, the District Court

fingerprint capture device. (These prints

conducted a lengthy Daubert hearing on

are, for obvious reasons, also referred to

the admissibility under Fed. R. Evid. 702

as “rolled prints” or “full-rolled prints.”)

of the government’s expert testimony

A full set of full-rolled fingerprints on a

(and Mitchell’s counter-experts) on the

card—as would be taken during a police

identification of fingerprints found on

booking, for example—is known as a

the gear shift lever and driver’s side door

“ten-print card.” Ten-print cards usually

of the beige getaway car. This hearing

also have space at the bottom of the card

was to adjudicate a major attack mounted

by Mitchell on the government’s

fingerprint evidence. As with any expert 1

In the jargon, artifacts are generally

testimony, some background in the field

small amounts of dirt or grease that

and an introduction to the jargon is

masquerade as parts of the ridge

helpful, and so we discuss the field of

impressions seen in a fingerprint, while

latent fingerprint identification in general

distortions are produced by smudging or

before turning to the particulars of the

too much pressure in making the print,

Daubert hearing.

which tends to flatten the ridges on the

finger and obscure their detail.

5

for “flat impressions” or “plain where ridges terminate or bifurcate are

impressions,” where all four fingers of often referred to as “Galton points,”

the hand are pressed at once onto the whose eponym, Sir Francis Galton, first

card without rolling. developed a taxonomy for these points.

The typical human fingerprint has

Rolled prints and latent prints alike

somewhere between 75 and 175 such

are subject to artifacts and distortions,

ridge characteristics. Level 3 detail

though the problems with latent prints

focuses on microscopic variations in the

are more acute because they are smaller,

ridges themselves, such as the slight

and left more carelessly than full-rolled

meanders of the ridges (the “ridge path”)

prints, and are left on surfaces that many

and the locations of sweat pores. This is

other fingers have also touched.

the level of detail most likely to be

Appellant Br. at 10-11. See Andre

obscured by distortions.

Moenssens et al., Scientific Evidence in

Civil and Criminal Cases, § 8.08 at 514 The FBI—the agency that made the

(4th ed. 1995) (“Many latent impressions primary identification in this case—uses

developed at crime scenes are badly an identification method known as ACE-

blurred or smudged, or consist of V, an acronym for “analysis, comparison,

partially superimposed impressions of evaluation, and verification.” The basic

different fingers.”). steps taken by an examiner under this

protocol are first to winnow the field of

Fingerprints are left by the depositing

candidate matching prints by using Level

of oil upon contact between a surface and

1 detail to classify the latent print. Next,

the friction ridges of fingers. The field

the examiner will analyze the latent print

uses the broader term “friction ridge” to

to identify Level 2 detail (i.e., Galton

designate skin surfaces with ridges

points and their spatial relationship to

evolutionarily adapted to produce

one another), along with any Level 3

increased friction (as compared to

detail that can be gleaned from the print.

smooth skin) for gripping. Thus toeprint

The examiner then compares this to the

or handprint analysis is much the same as

Level 2 and Level 3 detail of a candidate

fingerprint analysis. The structure of

full-rolled print (sometimes taken from a

friction ridges is described in the record

database of fingerprints, sometimes taken

before us at three levels of increasing

from a suspect in custody), and evaluates

detail, designated as Level 1, Level 2 and

whether there is sufficient similarity to

Level 3. Level 1 detail is visible with the

declare a match. In the final step, the

naked eye; it is the familiar pattern of

match is independently verified by

loops, arches, and whorls. Level 2 detail

another examiner, though there is some

involves “ridge characteristics”—the

dispute about how truly independent this

patterns of islands, dots, and forks

verification is.

formed by the ridges as they begin and

end and join and divide. The points The standards used by the FBI at the

6

evaluation stage of the ACE-V protocol has the advantage of allowing an

are somewhat less concrete than the examiner to find a match in situations

numerical descriptions found in where an examiner using a strict point-

television police dramas that extol based standard would not find one, this

“twenty-point matches” and the like. An flexibility comes at the price of

n-point match refers to a match between substituting a degree of subjectivity for

an unknown latent print and a known full an objective numerical standard.

print in which the examiner has

2. The Daubert Hearing

identified n corresponding Galton points

in the correct geometry relative to one The District Court held a five-day

another. A number of jurisdictions both hearing pursuant to Daubert v. Merrell

outside the United States and within Dow Pharmaceuticals, Inc., 509 U.S.

seem to rely on a system where a 579 (1993), to rule on the admissibility

minimum number of corresponding of the government’s and M itchell’s

points must be found before a match may proposed expert testimony. The record

be declared, irrespective of Level 3 of this marathon hearing alone comprises

detail. See, e.g., 2 Paul C. Giannelli & nearly one thousand pages of testimony

Edward Imwinkelried, Scientific and a similarly voluminous array of

Evidence § 16-7(A), at 768 (3d ed. 1999) exhibits. The government called six

(“In France, the required number [of witnesses (plus one rebuttal witness), and

points for a match] used most often is 24 Mitchell, four. The District Court found

while the number is 30 in Argentina and all the offered expert witnesses to be

Brazil.”). Such jurisdictions are said to qualified in their respective fields, and

use a “point system.” On the other hand, neither party raises a challenge to the

Canada does not have a minimum point qualifications, as such, of the witnesses.

threshold for identification, and the Rather, both sides’ issues lie with the

United Kingdom recently eliminated a content of the testimony accepted by the

minimum point threshold. See United District Court. We briefly describe the

States v. Llera Plaza, 188 F. Supp. 2d areas of testimony of each of the

549, 569-70 (E.D. Pa. 2002) (quoting witnesses, starting with the government’s

Lord Lester of Herne Hill’s colloquy witnesses.

with Lord Rooker). The alternative

a. The Government’s Experts

approach—which gained favor with the

FBI in the late 1940s, App. 378a— is to Steven Meagher, an FBI special

use a combination of quantity and agent, testified at the hearing about Level

quality: If ridge characteristics are 1, Level 2, and Level 3 detail (as

abundant, then the quality of Level 3 described above), and other aspects of

detail is unimportant; but a paucity of fingerprint identification. With regard to

Galton points can be compensated for by the FBI’s practices, technology, and

high-quality Level 3 detail. While this operations, he testified about the ACE-V

7

protocol; that the FBI does not rely on a cellular bases for the permanence of

minimum “points” standard for matching friction ridge arrangements. Ed German,

fingerprints (and why it does not); and of the United States Army Criminal

about the Automated Fingerprint Investigation Laboratory, testified to the

Identification System (“AFIS”) computer lack of similarity found between

system (which automates some corresponding fingerprints of identical

preliminary aspects of fingerprint twins, a conclusion established by his

matching). Meagher also described a own research on identical twins and

survey (which we discuss, infra) of state confirmed by other studies of identical

fingerprint identification agencies that he twins.

prepared and circulated for the purpose

The government also offered David

of demonstrating that the fingerprint

Ashbaugh, of the Royal Canadian

match in this case was, by wide

Mounted Police, who testified broadly

consensus, correct. He also described an

about the development, comparison, and

experiment (which we also discuss,

identification of friction ridge skin and

infra) designed and run in cooperation

impressions. Like the other government

with the contractor for the FBI’s AFIS

witnesses who were examined on the

computer system, Lockheed Martin, that

matter (viz., Agent German, Agent

would search a portion of the AFIS

Meagher, and Dr. Budowle) he

database for identical fingerprints.

responded that it was his opinion that

Donald Zeisig, of Lockheed Martin, and

friction ridge arrangements were unique

Bruce Budowle, a statistician and

(the “uniqueness proposition”) and

population geneticist with the FBI, were

permanent (the “permanence

also involved in this experiment, and

proposition”), and that positive

both testified at the Daubert hearing.

identifications can be made from

Zeisig also testified in greater detail

fingerprints containing sufficient

about the technical background of the

quantity and quality of ridge detail. Dr.

AFIS computer system.

Babler also opined that friction ridge

The government offered two arrangements are unique and permanent.

witnesses focusing principally on the These propositions were the foundation

biological aspects of fingerprints. Dr. of the government’s argument that latent

William Babler, of Marquette University, fingerprint identification evidence

testified about the prenatal development satisfies Daubert.

of friction ridges, opining that unique

The government conducted two

arrangements of friction ridges develop

experiments in anticipation of the

in the womb within a matter of months

Daubert hearing: (1) a survey of state

after conception. He also testified to the

fingerprint identification agencies asking

medical community’s accepted

them, inter alia, if they could match the

understanding of the anatomical and

latent prints in this case to Mitchell’s ten-

8

print card; and (2) a search for identical Part B of the survey was designed as

fingerprints using data in the AFIS a demonstration of the ACE-V

computer system.2 The specifics of these identification protocol, and it used the

experiments bear on their relevance as latent fingerprints at issue in this case.

expert evidence, and so we describe them Part B offered each agency photographs

in some detail. of the two latent prints and of Mitchell’s

ten-print card. Agencies were asked first

For purposes of this case, Meagher

to attempt to identify the ten-print card

created a survey packet that was sent out

using their own computerized fingerprint

to the principal law enforcement agency

database. It is common practice (for

of each of the fifty states, plus the

efficiency’s sake) to “filter” the database

District of Columbia, Canada’s Royal

in making an identification, by

Canadian Mounted Police, and the

considering only the subset of records

United Kingdom’s Scotland Yard. The

(by race, sex, date of birth, etc.) that are

survey contained three parts: Part A

likely to result in a match. Meagher

involved questions about whether the

requested that agencies not filter their

agency currently accepts fingerprints as a

database for this test, to ensure that the

means to individualize (i.e., make an

prints were compared against the

identification), and about whether the

maximum possible number of print

agency regards fingerprints as unique

records. Of the forty-seven agencies that

and permanent. All fifty-three recipients

responded, the only match that was found

responded in the affirmative to both

was in Pennsylvania, where Mitchell’s

queries. Joint Supp. App. at 56. Part C

ten-print record was already on file.

inquired whether the agencies had ever

found two individuals to have the same In the second segment of Part B,

fingerprint; the response was, agencies were asked to attempt to match

unanimously, no. Part C also revealed the latent prints to their existing records.

that, in the aggregate, the ten-print The only “hits” were made by the two

records of nearly 70 million agencies (Mississippi and South Dakota)

individuals—or about 700 million that inputted the ten-print card supplied

fingerprints—have been examined by Meagher into their system prior to

during the course of the agencies’ running the search (and thus raised the

operations. likelihood of a match). Pennsylvania

was unable to run this search because of

equipment troubles, but represented that

2

We note that these experiments—and, it undoubtedly would have made a match

indeed, much of the expertise marshaled if its system were fully operative.

both by the government and by

The third segment of Part B asked

Mitchell—required resources and

agencies to perform manual comparisons

preparation that are far from typical in

of the latent prints to the ten-print card

federal criminal trials.

9

provided to them. This survey was than M itchell; and in the third segment,

single-blind, i.e., while Meagher knew no agency matched a latent print to any

that the latent prints had been identified finger other than the one to which the

as Mitchell’s, knew that the ten-print FBI had matched the latent print.

card was Mitchell’s, and believed the

The second experiment conducted by

latents could be matched to the ten-print

the government’s experts was known as

card, none of the survey recipients was

the “50/50” experiment. This was an

told any of this. Roughly two thirds of

empirical examination by computer of a

the agencies responded to this portion.

subset of the FBI’s fingerprint records to

Over three quarters of the responding

search for pairs of very similar

agencies matched both prints consistently

fingerprints taken from different sources.

with the FBI’s identification. Of those

Finding such a pair would undermine the

that did not match both prints, half

uniqueness proposition, see supra page

matched only one print consistent with

8, that the government’s other experts

the FBI’s identification, and half

testified was well-established. The

matched neither print. In followup

experiment data set was a set of fifty

communications, the FBI either

thousand prints (out of about 340 million

convinced these non-identifying agencies

in the FBI’s AFIS computer system).

that a match did exist and they so

Rather than select these fifty thousand

acknowledged (though it took the strong

prints at random, the experimenters

suggestion of annotated blown-up

(Agent Meagher, Mr. Zeisig, and Dr.

photographs of the prints), or otherwise

Budowle) took them from the subset of

established reasons for the non-

prints that were from white males and

identification (e.g., the examiner deemed

exhibited a left-sloped whorl pattern at

the quality of the supplied photographs to

Level 1 detail. The experimenters also

be too poor to make an identification,

ensured that multiple prints from the

and would have preferred an original; or

same person were included in the set of

the comparison was performed by an

fifty thousand. The effect of these

inexperienced examiner, and on review,

restrictions was to bias, from the outset,

a senior examiner was able to find a

the prints toward being more similar (and

match).

hence more likely to contain a matching

A critical summary point is that no pair).3

agency ever registered a “false” positive

(i.e., a positive match that contradicted

3

the FBI’s result): In the first segment of An analogy may illustrate this biasing

Part B, no agency matched Mitchell’s effect: Consider a large multicolored pile

ten-print card to someone else’s ten-print of crayons produced by mixing several

card; in the second segment, no agency boxes of crayons. If one chooses a dozen

matched the latent prints to anyone other “dark” crayons at random, one is more

likely to find among those dozen crayons

10

In the first part of the test, a computer the world other than the person who

program—using the same algorithms as deposited the print at approximately one

the FBI’s AFIS computer system uses to in ten to the eighty-sixth power (i.e., 1

match prints—attempted to match each chance in 1 followed by 86 zeroes), a

of the fifty thousand prints against the very low probability indeed.

full set of fifty thousand prints (hence the

Apparently recognizing that analysis

moniker “50/50”). Thus, a total of

of full-rolled prints was not particularly

50,000 x 50,000, or 2.5 billion,

germane to the question of the

comparisons were performed. For each

identification of latent partial prints, the

print, the best match was, by an

government’s witnesses conducted a

enormous margin, itself.4 Based on

second experiment. From each of the

statistical extrapolation from these

fifty thousand prints, they had the

results, the experimenters put the chances

computer create a simulated latent print

of a single full-rolled print matching

(referred to as a “pseudolatent print” or

another full-rolled print from anyone in

simply a “pseudolatent”), as might be

recovered from a crime scene, by taking

a pair of exactly the same color than one only about a fifth of the full-rolled print.5

is to find such a pair if one selects a They then ran a similar fifty thousand-

dozen crayons at random from the pile at by-fifty thousand comparison to see how

large. strongly the pseudolatent prints matched

full prints from which they had not been

4

We note that the comparisons were derived. With one exception which we

run for each print against all 50,000 identify in the margin, each pseudolatent

prints, not against the other 49,999 was a strong match with the full print

prints. Thus, every print was assured of from which it had been derived, by a

having a tautologically perfect match wide margin over any other full print. 6

(i.e., itself) that could serve as a baseline

for statistical comparisons. This was

done to quantify statistically how much 5

The pseudolatents were 21.7% of the

better the perfect match was than all areal size of the full print, a figure which

other comparisons. The cases in which a Meagher determined was the average

print was a strong match for a print other size of a set of actual latent prints that he

than itself were subsequently discovered had previously used for testing.

to be the product of a double-entry in the

6

database (i.e., a set of prints from the Meagher explained that the sole

same person had been entered into the exception was caused by a poorly created

database twice). The experimenters fingerprint card. On the card in question,

testified that the system’s ability to catch the flat impression had strayed out of the

this unintentional duplication bolstered region on the card designated for the flat

their confidence in its capabilities. impression, and had left part of a print in

11

Statistical computations based on this described which agencies adhered to a

experiment put the probability of a latent point system, how many points they

partial print matching the full print of required to make an identification, and

anyone in the world other than the person noted that the agencies that did not find a

who deposited the print at approximately match generally reported that they had

one in ten to the sixteenth power (i.e., 1 found an insufficient number of points of

in 10,000,000,000,000,000), also a very similarity between the latent print and the

low probability. ten-print card. Ms. Peterman also

reported on the varying levels of

b. Mitchell’s Experts

experience and accreditation of the

Mitchell’s first witness at the examiners who performed the

Daubert hearing was Marilyn Peterman, comparisons for the agencies.

an investigator with the Defender

The first of Mitchell’s three major

Association of Philadelphia who took

experts was Dr. David Stoney, the

statements from those fingerprint

director of the McCrone Research

examiners at state agencies who had

Institute in Chicago, a not-for-profit

failed to match the latent prints to

organization engaged in teaching and

Mitchell’s ten-print card in completing

research in the forensic sciences. Dr.

Part B of the FBI’s survey. 7 She

Stoney was, in Mitchell’s counsel’s

words at the Daubert hearing, offered as

the box designated for one of the rolled an expert “with respect to whether a

impressions. Consequently, one of the fingerprint examiner’s conclusion that a

boxes for a rolled print actually latent fingerprint came from a particular

contained a rolled print, plus a fair-sized individual is a scientific determination.”

piece of a flat print of a different finger. App. 763a. The nucleus of Dr. Stoney’s

As a result, the strong match found by opinion is summarized in a portion of his

computer was actually a match between testimony at the hearing:

the pseudolatent print and the stray The determination that a

portion of the flat print. As with the fingerprint examiner . . . makes

database error discovered in the first when comparing a latent

stage of the 50/50 experiment, the fingerprint with a known

experimenters found this mistaken match fingerprint, specifically the

to be evidence of the robustness of their determination that there is

computer system. sufficient basis for an absolute

7 identification, is not a scientific

It appears that, in the interest of

efficiency, the parties consented to

introducing hearsay from the examiners

who completed the FBI Meagher for the government, and

survey—primarily through Agent through M s. Peterman for Mitchell.

12

determination. . . . It is a that it is wrong.” App. 781a.

subjective determination

Dr. Stoney also criticized the 50/50

without objective standards to

experiment. He noted first the

it.

undisputed proposition that two

Now, by “subjective” I mean impressions of the same friction ridges

that it is one that is dependent on will not be identical—artifacts and

the individual’s expertise, distortions will invariably appear.8 In

training, and the consensus of that experiment, see supra page 10 and

their agreement of other note 4, a fingerprint was compared

individuals in the field. By “not against itself and 49,999 other

scientific” I mean that there is not fingerprints taken from the FBI’s

an objective standard that has database. Hence, Dr. Stoney explained,

been tested; nor is there a the simulated task modeled by the 50/50

subjective process that has been experiment was that of matching Print 1

objectively tested. It is the and (the identical) Print 1 of Finger A.

essential feature of a scientific In his submission, the task in real-world

process that there be something to fingerprint identification is one of

test, that when that something is matching Print 1 and Print 2 of Finger A.

tested, the test is capable of Thus, Stoney reasoned, the 50/50

showing it to be false. experiment as executed assessed how

much better a match is found between

App. 765a. Dr. Stoney opined that the

Print 1 and (the identical) Print 1 of

evaluation phase of the ACE-V protocol

Finger A than between Print 1 of Finger

requires the examiner to make a binary

A and Print 1 of Finger B. A more

determination: Either two prints match

meaningful version of the 50/50

sufficiently to make an absolute

experiment, Dr. Stoney explained, would

identification, or they do not. This Dr.

Stoney contrasted to certain other

forensic disciplines in which 8

This point also underpins Dr.

intermediate determinations are

Stoney’s more general criticism of the

expressed in probabilistic terms. Dr.

discipline of latent fingerprint

Stoney further objected to any

identification: Dr. Stoney agreed that

characterization of fingerprint

human friction ridges are unique and

identification as having a “zero error

permanent, including small areas, App.

rate,” explaining that “something with a

914a, but suggested that this alone is

zero error rate cannot be a science . . . .

unhelpful on the question whether prints

[I]f we start out saying fundamentally

are identifiable, because fingerprints are

something can’t be shown to be wrong,

so subject to distortion and the forensic

then it means that we can’t test it. If we

identification process is so flawed, App.

can’t test it, . . . there’s no way to show

917a-920a.

13

have asked how much better a match is opinion as to whether latent fingerprint

found between Print 1 and Print 2 of examination meets the criteria of

Finger A than between Print 1 of Finger science.” App. 813a-814a. Like Dr.

A and Print 1 of Finger B.9 Stoney, Prof. Starrs testified that it was

his opinion that “[the current practice of]

Dr. Stoney further criticized the

fingerprint comparison and analysis is

method used to create the pseudolatent

not predicated on a sound and adequate

prints in the second part of the

scientific basis for purposes of making

experiment. Dr. Stoney explained that it

an individualization to one person from a

was established in the literature that

fragmentary print to the exclusion of all

simple masking, and even computer-

other persons in the world.” App. 828a.

generated blurring, of full prints cannot

adequately simulate real latent partial To support his conclusion, Prof.

prints. Dr. Stoney’s ultimate conclusion Starrs highlighted five aspects of

was that these experimental defects fingerprint examination that in his

rendered the probabilities derived by the opinion were inconsistent with a

government experts meaningless. scientific discipline: (1) claims to

“absolute certainty”; (2) “the failure to

The defense’s second principal expert

carry out controlled empirical-data-

was James Starrs, a professor in the

searching experimentation”; (3) a failure

Department of Forensic Sciences and the

to engage in error-rate analysis; (4) the

law school at George Washington

lack of uniformity, objectivity,

University. Prof. Starrs has had a long

systematization, and standards; (5) “a

career at the intersection of law and

failure to show a due regard to a

forensic science; indeed, an article by

vigorous and uncompromising

Prof. Starrs was cited by the Supreme

skepticism.” App. 828a-829a. In

Court in Daubert. See Daubert, 509 U.S.

elaborating on each of these points, Prof.

at 591 (citing James E. Starrs, Frye v.

Starrs gave illustrations. For example, he

United States Restructured and

briefly described a case of false

Revitalized: A Proposal to Amend

identification; he described some of the

Federal Evidence Rule 702, 26

subtle and non-systematized aspects of

Jurimetrics J. 249, 258 (1986)). Prof.

analyzing Galton points, see supra page

Starrs was offered as an “exert [sic] in

6, and he criticized some aspects of the

forensic science qualified to provide an

training of new fingerprint examiners.

Prof. Starrs also explained that he viewed

the government’s testimony and

9

We note, however, that such an experiments involving full-rolled prints

experiment was beyond the immediate as irrelevant to the question of latent

capability of the government because its partial print identification. However,

database, by design, does not have under cross-examination Prof. Starrs was

multiple prints from the same finger.

14

agnostic on whether the propositions he group, a notable difference, Dr. Cole

challenged as unproven might, in the explained, between fingerprint

end, be scientifically supportable. identification and, say, psychiatric

diagnosis. Dr. Cole also opined that

Mitchell’s final expert at the

fingerprint identification was not

Daubert hearing was Simon Cole, a post-

scientific because, inter alia, the

doctoral fellow at Rutgers University,

fingerprint identification community had

with expertise in “science and technology

not engaged in studies that attempt to

studies with particular expertise

falsify the discipline’s premises; did not

regarding the fingerprint profession.”

engage in anonymous, critical (as

App. 939a. Dr. Cole had no experience

opposed to positive) peer review; and did

in latent print examination. From his

not recognize error rates.

research, Dr. Cole identified four

explanations for the widespread c. Mitchell’s Exhibits

acceptance of fingerprint identification

As part of the Daubert hearing,

evidence: First, from the earliest days of

Mitchell also introduced several hundred

the discipline, fingerprint examiners have

pages of documentary exhibits,

developed an “occupational norm of

principally journal articles and other

unanimity,” i.e., examiners would not

excerpts from the corpus of literature

publicly disagree with one another about

criticizing the practice and theory of

an identification. Second, in terms of the

latent fingerprint identification, authored

way in which the fingerprint examination

by his experts and by others. Also

community handled the instances of

introduced were the results of some

known misidentification, such cases

fingerprint proficiency tests, which

would, Dr. Cole explained, be blamed on

suggested that examiners were prone to

practitioner incompetence or

both false negatives (i.e., declaring a

misconduct.10 Third was a simple lack of

nonidentification where an identification

judicial scrutiny— a sort of snowball

should have been made) and false

effect of string citations to cases and

positives (i.e., making an incorrect

treatises approving fingerprint

identification). App. 3014a, 3063a.

identification evidence. Fourth was a

Finally, the defense introduced a survey

lack of an organized counter-expert

of jurors that found that 93% agreed with

the statement “fingerprint identification

10 is a science” and 85% agreed with the

Dr. Cole noted that both of these first

statement “fingerprints are the most

two explanations were well illustrated by

reliable means of identifying a person.”

the FBI’s survey: Agent Meagher

App. 3047a-3048a.

followed up with each agency until a

match was agreed to, or otherwise d. The Government’s Rebuttal Witness

identified inexperienced examiners as the

To respond to defense testimony

source of nonidentifications.

15

regarding the “occupational norm of specialized knowledge.

unanimity” among fingerprint examiners,

***

the government offered Pat Wertheim, a

fingerprint examiner, as a rebuttal Further, pursuant to this

witness. Wertheim testified that he and Court’s ruling, this Court finds

David Grieve (who was present but did that the government’s fingerprint

not testify) were involved as defense evidence is highly probative and

experts in a case of false identification in substantially outweighs any

the United Kingdom. Based on their danger of unfair prejudice to

examination of the evidence in that defendant.

case—which was both independent of

***

the U.K. authorities and independent of

each other—they testified, in opposition We find that the government’s

to the prosecution’s expert, that the latent expert witness—at this juncture it

print in that case could not be matched to appears it’s Duane Johnson [sic

the defendant. The purpose of this Wilbur Johnson?], an FBI latent

testimony was to counter Dr. Cole’s fingerprint examiner who testified

contentions about the occupational norm first in the previous trial, and

of unanimity within the discipline. those other latent experts that

testified in the Daubert

3. The District Court’s Daubert and

hearing—are capable of testifying

Judicial Notice Rulings

in these proceedings, and in that

Two months after the Daubert regard, I am not going to limit the

hearing concluded, the District Court defense from calling latent

ruled from the bench on the admissibility fingerprint experts to testify as to

of expert testimony at trial. In relevant the ability not to identify or make

part, the Court stated: an identification from the

fingerprints, and I am also going

The matter presently pending

to allow the defense to call any

before the Court is in reference to

latent fingerprint expert who

the defense motion to exclude the

indicates that fingerprints are not

government’s fingerprint

reliable sources of information.

identification evidence, and based

on the Daubert hearing and also Only for that limited purpose

Kumho, this Court denies the and I am going to exclude

defendant’s motion. And evidence as to whether or not

pursuant thereto, this court is not [latent fingerprint identification

going to make a determination as is] scientific, technical, or

to the particular area of scientific whatever. It has no relevance

knowledge and technical or before the jury here. The question

16

is whether or not an called upon, we will instruct the

identification can be made by jury as so.

examination of

App. 1031a (repunctuated for clarity).

fingerprints—latent

The Court so instructed the jury. On

fingerprints.

appeal, Mitchell asserts that it was error

App. 1029a-1031a (repunctuated for for the District Court to take judicial

clarity). notice of these matters.

As we understand the ruling, the C. Mitchell’s Second Trial

District Court held that the government’s

1. The Government’s Case

expert witnesses and M itchell’s expert

witnesses could testify, but with the The case against Mitchell rested on

caveat that the latter could not testify to eleven lay witnesses and two experts.

the question whether latent fingerprint The government’s star witness was

identification is a “science.” This ruling Bookie’s girlfriend, Kim Chester. Ms.

forms at least the baseline of two of Chester testified that she was present

Mitchell’s issues on appeal: the when Bookie and T were planning the

admission of government experts, and robbery, and that she helped Bookie

the restriction of his own experts. The watch the comings and goings of the

Court again discussed the admissibility armored car in the weeks before the

of the defense’s expert witnesses in a robbery. Ms. Chester said that she and T

colloquy with counsel immediately first met Mitchell and his wife at

before jury voir dire, an exchange that Mitchell’s house, where she heard

we will discuss in greater detail, infra Mitchell and T discussing plans for the

Part IV. robbery. Mitchell’s wife, Anita, invoked

her spousal privilege and did not testify.

Immediately following its ruling on

Eileen Lambert, T’s girlfriend at the

the admissibility of expert testimony, the

time, testified that she also witnessed

District Court addressed what would

meetings between T and Mitchell.

become another ground of Mitchell’s

appeal. Again from the bench, the Court Ms. Chester testified that the night

ruled: before the robbery, Mitchell, Bookie, and

T discussed the need to obtain a stolen

This Court will take judicial

car to use in the robbery. She explained

notice that human friction ridges

that the next morning—September

are unique and permanent

12th—Bookie, T, and Mitchell drove her

throughout the area of the friction

to work. She described how M itchell

ridge skin, including small friction

and Bookie were arguing about what car

ridge areas, and further that

to use in the robbery—the car they were

human friction skin arrangements

in was Mitchell’s wife’s car, and he did

are unique and permanent, and if

17

not want to use it in the robbery. Ms. from the driver’s side door handle—that

Chester testified that they dropped her he later identified as matching Mitchell’s

off at her work, and that when she next ten-print card as the right and left

spoke to Bookie, he indicated that they thumbs, respectively.

had gone through with the planned

Mitchell was arrested the afternoon of

robbery. At that time, he had a

September 12th. Special Agent Kevin

substantial amount of cash, some of

Mimm and Special Agent Daniel

which he used to purchase a car and

Murphy, both of the FBI, testified to the

redeem several pieces of jewelry from a

circumstances of the arrest. They

pawn shop.

explained how they had been conducting

Alma Shaw testified about her car surveillance operations in Philadelphia as

being stolen the morning of September a result of a number of armored car

12th. Emanuel Glover and Vernon robberies; Agent Murphy was in charge

Muse, the armored car guards, and Kim of these operations. Agent Mimm

Kover-Jacobs, the check cashing agency testified that while he was engaged in

manager, testified about the robbery covert surveillance of Mitchell and

itself. Messrs. Glover and M use both tailing Mitchell’s car, Mitchell began to

identified Ms. Shaw’s car as the getaway flee; Mimm described how he chased

car; also, a fragment of the getaway car’s Mitchell at high speed for several blocks,

license plate was noted by a bystander, and was ultimately able to stop him.11

Regan Wiggins, and this fragment was Mitchell was arrested, and $1400 in five

consistent with Ms. Shaw’s car’s license and ten dollar bills was recovered from

plate. him. This currency was never identified,

however, as having been part of the

Laura Barnett, a Philadelphia police

armored car delivery.

officer, testified that she recovered Ms.

Shaw’s car shortly after the robbery. It Agent Meagher returned to testify at

was found (with a bullet hole through the trial about many of the matters brought

trunk) a few blocks from the check out by the government at the Daubert

cashing agency. FBI Special Agent hearing. He discussed the embryology of

Donald Halfpenny testified that Ms. friction ridge skin, the fingerprints of

Shaw’s car had been secured by the identical twins, and the biological basis

Philadelphia police at the time he took

control of it. Wilbur Johnson, an FBI

fingerprint examiner whom the Court 11

The anonymous note that was the

qualified as an expert, testified that in

subject of the previous appeal in this case

Ms. Shaw’s car he found, photographed,

was the critical link: That note connected

and preserved two latent

the robbery getaway car to Mitchell’s

fingerprints—one from the gearshift

own car, allowing the FBI to monitor and

knob on the steering column, and one

capture Mitchell so quickly.

18

for the permanence of fingerprints. He to identify one or both of the latent prints

described how latent prints are left and as belonging to M itchell. 12

how they are processed by examiners,

Mitchell also cross-examined the

and the various conclusions that

government’s experts, Agents Johnson

examiners can draw from a comparison

and Meagher. Cross-examination of

of prints. During M eagher’s testimony,

Johnson concentrated on questions about

the government invoked the Court’s

his presentation to the jury of the

promise to take judicial notice of the

fingerprints he matched— Johnson’s

uniqueness of small areas of friction

demonstrative exhibits identified only

ridge skin. The government also read a

nine points of Level 2 similarity between

stipulation detailing some of the results

the latent prints from the car and

of the survey that Meagher testified

Mitchell’s ten-print card, despite

about at the Daubert hearing, and the

Johnson’s and Meagher’s claims of a

prosecutor examined Meagher regarding

greater number of similarities. Through

the agencies that did not make a positive

cross-examining Agent Johnson,

identification of the latent prints.

Mitchell also probed the existence and

Meagher then demonstrated to the jury in

maintenance of minimum-point standards

some detail his use of the ACE-V

and other quality-control measures at the

technique in matching the latent prints to

FBI in particular, and in the discipline

Mitchell’s ten-print card. He stated

more generally. Cross-examination of

definitively that the fingerprints from the

Agent Meagher ranged into more general

beige car matched Mitchell’s ten-print

considerations, most notably the limited

card. Agent Johnson also stated

studies performed specifically to

definitively that he had matched the

establish an error rate for fingerprint

latent prints from the beige car to

identification, and the limited means for

Mitchell’s ten-print card, though he did

detecting errors in particular

not give an in-depth demonstration to the

examinations. Meagher was also cross-

jury as Agent Meagher did.

2. Mitchell’s Case and

12

Cross-Examination of These witnesses (and their states)

the Government’s Experts were: John Otis (Maine); Janice

Williams and Michael McSparrin

The entirety of Mitchell’s case was

(Mississippi); Ralph Turbyfill

the testimony of individuals at state

(Arkansas); Donald Lock (Missouri);

agencies who examined or supervised the

Russell McNatt, Jr. (Delaware);

examination of the latent prints sent by

Raymond York (Idaho); John Artz

Agent M eagher in the survey.

(Nevada); Janice Reeves (Louisiana);

Specifically, Mitchell called thirteen

and Richard Higgins, Edward Pelton,

latent fingerprint experts from nine

Robert McAuley, and James Ruszas

states, all of whom were initially unable

(New York).

19

examined on his highly suggestive of the solicitation . . . admitted . . . [to]

follow-up communications to those state serious shortcomings in fingerprinting as

agencies that did not match Mitchell’s it has been done up to this time.” App.

prints in the survey. 2325a.

D. Withholding of the NIJ Moreover, Mitchell suggested that

Solicitation and Mitchell’s even the government regarded the

Post-Trial Motion solicitation as material. His most

damaging evidence came from Dr.

On February 7, 2000, the jury

Richard Rau of the NIJ, who coordinated

returned a verdict of guilty on all counts.

the drafting of the solicitation. Rau

Mitchell’s May 15, 2000 motion for a

testified to conversations at a September

new trial pursuant to Fed. R. Crim. P. 33

1999 meeting among himself, Donald

was founded on the discovery of a

Kerr (the Assistant Director of the FBI in

research proposal solicitation released by

charge of the FBI crime laboratory),

the National Institute of Justice (an arm

David Boyd (the Deputy Director of the

of the United States Department of

NIJ), and others. Rau claimed that at that

Justice) entitled Forensic Friction Ridge

meeting Kerr and Boyd agreed to

(Fingerprint) Examination Validation

withhold release of the solicitation until

Studies (the “solicitation”). The

the end of Mitchell’s trial. In response to

solicitation sought proposals for research

Dr. Rau’s testimony, the government

studies on “validation of the basis for

called Kerr, Boyd, and the other

friction ridge individualization and

individuals at the meeting to testify that

standardization of comparison criteria.”

Dr. Rau’s account of the delay in

App. 3078a. Creation of the solicitation

releasing the solicitation was incorrect

had been underway before Mitchell’s

and that the delay was caused by

trial, but the solicitation was not released

budgetary issues.

until March 2000—after Mitchell’s trial

had concluded. The District Court denied M itchell’s

motion, reasoning that the solicitation

The District Court held a four-day

was not material for two independently

hearing to take testimony and receive

sufficient reasons: First, the solicitation

exhibits on the creation and import of the

would not have been admissible at trial

solicitation. At that hearing, Mitchell

because attacks on the reliability of latent

established that Agent Meagher (as well

fingerprint identification were not

as some of the government’s other

permitted at trial based on the Court’s

witnesses at the Daubert hearing) had

Daubert ruling; and second, the

been involved in drafting the solicitation.

solicitation was “not meant to set forth

Prof. Starrs testified that he regarded the

the state of the current research” and so

solicitation as “a bolt out of the blue”

its “claimed impeachment value . . .

that suggested to him “that the sponsors

either during the trial or for Daubert

20

purposes is questionable at best.” App. lay witness, Ms. Chester. We will

12a-13a. On appeal, the government address each of these contentions in turn.

disclaims the first ground, but defends

the District Court’s ruling on the second

ground, as well as on alternative grounds III. Admissibility of the

not reached by the District Court. Government’s Expert Testimony

E. This Appeal A. Standard of Review

The District Court had jurisdiction The parties disagree about the

over this case under 18 U.S.C. § 3231. standard of review we should apply in

Mitchell filed a timely appeal from the evaluating the District Court’s decision

final judgment of conviction and to admit the government’s expert

sentence, and we have jurisdiction under testimony. It is well-settled that, as a

28 U.S.C. § 1291. general matter, we review a district

court’s decision to admit expert

On appeal, Mitchell asserts that the

testimony for abuse of discretion. See In

District Court committed five errors.

re TMI Litig., 193 F.3d 613, 666 (3d Cir.

First, he challenges the District Court’s

1999). We exercise plenary review,

ruling following the Daubert hearing that

however, over a district court’s legal

admitted the prosecution’s expert

interpretation of Fed. R. Evid. 702, under

testimony on fingerprint identification.

which the evidence in question was

Second, Mitchell claims that the District

admitted. See id. On this much the

Court erred in precluding his experts

parties agree.

from testifying at trial that fingerprint

identification is not a science, and is Disagreement arises about the

otherwise unreliable. Third, Mitchell standard of review where, as here, the

finds error in the District Court’s District Court made no findings of fact to

decision to take judicial notice of the support its admission of the testimony;

uniqueness of small areas of friction indeed, after the lengthy Daubert

ridge skin. Fourth, Mitchell contends hearing, the District Court elected not to

that the government’s withholding of the make findings of fact or conclusions of

NIJ solicitation, which could have been law (written or oral), and simply ruled

used as impeachment evidence, merited a from the bench. This absence of factual

new trial under Fed. R. Crim. P. 33, or findings, Mitchell contends, requires

that this nondisclosure violated the plenary review. We reject the rule that

government’s obligation under Brady v. Mitchell urges for four reasons. First,

Maryland, 373 U.S. 83 (1963). Fifth, Mitchell has provided no precedent for

Mitchell asserts that the District Court such a heightened standard of review

improperly admitted hearsay in the over a field historically committed to the

testimony of the government’s principal

21

sound discretion of district courts.13 Appeals to assess the matter. See

Second, the exception that Mitchell Ruggero J. Aldisert, The Judicial

proposes would swallow the rule that Process 728-29 (2d ed. 1996) (quoting

district courts’ evidentiary rulings are Maurice Rosenberg, Judicial Discretion

generally reviewed only for abuse of of the Trial Court, Viewed from Above,

discretion. The vast majority of 22 Syracuse L. Rev. 635, 663 (1971)

evidentiary rulings are made on-the-fly (“[P]robably the most pointed and

and without written findings of fact, yet helpful [reason] for bestowing discretion

this Court routinely affords deference to on the trial judge is [that] . . . . he sees

such judgments. Third, Mitchell’s more and senses more [than the Court of

argument misconceives the rationale for Appeals].”)). This case is a good

using a deferential standard of review. example: The District Court assessed

Deferential review is employed not extensive live testimony, while we work

because the court being reviewed labored from a cold record. Fourth, the Supreme

to produce a long opinion— there are Court has in other contexts rejected

lengthy but incorrect opinions just as heightened appellate review of district

there are brief but sagacious ones. court rulings on expert testimony. See

Rather, deferential review is used when Gen. Elec. Co. v. Joiner, 522 U.S. 136

the matter under review was decided by (1997).

someone who is thought to have a better

Thus we reject Mitchell’s proposed

vantage point than we on the Court of

standard of review, and adhere to the

usual precepts of abuse-of-discretion

review over the District Court’s decision

13

The case Mitchell cites in his brief to admit the government’s expert

and relied on at oral argument, United testimony.

States v. Ellis, 121 F.3d 908, 927 (4th

B. Standard for Admissibility

Cir. 1997), is inapposite. Ellis applied

under Rule 702

plenary review not to the admission of

expert testimony, but rather to a claim of The pathmarking Supreme Court

prosecutorial misconduct where the cases interpreting Fed. R. Evid. 702 are

district court had made no findings of Daubert v. Merrell Dow

fact. Apart from the fact that the issue in Pharmaceuticals, Inc., 509 U.S. 579

Ellis has strong Constitutional overtones (1993), and Kumho Tire Co. v.

that the Rule 702 issue in this case lacks, Carmichael, 526 U.S. 137 (1999). The

this Court does not agree with the Fourth version of Rule 702 in effect at the time

Circuit on this point. See United States of the Daubert hearing and the trial

v. Ismaili, 828 F.2d 153, 163 (3d Cir.

1987) (reviewing District Court’s

rejection of a prosecutorial misconduct

claim for abuse of discretion).

22

provided:14 Daubert, 509 U.S. at 589-92. Daubert

was “limited to the scientific context

If scientific, technical, or other

because that [wa]s the nature of the

specialized knowledge will assist

expertise offered [t]here,” id. at 590 n.8,

the trier of fact to understand the

but Kumho Tire extended Daubert’s

evidence or to determine a fact in

“general principles” to all of “the expert

issue, a witness qualified as an

matters described in Rule 702.” Kumho

expert by knowledge, skill,

Tire, 526 U.S. at 149. Thus “technical

experience, training, or education,

knowledge,” under which heading the

may testify thereto in the form of

discipline of latent fingerprint

an opinion or otherwise.

examination and identification seems to

Daubert identified the twin concerns fall, is generally subject to the same

of “reliability” (also described as “good considerations as “scientific” expertise.

grounds”) and “helpfulness” (also

The “general principles” adverted to

described as “fit” or “relevance”) as the

in Kumho Tire comprised not only the

“requirements embodied in Rule 702.” 15

fundamental concerns of reliability and

helpfulness, but also a method for

14

The rule was subsequently amended, assessing reliability. The Daubert Court

effective December 1, 2000, to codify articulated “general observations” to this

aspects of Daubert and its progeny. The end by offering a nonexclusive list of

Advisory Committee’s note five factors that a district court might

accompanying that amendment is a consider in deciding whether to admit

useful consolidation of commentary and evidence under Rule 702. The Advisory

precedent on the version of Rule 702 that Committee summarized these factors:

applies in Mitchell’s case, and so we will The specific factors explicated by

refer to it at points in our opinion. the Daubert Court are (1) whether

15 the expert’s technique or theory

In applying the teachings of Daubert

can be or has been tested—that is,

in In re TMI Litigation, we explained that

whether the expert’s theory can be

Rule 702 was addressed to two issues:

challenged in some objective

first, the qualification of the experts

sense, or whether it is instead

themselves, and second, the reliability

simply a subjective, conclusory

and helpfulness of their testimony. See

approach that cannot reasonably

In re TMI Litig., 193 F.3d at 664 (citing

be assessed for reliability; (2)

In re Paoli R.R. Yard PCB Litig., 35 F.3d

whether the technique or theory

717, 749-50 (3d Cir. 1994) (Paoli II)).

Daubert addresses the latter. As noted

above, the former is not at issue in this

appeal, as the District Court qualified all areas of expertise, and neither party

experts on both sides in their proffered challenges any of these rulings.

23

has been subject to peer Paoli II, 35 F.3d at 742 n.8.

review and publication; (3) the

These factors address only reliability,

known or potential rate of

and not “helpfulness” or “fit.” But the fit

error of the technique or

inquiry in the case of fingerprint

theory when applied; (4) the

identification is not a significant factor,

existence and maintenance of

because identity evidence is the

standards and controls; and (5)

archetypal relevant evidence in criminal

whether the technique or

cases. Thus, the analysis that follows

theory has been generally

only addresses the reliability prong of

accepted in the scientific

Daubert.

community.

C. Application of Daubert Factors

Fed. R. Evid. 702 advisory committee’s

to Government’s Expert Testimony

note.

1. Testability

Citing Kumho Tire, the Advisory

Committee noted that “[o]ther factors We first consider whether the

may also be relevant,” id., and indeed, premises on which fingerprint

courts have augmented this list. In Paoli identification relies are testable— or,

II we drew on Daubert and our earlier better yet, actually tested. “Testability”

decision in United States v. Downing, has also been described as

753 F.2d 1224 (3d Cir. 1985), to lay out “falsifiability.” See, e.g., Daubert, 509

an expanded list of factors: U.S. at 593 (citing Karl R. Popper,

Conjectures and Refutations: The

(1) whether a method consists of a

Growth of Scientific Knowledge 37 (5th

testable hypothesis; (2) whether

ed. 1989)). A proposition is “falsifiable”

the method has been subject to

if it is “capable of being proved false;

peer review; (3) the known or

defeasible.” Webster’s Third New

potential rate of error; (4) the

International Dictionary 820

existence and maintenance of

(unabridged ed. 1966). Proving a

standards controlling the

statement false typically requires

technique’s operation; (5) whether

demonstrating a counterexample

the method is generally accepted;

empirically—for instance, the hypothesis

(6) the relationship of the

“all crows are black” is falsifiable

technique to methods which have

(because an albino crow could be found

been established to be reliable; (7)

tomorrow), but a clairvoyant’s statement

the qualifications of the expert

that he receives messages from dead

witness testifying based on the

relatives is not (because there is no way

methodology; and (8) the

for the departed to deny this).

non-judicial uses to which the

method has been put. In this case, the relevant premises

24

were posed as explicit questions to many on identical twins (testified about by

of the government experts: (1) Are Agent German) showed unique

human friction ridge arrangements fingerprints. While this is a small

unique and permanent? and (2) Can a sample, there are independent and solid

positive identification be made from genetic grounds for believing that if

fingerprints containing sufficient identical friction ridge arrangements are

quantity and quality of detail? The to be found, they are most likely to be

government’s experts responded in the found in identical twins. Third, in the

affirmative. We must consider not course of routine fingerprint

whether we agree as a factual matter with examination, there are certainly

their responses, see Paoli II, 35 F.3d at opportunities to encounter identical

744, but rather whether these hypotheses fingerprints; as several witnesses

are testable (or tested). We conclude that testified, such a discovery would be very

they are. notable and word would spread quickly

throughout the fingerprint examiner

Consider the first premise (which is

community. Yet no reports of non-

really two hypotheses in one)—that

unique friction ridge arrangements were

human friction ridge arrangements are

introduced, and, indeed, the FBI survey

unique and permanent. The uniqueness

sent to state agencies revealed that none

proposition is testable because it would

had ever encountered two different

immediately be shown false upon the

persons with the same fingerprint. Joint

production of identical friction ridge

Supp. App. at 55.

arrangements taken from different

fingers (either from different fingers on The permanence component of the

the same person, or from two different first hypothesis is also easily

people). The uniqueness proposition has testable—simply take fingerprints from

also been tested in several ways: First, an individual at one time and compare

the full-print matching portion of the them to the prints taken at another time.

FBI’s 50/50 experiment tested it and The Daubert hearing did not provide

found no true matches. 16 Second, studies much evidence of actual testing of this

16

The experiment had its limitations, circumstances, we are unsure if it is

though. First, the test sought to match adequate here. There is limited evidence

fingerprints, not friction skin on the record of why the government’s

arrangements on actual fingers. Second, experts chose a 50 thousand fingerprint

it was only a sample—50 thousand set, and why they could confidently

fingers tested, out of about 60 billion in extrapolate from it. Indeed, there is

the world. While this sample size seems some suggestion that purely practical

quite large, and doubtless would be technical concerns may have dominated

adequate in many if not most this choice. See infra note 18.

25

hypothesis, however. Just how much testing has been done

to this end is unclear from the testimony

We turn next to the testability of the

at the Daubert hearing. On the one hand,

second hypothesis—that positive

it might be that examiners compare a

identification can be made from

latent print to a series of full-rolled prints

fingerprints containing sufficient

until a match is found, and then terminate

quantity and quality of detail. Much of

the process. If this protocol is used for

the debate in this case is masked by the

routine examinations, those examinations

word “sufficient.” For example, a

will not tend to turn up multiple matches,

sufficiency standard of “100 points of

because the examiner stops work after

matching Level 2 detail in an undistorted

finding one match. In essence, the

fingerprint lifted from a clean, smooth

surface” would surely attract less

objection than a sufficiency standard of

“four points of matching Level 2 detail Meagher identified fourteen points of

and passable quality.” The actual Level 2 detail (and unspecified

standard employed by any given FBI supporting Level 3 detail, which we

examiner falls somewhere between these leave aside for simplicity) that matched

extremes, yet the FBI’s reliance on an Mitchell’s right thumbprint to the latent

unspecified, subjective, sliding-scale mix print taken from the gearshift knob.

of “quantity and quality of detail” makes Thus, for purposes of this particular

meaningful testing elusive, for it is identification, “sufficient quantity and

difficult to design an experiment to test a quality of detail” really means “fourteen

hypothesis with unspecified parameters. points of Level 2 detail.” The hypothesis

Two things rescue fingerprint that “fourteen points of Level 2 detail is

identification from this apparent failure enough to make an identification” is

of testability: First, the examiner can falsifiable because one might be able to

testify to how much detail (quantitative show that some latent print matches more

and qualitative) was necessary for the than one full-rolled print under the

particular identification at issue; and “fourteen points of Level 2 detail”

second, any testing directed toward standard.

falsifying the premise that a greater or Actual testing (as opposed to mere

equal amount of detail is sufficient to testability) is harder to come by, probably

make an identification will serve as an because someone seeking to falsify this

attempt (albeit an imperfect one) to hypothesis has no a priori reason to

falsify the adequacy of the identification choose 14 points instead of 13 or 15 as

standard actually used.17 the standard. Nonetheless, any showing

that a more stringent standard (e.g., a 20-

point standard) is fallible necessarily

17

A concrete example may provide implies that the 14-point standard is also

some clarity. In this case, Agent fallible.

26

examiner has assumed the computations extrapolating this to a

conclusion—that no other prints will much larger population of prints

match the latent, and therefore no further suggested that such duplicate matches

search is required. On the other hand, would still be highly improbable.

testimony at the Daubert hearing about

Mitchell’s experts, however, attacked

the AFIS computer system suggests that

the design of the 50/50 experiment, most

the system tests a given latent print

effectively on the ground that

against its entire database (or a selected

pseudolatents are poor approximations of

subset) of full-rolled prints, and returns a

real latent prints.18 This lack of

set of the best candidate matches. This

correspondence undermines the utility of

protocol would tend to expose multiple

the experiment because the issue for

full-rolled prints that match a given

Daubert purposes is the testing of the

latent. Consequently, a lack of multiple

hypothesis that positive identification be

matches from AFIS searches can

made from actual latent fingerprints

constitute testing of the hypothesis that

containing sufficient detail. As we

single positive identifications can be

recount above, see supra page 13,

made from latent fingerprints. Whatever

Mitchell’s experts (particularly Dr.

the case, no state agency claimed in

Stoney) convincingly explained why the

response to the FBI survey that it had

found a latent fingerprint that was

“identified with two different fingers of 18

They also contended that actual tests

the same person or even different

on a larger data set (i.e., more

persons.” Joint Supp. App. at 55. This is

fingerprints) would have been preferable

perhaps the strongest support for the

to statistical extrapolations. However,

government on this point.

significantly larger data sets may be

Modest support also comes from the computationally intractable: The

second part of the government’s 50/50 experiments conducted for this case took

experiment, which matched simulated on the order of a day to run on the

latent prints (pseudolatents) against the computer. But for larger sets of

50,000 full-rolled prints in the sample fingerprints, the number of comparisons

under examination. Setting aside goes up as the second power (i.e., the

spurious results due to mistakes in the square) of the number of prints in the

FBI’s database, the experiment found sample. Thus, a 1 million / 1 million

that each pseudolatent strongly matched experiment would take 20 x 20 = 400

one and only one full-rolled print. In times longer than a 50 thousand / 50

other words, the experiment did not thousand experiment—or on the order of

reveal any counterexample to the a year to complete, given the same

hypothesis that identifications can be computing power. An experiment with

made. Moreover, statistical the FBI’s full AFIS database would take

millennia.

27

process used by the government experts reliability. In sum, the hypotheses that

to generate the pseudolatents for the undergird the discipline of fingerprint

50/50 experiment renders them poor identification are testable, if only to a

substitutes for actual latent prints. In lesser extent actually tested by

brief, the failing flagged by Dr. Stoney is experience, and so we find this factor to

that actual prints are subject to weigh in favor of admitting the evidence.

distortions and artifacts that were not

2. Peer Review

simulated by the pseudolatent generator.

Arguably, the pseudolatents resembled The evidence at the Daubert hearing

actual latents only in that the former were on peer review was not particularly

similar in areal size to the latter. Dr. extensive. Much of the testimony

Stoney’s contention rings true: Distorted, centered around the question whether the

real-world latent prints should tend to be “verification” step in the ACE-V

harder to match to full-rolled prints than protocol—where a second examiner

should computer-generated simulated confirms the identification made by the

latents. Since the 50/50 experiment did first examiner—constitutes effective peer

not adequately model real-world review. On the one hand, this could be

conditions, we cannot say that it viewed as stringent peer review,

significantly supports the government’s equivalent to the best sort used in, for

position. example, the physical sciences, where

peer review most often consists of

In sum, if directed, specific actual

anonymously reviewing a given

testing were the requirement of Daubert,

experimenter’s methods, data, and

we might be hesitant to find this factor

conclusions on paper. Sometimes the

weighing in favor of the government.

review takes the form of reproducing in

There is some force to Budowle’s point

full the results under review—that is, a

that “[n]o one would say any one test or

second investigator repeats the entire

any kind of thing [that] has been done in

course of experiments. Thus the

one hundred years proves uniqueness.”

verification step of ACE-V seems

App. 1013a. But his further point about

usually to be akin to this heightened form

a long history of implicit testing is

of peer review: The government’s

equally forceful: “It’s the culmination of

experts testified that verification often

all of the experiences that [demonstrate

amounts to repeating the whole

uniqueness].” App. 1013a. Moreover,

identification process de novo, though

testability—which assures the opponent

sometimes the verifying examiner will

of proffered evidence the possibility of

merely confirm the match found by the

meaningful cross-examination (should he

initial examiner. See App. 161a.

or someone else undertake the

Moreover, in this particular case, the

testing)—is one of the factors announced

survey of state law enforcement agencies

by the Daubert Court as an indicium of

constitutes verification many times over

28

of the match of Mitchell’s fingerprints. by the government and by Mitchell—that

address more theoretical/foundational

Mitchell’s experts, however, (Dr.

questions, such as an appropriate

Cole in particular) cast some doubt on

minimum point standard, the likelihood

the purity of the verification step.

of two persons having identical friction

Backed by his research, Dr. Cole

ridge arrangements, and so on. Thus the

suggested that fingerprint examiners

publication facet of peer review is not a

have developed an “occupational norm

strong factor, and neither reinforces nor

of unanimity” that strongly discourages

detracts from our conclusion that the peer

the verifying examiner from challenging

review factor favors admission.

the identification made by the initial

examiner. Moreover, Dr. Cole criticized 3. Error Rate

peer review of latent fingerprint

The parties have waged a

identification conclusions for not being

considerable battle of experts over

anonymous. We also acknowledge that

whether a known error rate exists for

the cultural mystique attached to

latent fingerprint identification.

fingerprint identification may infect the

Assuming that such a rate has been

peer review process. But the

soundly established, it is surely a low rate

government’s experts countered that they

of error. But the existence of any error

were aware of cases where the results of

rate at all seems strongly disputed by

the verification step caused the initial

some latent fingerprint examiners.

examiner to withdraw his initial

identification. Looking at the entire The question whether an error rate

picture, the ACE-V verification step may can be established on the existing data is

not be peer review in its best form, but, subtler than the parties seem to

on balance, the peer review factor does acknowledge. Preliminarily, we must

favor admission. distinguish between two error rates: false

positives and false negatives. In this

The peer review factor also

context, false positives are incorrect

encompasses publication, as the

affirmative identifications, and false

dissemination of a work tends to subject

negatives are incorrect findings of

it to scrutiny in the same way that

dissimilarity. A fair amount of the

prepublication peer review does. See

government’s evidence—and also much

Daubert, 509 U.S. at 593-94. On the one

of Mitchell’s response—centers on the

hand, a significant fraction of the

existence vel non of failed

publications in the field concern articles

identifications. For example, the

on technique—for example, the best

government stresses the large number of

practices for preserving latent

state agencies that confirmed its

prints— and such materials say little

identifications, and Mitchell counters by

about the field’s reliability. On the other

pointing to the agencies that failed to

hand, there are articles—introduced both

29

identify the prints. But these Thus we must focus on evidence that

observations go to the rate of false is probative of the rate of false positives.

negatives: While a system of Perhaps the government’s most powerful

identification with a high false negative evidence is the fact that, in the course of

rate may be unsatisfactory as a matter of the FBI survey of state agencies, no

law enforcement policy, in the courtroom jurisdiction ever matched the latent prints

the rate of false negatives is immaterial from the gearshift knob and door handle

to the Daubert admissibility of latent to anyone other than M itchell

fingerprint identification offered to prove himself—despite searches run against (in

positive identification because it is not the aggregate) nearly 70 million ten-print

probative of the reliability of the records. Assuming that every record had

testimony for the purpose for which it is 10 fingerprints, and that the latents

offered (i.e., for its ability to effect a actually were left by Mitchell, the test of

positive identification). 19 the two latent prints against these records

implies something on the order of 1.4

billion comparisons resulting in no false

19 positives. The government can also draw

Moreover, evidence of the false

support from the very limited number of

negative rate is often equivocal. While it

reports of false positive identifications

might suggest a generally error-prone

throughout the many decades that the

method, it is equally consistent with a

technique has been in use. Furthermore,

very conservative method with a low

the government’s 50/50 experiment using

false positive error rate. That is, a

pseudolatents, representing 2.5 billion

method may be designed to lower its

comparisons, also did not register any

false positive error rate by accepting a

false positives, though as we have noted,

large number of false negatives out of an

see supra page 27, it had flaws.

abundance of caution. One very familiar

example of such a system is the criminal Mitchell counters this evidence in

jury using the “beyond a reasonable two different ways, but neither of them

doubt” standard: As the adage (attributed fully refutes the government’s evidence.

to Blackstone) says, “It is better that ten First, he raises a legal challenge,

guilty escape [false negatives] than one claiming that the burden of proof under

innocent suffer [a false positive].” The Fed. R. Evid. 104(a) is up-ended by

same may be true for latent fingerprint effectively requiring him to come

identification—the examiners who forward with examples of false positives.

declared they could not match the latent While Mitchell is correct that Rule

prints in the FBI’s survey (the examiners 104(a) places the burden of proof on the

responsible for the putative false proponent of the evidence (here, the

negatives) may have done so because

they would rather commit a likely false

negative error rather than risk a small chance of a false positive identification.

30

government), see Bourjaily v. United We therefore accept that the error rate

States, 483 U.S. 171, 175 (1987), this has been sufficiently identified to count

does not mean that the burden is static, at this factor as strongly favoring admission

least in terms of a burden of going of the evidence. The error rate has not

forward. Particularly in a case like this, been precisely quantified, but the various

where what is sought to be proved is methods of estimating the error rate all

essentially a negative (i.e., the absence of suggest that it is very low. This follows

false positives), it seems quite from three pieces of evidence we identify

appropriate to us to use a burden-shifting above as favoring the government: (1)

framework. Such a framework was

applied here: The government’s

experts—qualified as knowledgeable in

makes clear that error rates and the

matters pertaining to fingerprint

qualification of the expert are distinct

identification—testified to their being

inquiries. 35 F.3d at 742. The corollary

unaware of significant false positive

to this, however, raises an issue for any

identifications. At that point, it becomes

given fingerprint expert: His testimony

quite reasonable to shift the burden to the

would be more likely to be admitted

opponent of the evidence (here, Mitchell)

(because he would be more qualified) if

to counter this claim with affirmative

he himself demonstrated a low rate of

examples.

false positives in his own work and/or on

Mitchell’s second attack on the his own proficiency tests. Cf. Calhoun v.

government’s evidence of error rates is Yamaha Motor Corp., 350 F.3d 316, 322

factual. He presented evidence that (3d Cir. 2003) (holding that the scope of

fingerprint examiners sometimes make an expert’s testimony was properly

false positive identifications on circumscribed by the scope of his

proficiency examinations. This evidence expertise).

is troubling, but we view it as evidence As suggested above, known false

relating only to the competency of those positives have been attributed to malice

practitioners, leaving undisturbed the or incompetence on the part of the

government’s evidence about the near- examiner, and not to a deeper flaw in the

absence of false positive identifications.20 method itself. Dr. Cole testified that this

“circling the wagons” behavior is yet

another occupational norm of a

20

Mitchell’s experts respond by fingerprint identification community bent

denying the existence of a dichotomy on preserving the unimpeachability of its

between method error rate and methods. But even if every false positive

practitioner error rate, asserting that both identification signified a problem with

are part of a unitary inquiry. We reject the identification method itself (i.e.,

this view as a legal conclusion independent of the examiner), the overall

inconsistent with Paoli II. Paoli II error rate still appears to be microscopic.

31

the absence of significant numbers of Rules of Evidence, admission of expert

false positives in practice (despite the testimony was governed by the Frye test,

enormous incentive to discover them), which required that the evidence must

(2) the absence of false positives in the have gained “general acceptance in the

FBI’s state agency survey, and (3) the particular field in which it belongs.”

statistical computations based on the Frye v. United States, 293 F. 1013, 1014

50/50 experiment. (D.C. Cir. 1923). Daubert held that

Congress’s adoption of Rule 702

4. Maintenance of Standards

legislatively overruled Frye, see 509 U.S.

Closely related to the question of at 588-89, but at the same time

error rate is the maintenance of standards acknowledged that “‘general acceptance’

to guide the application of the method. can yet have a bearing on the inquiry,”

This is lacking here in some measure. id. at 594. Thus we consider as one

The FBI maintains that its flexibility to factor in the Daubert analysis whether

consider a mixture of Level 2 and Level fingerprint identification is generally

3 detail in making identifications renders accepted within the forensic

its method superior to and more flexible identification community. The answer is

than the minimum-points standards used yes, as demonstrated by the results of the

in some states and various foreign FBI’s survey of state agencies. See App.

jurisdictions. The tradeoff, though, is 383a. Mitchell’s only argument with

that the FBI’s method lacks a significant respect to this factor is that there is no

yardstick of standard-based objectivity. scientific community that generally

In contrast, with a minimum-point accepts fingerprint identification. But

standard there is at least some agreement the scientific/nonscientific distinction is

about what constitutes a Galton point and irrelevant after Kumho Tire, and

what does not. accordingly we reject the argument. We

also note that the Court of Appeals for

Some standards do remain: There are

the Fourth Circuit, in addressing the

procedural standards (such as ACE-V)

same question that we are considering

and terminological standards (such as the

here, relied heavily on general

naming conventions for Galton points).

acceptance to support the admission of

But these are insubstantial in comparison

fingerprint identification evidence. See

to the elaborate and exhaustively refined

United States v. Crisp, 324 F.3d 261 (4th

standards found in many scientific and

Cir. 2003). We likewise conclude that

technical disciplines. As such, we find

this factor weighs in favor of admitting

that this factor does not favor admitting

the evidence.

the evidence.

6. Relationship to Established

5. General Acceptance

Reliable Techniques

Prior to the adoption of the Federal

Although the parties have not

32

provided us with extensive analysis of Schneider ex rel. Estate of Schneider v.

the relationship of the principles and Fried, 320 F.3d 396, 407 (3d Cir. 2003)

practice of latent fingerprint (“[The defendant’s] argument appears to

identification to “‘more established challenge the qualification of [the

modes of . . . analysis,’” Paoli II, 35 F.3d plaintiff’s expert]; although we note that

at 742 (quoting Downing, 753 F.2d at ‘the degree to which the expert testifying

1238-39), it seems to us that this is the is qualified’ also implicates the reliability

best heading under which to consider the of the testimony.” (quoting Paoli II, 35

government’s evidence from the fields of F.3d at 742)).

developmental embryology and anatomy.

The qualifications of Agents Meagher

The testimony and documentary

and Johnson matter the most, because

materials introduced on these topics

they were the government’s experts at

during the Daubert hearing—especially

trial. Both had estimable qualifications.

through Dr. Babler—tended to establish

The putative blemish on their

biological bases for the uniqueness and

qualifications, which we hint at above,

permanence of areas of friction ridge

see supra note 20, is that neither testified

skin. Since no question was raised about

extensively about his own known error

the soundness and reliability of the work

rate as a practitioner (as might be

in these specialties, we are comfortable

revealed, for example, by proficiency

that the reliability of these fields is well-

tests they had taken). While this is by no

established. Independent work in these

means fatal to the admissibility of the

fields bolsters the underlying premises of

testimony, prosecutors would be well-

fingerprint identification, and so we find

advised to elicit testimony about their

that this factor lends additional support

experts’ personal proficiency, rather than

to admitting the latent fingerprint

relying on the discipline’s good general

identification evidence.

reputation among lay jurors. Failing that,

7. Degree to Which the Expert we are confident that defense counsel

Testifying Is Qualified will use cross-examination to expose

incompetent fingerprint examiners. In

As we have noted before, there were

this case, Agent Meagher’s uniquely

essentially no challenges to the

strong qualifications and the

qualifications of the government’s

confirmatory identifications from state

experts (or of Mitchell’s experts, for that

agencies are a surrogate for testimony

matter), but the binary question whether

about Agent Meagher’s and Agent

an expert is or is not qualified to testify

Johnson’s personal proficiency as

to a particular subject is analytically

examiners.21 Thus this factor supports

distinct, under Rule 702, from the more

finely textured question whether a given

expert’s qualifications enhance the 21

Mitchell’s counsel came close to

reliability of his testimony. See

inquiring on voir dire about Agent

33

admitting the government’s evidence. government offered some evidence of the

non-judicial uses of fingerprint

8. Non-Judicial Uses

identification, particularly through Dr.

We have recognized that evidence of Budowle. App. 639a-641a. In analyzing

the non-judicial uses of the technique in this factor, the government relies on

question is relevant to the Daubert three categories of non-judicial uses of

reliability inquiry. See Paoli II, 35 F.3d fingerprints: (1) the identification of

at 742. This is because non-judicial use arrested persons (e.g., checking an

of a technique can imply that third arrestee’s record at the time of booking);

parties—i.e., persons other than the (2) biometric identification as a security

proponent of the expert testimony, for measure (e.g., authenticated access to a

whom the testimony is typically self- computer system) or for regulatory

serving—would vouch for the reliability purposes (e.g., fingerprinting for driver

of the expert’s methods.22 The licensing as an anticounterfeiting

measure); and (3) identification of partial

remains following disasters. While at

Meagher’s results on proficiency first blush this seems like a factor

examinations administered internally by strongly supporting admissibility, the

the FBI, but did not actually ask a bloom recedes upon close analysis.

specific question. App. 1456a-1457a.

The government did ask Agent Johnson Latent fingerprint identification

about his results on FBI proficiency works from fingerprints that are partial

examinations, but defense counsel and subject to distortions. All the non-

objected and the Court sustained the judicial uses listed above either use full-

objection on the ground that Johnson had

already been qualified as an expert. App.

1652a-1653a. As our discussion in the introduced evidence of the widespread

text suggests, this question was commercial use of biometric

proper—even desirable—and the District identification technology based on

Court was wrong to sustain the objection. fingerprints. It is possible that

commercial adoption of the method

22

Keeping this rationale in mind is signals acceptance of its reliability. But,

helpful, because some non-judicial uses as Mitchell’s uncontradicted survey

will support the required inference of evidence showed, fingerprint

third-party confidence better than others. identification enjoys a near-mythical

For example, no one would argue that the reputation for reliability, and so the

commercial popularity of astrology for evidence of commercial adoption is

non-judicial use makes it fit for equally consistent with uncritical

admission under Rule 702. This case acceptance of a method that consumers

may provide another example: As we merely believe—but do not know—to be

discuss below, the government reliable.

34

rolled prints, or avoid the difficulties identification, the government’s experts

introduced by distortion— or both. Both did testify that fragments of friction ridge

differences are critical, as Mitchell’s skin have been used to make

experts testified and as the government’s identifications, but even those

experts acknowledged: It is significantly identifications still differ from latent

easier to match one clean full-rolled print fingerprint identification because

to another than it is to match a somewhat identification using actual skin eliminates

distorted latent fragment to a full-rolled the challenges introduced by

print. 23 Thus, in the case of identification distortions.25 Thus there is less here than

of arrestees, the booking officer will take meets the eye, and while this factor

a ten-print card with a full set of full- supports admitting the government’s

rolled prints, and if the prints do not evidence, it does so only weakly.

come out cleanly, the officer has the

D. Application to the Record of

opportunity to take a second set of

Core Daubert Principles

impressions. Likewise, the security and

regulatory uses of fingerprinting Although it is clear from the

generally rely on clean, full-rolled foregoing analysis of the Daubert factors

prints.24 As for disaster-victim

App. 639a. This makes such a technique

23

The government’s experts implicitly more akin to latent fingerprint

acknowledged this—even before the identification, but it still differs in

Daubert hearing—in the very design of significant ways. First, the fraction of

the 50/50 experiment: The first stage of the print will be distortion-free, unlike

that experiment was the matching of full- actual latent prints. Second, the 6%

rolled prints to full-rolled prints, but the portion is likely to be taken from a

ultimate aim of the experiment was to portion of the finger with a high areal

test pseudolatent prints against full-rolled density of Level 2 detail, a luxury that

prints to better simulate the more latent fingerprint examiners do not have.

demanding exercise of latent fingerprint

25

identification. Of course, as we have We also understand the task in

noted above, see supra page 14, even this disaster-victim identification as being

refined experiment used pseudolatents, (merely) to individualize one victim out

and thus failed to capture the of at most a few thousand victims, while

complexities of matching latent prints forensic criminal identification seeks to

marred by distortions and artifacts. individualize the defendant out of a pool

of millions of potential perpetrators.

24

Dr. Budowle testified that current Accordingly, there seems to be less of a

commercial research and development threat of a false positive in the context of

seeks to use as little as 6% of the area of disaster-victim identification than in

the full print to make an identification. forensic criminal identification.

35

that the government’s fingerprint scientifically sound and

evidence passes muster, Mitchell methodologically reliable fashion.

contends that the government’s inability

Ruiz-Troche v. Pepsi Cola Bottling Co.,

to establish that its evidence is correct,

161 F.3d 77, 85 (1st Cir. 1998) (citations

and its failure to show that its evidence

omitted) (quoting Daubert, 509 U.S. at

meets the standards required of

590) (citing Kannankeril v. Terminix

“science,” mean that the government’s

Int’l, Inc., 128 F.3d 802, 806 (3d Cir.

evidence must be excluded. M itchell is

1997); Paoli II, 35 F.3d at 744), quoted

wrong. This is established by Daubert

in part in In re TMI Litigation, 193 F.3d

itself, which requires no more than that

at 692. Good grounds for admission

the Court satisfy itself that “good

plainly exist here.

grounds” exist for the expert’s opinion.

See 509 U.S. at 590. To the extent that Mitchell’s attack

rests on his experts’ claim that latent

Judge Selya has put it well:

fingerprint examiners do not engage in

Daubert does not require that a “science,” he does not heed the text of

party who proffers expert Rule 702 or the Supreme Court’s

testimony carry the burden of teachings in Kumho Tire. Rule 702

proving to the judge that the “makes no relevant distinction between

expert’s assessment of the ‘scientific’ knowledge and ‘technical’ or

situation is correct. As long as an ‘other specialized’ knowledge.” Kumho

expert’s scientific testimony rests Tire, 526 U.S. at 147. The very holding

upon “good grounds, based on of Kumho Tire is that those categories

what is known,” it should be simply address what type of testimony is

tested by the adversary covered by the rule, and that, in

process— competing expert addressing admissibility under Rule 702,

testimony and active cross- the same factors generally apply to all

examination—rather than categories of expert testimony. Kumho

excluded from jurors’ scrutiny for Tire explicitly rejected as unworkable

fear that they will not grasp its and unnecessary any “distinction

complexities or satisfactorily between ‘scientific’ knowledge and

weigh its inadequacies. In short, ‘technical’ or ‘other specialized’

Daubert neither requires nor knowledge.” Id at 148. That a particular

empowers trial courts to discipline is or is not “scientific” tells a

determine which of several court little about whether conclusions

competing scientific theories has from that discipline are admissible under

the best provenance. It demands Rule 702; at best, there will be some

only that the proponent of the overlap between the factors that bear on

evidence show that the expert’s a field’s status as “science” and

conclusion has been arrived at in a Daubert’s factors addressed to reliability.

36

Reliability remains the polestar. referred to as a “gatekeeper.” This

metaphor is particularly apt because it

Mitchell seeks a significantly higher

works two ways: On the one hand, the

threshold of admissibility under Rule

court must exclude some evidence as a

702, and, consequently, a very different

gatekeeper, by “preventing opinion

allocation of responsibility between

testimony that does not meet the

judge and jury. Yet Rule 702 and

requirements of qualification, reliability

Daubert put their faith in an adversary

and fit from reaching the jury,”

system designed to expose flawed

Schneider, 320 F.3d at 404. But on the

expertise. Mitchell misconceives this

other hand, the court is only a

balance struck by the framers of Rule

gatekeeper, and a gatekeeper alone does

702 and the Daubert Court. As the

not protect the castle; as we have

Advisory Committee explained in the

explained, “[a] party confronted with an

context of the December 1, 2000

adverse expert witness who has

amendment to Rule 702, “Daubert did

sufficient, though perhaps not

not work a ‘seachange over federal

overwhelming, facts and assumptions as

evidence law,’ and ‘the trial court’s role

the basis for his opinion can highlight

as gatekeeper is not intended to serve as

those weaknesses through effective

a replacement for the adversary system.’”

cross-examination.” Stecyk v. Bell

Fed. R. Evid. 702 advisory committee’s

Helicopter Textron, Inc., 295 F.3d 408,

note (quoting United States v. 14.38

414 (3d Cir. 2002).

Acres of Land Situated in Leflore

County, Miss., 80 F.3d 1074, 1078 (5th Indeed, as our discussion of the

Cir. 1996)). Daubert itself emphasized various Daubert factors suggests, many

the point: “Vigorous cross-examination, of them are guarantees that cross-

presentation of contrary evidence, and examination and adversary testing will be

careful instruction on the burden of proof possible: Testability ensures the basic

are the traditional and appropriate means possibility of meaningful cross-

of attacking shaky but admissible examination. Peer review and

evidence.” 509 U.S. at 596. These trial publication also provide raw material for

practices and procedural devices like the the cross-examining attorney to confront

directed verdict, “rather than wholesale the expert with. The existence of a

exclusion under an uncompromising . . . known error rate may force an expert to

test, are the appropriate safeguards where admit to the limitations of his or her

the basis of scientific testimony meets methods. The maintenance of standards

the standards of Rule 702.” Id. We provides an objective benchmark to

echoed this in Paoli II, where we noted confirm that the expert did indeed follow

“Rule 702 mandates a policy of liberal her method. And so on. Since these

admissibility.” 35 F.3d at 741. factors were well-satisfied in this case, it

was with confidence that the baton was

In this context, the court is often

37

passed from the Court to the adversary Third, this case does not announce a

system. categorical rule that latent fingerprint

identification evidence is admissible in

The principle that cross-examination

this Circuit, though we trust that the

and counter-experts play a central role in

foregoing discussion provides strong

the Rule 702 regime has three important

guidance. And as we explain in

applications to this case. First is the core

Velasquez, both Rule 702 and the Sixth

holding of United States v. Velasquez, 64

Amendment’s Confrontation Clause

F.3d 844, 848-49 (3d Cir. 1995): Experts

permit any criminal defendant to put the

with diametrically opposed opinions may

prosecution to its proof at trial. None of

nonetheless both have good grounds for

this, however, should be read to require

their views, and a district court may not

extensive Daubert hearings in every case

make winners and losers through its

involving latent fingerprint evidence.

choice of which side’s experts to admit,

The Supreme Court has emphasized that

when all experts are qualified. Rather,

district courts “have the same kind of

the same standards of reliability and

latitude in deciding how to test an

helpfulness should be applied to both

expert’s reliability” as they do in

sides, with a “‘preference for admitting

deciding “whether or not that expert’s

any evidence having some potential for

relevant testimony is reliable.” Kumho

assisting the trier of fact.’” Id. at 849

Tire, 526 U.S. at 152. Thus a district

(quoting DeLuca v. Merrell Dow Pharm.,

court would not abuse its discretion by

Inc., 911 F.2d 941, 956 (3d Cir. 1990)).

limiting, in a proper case, the scope of a

We return to this in the next section,

Daubert hearing to novel challenges to

where we discuss the District Court’s

the admissibility of latent fingerprint

handling of Mitchell’s experts.

identification evidence—or even

Second, district courts will generally dispensing with the hearing altogether if

act within their discretion in excluding no novel challenge was raised.

testimony of recalcitrant expert

E. Conclusion on the Admissibility

witnesses—those who will not discuss on

of the Government’s Evidence

cross-examination things like error rates

or the relative subjectivity or objectivity We conclude, on the record before us

of their methods. Testimony at the read in light of the basic Daubert

Daubert hearing indicated that some principles, that most factors support (or

latent fingerprint examiners insist that at least do not disfavor) admitting the

there is no error rate associated with their government’s latent fingerprint

activities or that the examination process identification evidence. There are good

is irreducibly subjective. This would be grounds for its admission. We therefore

out-of-place under Rule 702. But we do conclude that the District Court did not

not detect this sort of stonewalling on the abuse its discretion in holding the

record before us. government’s evidence admissible.

38

IV. Admissibility of Mitchell’s does not disclose that Mitchell’s experts

Expert Testimony were excluded or the scope of their

testimony improperly limited. To the

A. Introduction

extent that the record is even ambiguous,

Mitchell asserts that he was not the onus was on Mitchell’s counsel to

permitted to put on all of his experts at make a clear record, especially given the

trial, and hence was not able to multiple, nuanced categories of

effectively counter or undermine the testimony being discussed in the

government’s fingerprint identification colloquies with the District Court on this

evidence. Specifically, Mitchell matter.

contends that his three principal experts

As in the previous section, we review

at the Daubert hearing— Dr. Stoney,

the District Court’s decision to admit or

Prof. Starrs, and Dr. Cole— were, as a

exclude expert testimony for abuse of

practical matter, excluded from the trial

discretion, see In re TMI Litig., 193 F.3d

by the District Court’s rulings limiting

at 666, but also note that an error of

the scope of their testimony. Mitchell

law—such as a failure to follow

argues that our holding in United States

Velasquez—is an abuse of discretion, see

v. Velasquez, 64 F.3d 844 (3d Cir. 1995),

Planned Parenthood v. Attorney Gen.,

requires that he be able to present

297 F.3d 253, 265 (3d Cir. 2002). We

qualified expert testimony before the jury

begin with a discussion of Velasquez and

to challenge the government’s expert

then turn to the District Court’s rulings.

testimony. The government does not

dispute this as a legal matter; instead it B. Velasquez

takes issue with Mitchell’s premise,

The defendant in Velasquez was tried

arguing that the District Court did not in

on federal drug, firearms, and conspiracy

fact exclude Mitchell’s witnesses. The

charges. A fact in issue at trial was the

foregoing discussion about the central

origin of certain packages with

role of adversary testing in expert

handwritten mailing labels, packages the

testimony has direct application.

government sought to connect to

If Mitchell were correct that his Vela

This text is long and has been trimmed here. Open the source document for the complete record.