Illinois v. Thomas Jennings

On September 19, 1910 Thomas Jennings was charged and convicted for murder based on fingerprint evidence.  It was the first time fingerprints had been presented in a criminal trial.  Jennings appealed to the Supreme Court of Illinois challenging the admissibility of fingerprint evidence as a means for positive identification. The Court considered the following information [70]:

“Standard authorities on scientific subjects discuss the use of fingerprints as a system of identification, concluding that experience has shown it to be reliable”.

“These authorities state that this system of identification is of very ancient origin, having been used in Egypt when the impression of the monarch's thumb was used as his sign manual, that it has been used in the courts of India for many years and more recently in the courts of several European countries; that in recent years its use has become very general by the police departments of the large cities of this country and Europe; that the great success of the system in England, where it has been used since 1891 in thousands of cases without error, caused the sending of an investigating commission from the United States, on whose favorable report a bureau was established by the United States government in the war and other departments”.

Based on this information, the Supreme Court ruled “there is a scientific basis for the system of fingerprint identification, and that the courts are justified in admitting this class of evidence; that this method of identification is in such general and common use that the courts cannot refuse to take judicial cognizance of it”.  The court affirmed the lower court decision for the murder conviction and Jennings was subsequently executed. 

The Court Fails to Consider the Propensity for Look-alikes

In Illinois v. Jennings the Supreme Court legitimized fingerprints as a means to establish  identification without considering the fact that fingerprint evidence, especially latent print evidence recovered from crime scenes, is often partial or fragmented and the fact that partial, fragmented fingerprints in two impressions from different individuals can on occasion be similar in appearance.  The court failed to ask about the propensity for small pieces and parts of fingerprints to duplicate in different individuals.  They failed to consider the propensity for look-alikes.

David Ashbaugh elaborates on "sufficiency to individualize" as an amount of corresponding ridge formations present in two impressions that exceeds the largest ever seen in a non-match [19].  The idea for an individualization to exceed the best look-alike is documented minimally, i.e. by Interpol [39], John "Dusty" Clark [43], Christophe Champod [102], and although it is presented in examiner training [19] and is consistent with documented standards for sufficiency to individualize by tool mark examiners [38], it has not been formally documented as criteria for individualization by the IAI, SWGFAST or FBI.   The absence of formal documentation has resulted in courtroom testimony that fails to elaborate on this idea and how much is enough to infer identification which has subsequently prevented fingerprint evidence to be allowed in court.  The February 2008 court ruling in Maryland v. Bryan Rose illustrates what can happen when the expert fingerprint witness fails to elaborate about how much is enough to establish sufficiency to individualize [55] [56].

Maryland v. Bryan Rose       

Bryan Rose was arrested and charged with murder based on fingerprint evidence.  The fingerprint evidence was subsequently challenged.  The issue presented to the State of Maryland Circuit Court for Baltimore County was whether ACE-V is a methodology, which establishes the reliability of the general practice of latent fingerprint identification.

Material about the Madrid was available for the court to review.  Subsequently the court recognized that the cause for the FBI’s misidentification of the Madrid fingerprint showed that “the unusual similarity between the latent print and Mayfield’s known print was a major factor in the misidentification that confused three experienced FBI examiners and a court appointed expert.”  The court recognized that look-alikes occur in fingerprints.  The court also recognized that the biological science concerning fingerprints is different than the practice of examiners to individualize two small, partial, fragmented amounts of apparent corresponding ridge formations with 100% certainty as from the same source.  The court recognized that “individualization is a determination that two friction ridge impressions, for example, the latent print and the exemplar(s) originated from the same source, to the exclusion of all others.  The court recognized that individualization is the result when the compared impressions contain “sufficient quality (clarity) and quantity of friction ridge detail in agreement.” 

However, the state’s expert fingerprint witness (Stephen Meagher) did not testify to any minimum threshold and on the contrary stated that positive identification could be effected using any amount of corresponding ridge detail.  This statement is easily disproved if considered in the extreme.  For example, hypothetically, based on a world fingerprint population of 6.6 billion, can positive identification be made with a small amount of matching ridge formations containing the excellent agreement of only, let's say, 2 strong diminishing area ending ridge units with 10 continuous ridge units with no distinct edge contours, no pores, and no other ridge features of any kind?   The answer is no.  This amount is not enough for purposes of individualization in this case because look-alikes displaying more matching ridge formations than these have been recorded, i.e. Chesapeake and Clark non-matches.  As a result positive identifications cannot be based on any amount of corresponding ridge formations which means the statement was incorrect.  

Examiners should be prepared to testify that amounts of corresponding ridge formations should [in general] exceed the largest and best look-alike ever recorded.  They should be prepared to define a minimum threshold.  In Maryland v. Bryan Rose, the Court ruled that the expert witness failed to elaborate on “sufficiency” which contributed to the fingerprint evidence not being allowed.  The court also recognized the issue as not being that of finding two fingerprints that are alike, but rather the finding of prints from two different fingers that can be mistakenly judged to be alike by a fingerprint examination. 

In reconsideration the Court declined to change its prior ruling and stated “The State’s sole witness who was not a scientist and did not have scientific training did not prove that the ACE-V method is generally accepted by the scientific community at this time”, and “The evidence in this case demonstrated that the ACE-V method is neither perfect nor perfectly applied to partial latent print identifications although the State’s sole witness testified to a 0% error rate.”, and “In this case, the Defendant demonstrated there are no studies of the ACE-V method to determine the reliability of the methodology.” [56]

Subjective Training and Experience v. Objective Statistics to Weigh Fingerprints

It is significant to emphasize a fundamental issue facing the fingerprint community and the need to be able to elaborate on sufficiency to infer identification, if necessary in court.  Examiners need to be aware of experiments that demonstrate the inaccuracy of using only the term "training and experience" to describe how ridge formations are weighted and how they compare with the largest and best look-alikes ever recorded (see Validation Study).    

It has been established that partial, fragmented fingerprint look-alikes have been recorded.  As a result the largest and best look-alikes represent amounts that every individualization (if defined as to the exclusion of everyone else in the world) should exceed.  The fingerprint weighing scale currently used to determine the relative weight of ridge formations and how they compare to the largest and best look-alikes ever recorded is “training and experience”.  With regards to how ridge formation values or weights are determined, Pat Wertheim states:  “In a truly scientific comparison, however, one accepts the idea that not all features in a fingerprint are exactly equal in the weight they contribute to the identification.  Some features contribute more to the conclusion, others less.  Some features may weigh as a grain of sand when tested on the scale, others may weigh as a cobblestone.  In Ridgeology, it is up to the expert doing the comparison to determine the relative weight of each feature.  The expert makes the determination based on training and experience.” [71]

A highly significant survey was performed by Christophe Champod, Cedric Neumann, and others, on groups of examiners to investigate their perceptions of the weight or value for certain types of ridge formations.  The results showed considerable disagreement and lack of consensus with regards to the value or weight given to the various ridge formation types.  Fingerprint examiners were given images of ridge formations displaying pores, ridge edges, scars, warts, creases, ridge widths, incipient ridges, and a dot.  The results from the survey showed “No agreement has been demonstrated with respect to the value to be accorded to the characteristics in question.  The answers from the participants were evenly distributed between the possible answers for most of the characteristics.” [72] 

Based on these surveys, the vague and unspecific measure of examiner "training and experience" is not an accurate, reliable weighing scale to use to define the relative weight for ridge formations.  As a result, a more accurate, reliable weighing scale is needed. 

It is significant to note here that there have been numerous studies showing that simple statistical methods are more accurate than humans and that the more complicated the problem the less likely an “expert” will beat a statistical algorithm.  One reason why humans perform poorer than statistical models is because they fail to assign the right weights to the overall “equation”, but they think they do [53] [74][75][76].

Fingerprint examiner “experts” currently rely almost exclusively on a non-statistical professional judgment or personal opinion approach to define relative weights for fingerprint ridge formation types and to make decisions regarding positive identification or exclusion.  The Champod/Neumann study illustrates how fingerprint examiners consistently fail to agree on relative weights given to specific fingerprint ridge formation types.  The results from this study shows that the attribution of fingerprint ridge formation weights based on expert opinion cannot be accurate or reliable.  If examiners who rely only on expert opinion cannot reliably make decisions about the weight accorded to different amounts of ridge formation types, then the decisions they make regarding what is the threshold for sufficiency to establish positive identification and at what point it is exceeded must also be unreliable.

This is a significant issue for the field of fingerprint identification especially in light of the Maryland v. Bryan Rose case in which fingerprint evidence was not admitted based on the State’s witness failure to elaborate on “sufficiency”.  However, this issue is easily remedied by assigning quantitative weights to specific ridge formation types based on frequency of occurrence and application of the product rule to calculate aggregate values and compare these values against the largest and best amounts ever recorded in a non-match.  Total values that exceed the total value for the largest and best look-like ever recorded [even for a given fingerprint population] may be considered valid, scientific basis to establish sufficiency to individualize.

Based on the work of Meehl, Ayres, Trout and Bishop [76], and others [74] it has been demonstrated that in general statistical models are more accurate than non-statistical models.  Subsequently, there is overwhelming evidence that a fingerprint weighing scale based on statistics should be superior to fingerprint examiner expert opinion in order define relative weights for individual and aggregate fingerprint ridge formations. 

Fingerprint examiners are encouraged to take this information into consideration, do not merely accept it as true or valid, but think it through, perform your own research and experiments in order to verify whether or not it is correct. 

NEXT PAGE >>> 

The T Model is a statistical weighing scale that has demonstrated to be more accurate and more reliable than human experts to identify the largest and best look-alikes as insufficient to infer identification with (so far) a zero error rate. 

The T-Model uses an objective statistical approach to fingerprints in order to define fingerprint match probability and "T-Values" for  amounts of matching ridge features, and number of close matches or look-alikes likely to occur in the relevant population for the case at hand.

Examiners need to be prepared to elaborate how they define relative individual and aggregate weights for ridge formations. Examiners should be familiar with the T-Model, and perform their own frequency studies, in order to better understand the relative weights that may be attributed to various fingerprint ridge feature types in position.