January 2006 - Volume 8 - Number 1

Research and Technology

Review of the Scientific Basis for Friction Ridge Comparisons as a Means of Identification: Committee Findings and Recommendations

Bruce Budowle
Chief Scientist
FBI Laboratory
Quantico, Virginia

JoAnn Buscaglia
Research Chemist
Counterterrorism and Forensic Science Research Unit
FBI Laboratory
Quantico, Virginia

Rebecca Schwartz Perlman
Research Chemist
Counterterrorism and Forensic Science Research Unit
FBI Laboratory
Quantico, Virginia

Introduction | Basic Assumptions | Quality | Image Capture and Quality | Generating Data Sets | Simultaneous Impressions | Exclusions | IAFIS Searches | Sourcebook | Conclusions | Acknowledgments | Suggested Readings

I. Introduction

In response to the misidentification of a latent print, senior management of the FBI Laboratory tasked a three-member review committee to evaluate the fundamental basis for the science of friction ridge skin impression pattern analysis and to recommend research to be considered to test, where necessary, the hypotheses that form the bases of this discipline. The committee’s evaluation followed a proactive approach, identifying areas where research and development might provide enhancements to current analytical capabilities in the field of friction ridge pattern analysis. This committee was not able to find a single peer-reviewed publication that definitively addressed all of the basic assumptions of friction ridge impression analysis, nor was that expected. Science is built on many studies, and one needs to review the totality of data. Thus this task was greater than the time and resources provided. For practical reasons, the committee was able to review only a small portion of the literature to define the current practices, scientific bases, and philosophies of the discipline. Further assessments were gained by interviewing experts in fingerprint analyses, forensics, statistics, and legal matters, as well as by relying on the experiences of the committee members to understand the fundamentals and to derive recommendations for documentary and validation studies. The findings and recommendations that follow are therefore not exhaustive, but instead focus on the primary foundations of the science of friction ridge skin impression pattern analysis.

II. Basic Assumptions

The committee reviewed the scientific basis for comparing a latent print found at a crime scene with a reference print obtained by a more controlled process (inking method, live scan, etc.) and the ability to render an interpretation of whether or not the two originate from the same source. There is indisputable evidence supporting that such practices can be carried out reliably and that the general process should not be rejected.

All forensic analyses have a subjective component, in which the analyst decides whether or not to interpret the evidence and the thresholds to institute during the evaluation. The latent print ACE-V (Analysis, Comparison, Evaluation-Verification) process has a greater component of subjectivity than, for example, chemical analyses or DNA typing. Yet this does not in itself call into question the reliability of the latent print analysis methodology. However, at some level, the examiner might be considered a “black box.” The examiner makes an interpretation, and one may not know, understand, or appreciate the machinations that the examiner made to arrive at a conclusion. One also may not be able to codify the data used to make that interpretation. But reliable results have been obtained, and thus there can be confidence in the process. Alternatively, some suggest that more objective criteria would be useful to set minimum criteria across the field, provide greater confidence in the process, and provide better evaluation criteria to review cases critically. Both of these positions (i.e., the black box and objective criteria) have merit and should be considered to address the scientific underpinnings of friction ridge skin impression pattern analyses.

The use of friction ridge skin comparisons as a means of identification is based on the assumptions that the pattern of friction ridge skin is both unique and permanent. The assumption of uniqueness is grounded in the belief that the stresses, strains, and tensions that occur during ridge formation are infinite, random, and independent and that these forces yield tremendous variation in the population of fingerprint ridge formations produced. However, it is well accepted that wide variations in the amount of detail transferred during any given contact from the three-dimensional world of a finger to the two-dimensional realm of a fingerprint may not permit individualization. Thus, although the ridge pattern arrangement on friction ridge skin is unique, one may not be able to render an identification or an exclusion of a source from the limited amount of detail in certain latent prints. The second assumption, that friction ridge skin detail is permanent, is supported by basic biology (i.e., the structure of friction ridge skin) and by empirical observation. The patterns on friction ridge skin do not change over time, except that they become larger during growth to adulthood or may change as a result of a serious injury (which may produce scarring, for example) or some disfiguring disease. These two assumptions, uniqueness and permanence, are based to a lesser or greater degree on empirical research, probabilistic models, anecdotal evidence, and extrapolation.

III. Quality

It is compelling to focus on a quantifiable threshold; however, quality/clarity, i.e., distortion and degradation of prints, is the fundamental issue that needs to be addressed. Variability is inherent in the production of any two prints from the same source, due to a number of factors (surface, environmental factors, size, etc.). Latent prints in particular are not produced in a controlled manner and are subjected to various development processes that may add to the variation between the latent print and the source fingerprint. One has to accept a certain amount of explainable variation in the representation of a print; otherwise, everything would be excluded and no effective print comparisons could be made.

The human eye is quite good at correcting for distortion and degradation, much better than current computer systems. Although the human expert may be better at identifying and accounting for distortion, this process is somewhat subjective and dependent on the individual examiner. Some distortion also is tolerated for minutiae searching by an automated fingerprint identification system (AFIS). A certain amount of distortion is tolerated in the relationship of features between a reference print and a latent print. In fact, a “wider net” is cast when searching and identifying candidates in order to reduce the number of false exclusions in the candidate list. Automated matching algorithms account well for a degree of distortion, and these algorithms can be subjected to rigorous testing. Perhaps some validation studies have been carried out on the degree of distortion (at Level II detail) that can be tolerated before a true mate falls from the top position of the candidate list. If such studies have not been carried out, they could be considered. The concern is that automated encoding algorithms generate false-positive and false-negative minutiae. Thus the detected minutiae pattern may have to be artificially generated or require intensive manual encoding.

To determine if a print can be used for comparison, an assessment of the clarity and quantity of information is made (i.e., the Analysis step in the ACE-V method). This is a sliding heuristic practice: as quality declines, a greater quantity of features is needed for the print to be considered “of value” for identification. There is/are no defined quality metric(s). Quality metrics are difficult concepts to define and convey. Perhaps some quality metrics could be (1) a demonstrable and recognizable feature, (2) general clarity/blurriness, (3) grayscale requirement, and (4) defined ridge/valley. Perhaps guidelines for the Comparison and Evaluation phases should be developed for distinguishing the number of explainable dissimilarities versus unexplainable features.

A set of guidelines describing quality metric features should be established. There is some attempt by the FBI LPUs to address quality by invoking a minimum 12-point guideline for requiring a supervisor’s approval for a rendered identification because quality may be low. This 12-point system should be tested to determine if a correlation exists between the number of points and clarity. If so, the number of appropriate points to invoke additional review could be codified.

It may be difficult to prescribe quality metrics for every case, but some guidelines could be developed. If not, a minimum quantity threshold (if possible) with a requirement of recognizable and identifiable features might suffice. It is important to stress that under the minimum quantity threshold approach, tabulation of features occurs at two stages. The first is during the analysis of a latent print, and the second is a refinement after comparison with a reference print. At first glance, the refinement may be considered a biased practice to be avoided. The committee disagrees, provided that the analysis stage is carried out independently and, when possible, prior to that of the reference print.

An analogy with DNA is provided for clarification. When assessing a DNA evidence profile as being a single-source or mixed sample, the number of alleles per marker is counted. A single-source sample should display only one or two alleles per marker. Suppose only one of the 13 CODIS (Combined DNA Index System) markers in an evidence profile displays three alleles (call it marker TPOX) and all other markers display one or two alleles (excluding identical twins, a 13-marker profile from a single-source sample is typically considered sufficient for source identification). Because the phenomenon of three alleles has been observed (although not frequently), the examiner may not rule out the hypothesis that the evidence is from a single source. At the same time, the possibility of a mixed sample, although less likely, might be entertained. A suspect is identified and his or her reference DNA profile matches at all 13 CODIS markers and displays three alleles at marker TPOX. Certainly, such results would be consistent with the interpretation that the sample is not a mixture and would provide stronger weight toward source attribution. There is no distinction for latent print examinations, and more documentation of meaningful scenarios would be useful.

IV. Image Capture and Quality

Early on, the committee assumed that some studies have assessed the accuracy of representation of the friction ridge detail on the finger using the image-capture systems that record reference prints (rolled inked prints, live scan, flat inked prints, etc.). However, there apparently are little or no published data addressing this assumption. Some image-capture systems, such as live scans, may not accurately capture the features and their arrangements on a finger. If such low-quality images are being accepted into the fingerprint repository, it could hamper the identification process, particularly for partial prints. Most discussions of the issue of the ability to effect an identification from a partial latent print have considered only the clarity and quantity of features in the latent print. However, it is important to consider image quality, accuracy of recorded detail, and information content in the known source or file prints as well, in particular when performing an AFIS search. This gap needs to be rectified.

V. Generating Data Sets

In order to carry out effectively some of the studies suggested herein, populations of latent and reference print data need to be available. Attempts have been made to collect such data, and images can be purchased. But the committee could not find any well-defined protocol(s) describing the process for recording, collating, evaluating, and editing such research materials. For example, for a minimum quantity threshold study, it is suggested that high-quality data be used first. To extract the feature data, it would be desirable to hold distortion and degradation to a minimum. It would then be possible to degrade the data in a controlled manner to explore the effects of poor quality on the threshold value.

VI. Simultaneous Impressions

There is a practice of using simultaneous impressions to make an identification. Simply stated, simultaneous impressions are two or more friction ridge impressions from the fingers and/or palm of one hand that are determined to have been deposited at the same time. Considerable variation in the definition of simultaneous prints as well as the practices for interpreting such evidence was found within the FBI LPUs. This makes it difficult to effectively address the subject and to critique the practice. Therefore, and foremost, an explicit definition and protocol need to be written. In the meantime, the committee focused on the simple model of two latent prints from two fingers that may have originated from one hand of a single person and could have been placed on an object contemporaneously. Two assumptions are made: (1) the two impressions are related contemporaneously; and (2) even though there are not sufficient quantity and clarity in any one impression, the weight of the combination of features from two or more impressions is equal to or greater than an equivalent amount of data from a suitable identification that could be made if all of the features were located in a single impression. In other words, the features and relationships are not restricted to a small region of friction ridge skin.

Assuming hypothetically that an examiner requires a minimum of seven points in any configuration sufficient for effecting an identification, in simultaneous prints, the seven points could be apportioned between the two fingers. In assessing the independence of features or lack thereof (as described above), one also should consider the independence (or relationship) of features across fingerprints and lower joints on the same hand, not only those in the same fingerprint, if simultaneous impression interpretations are to be used. To justify extracting partial information from two or more impressions and then combining them requires testing that the combination (of less-than-threshold features per impression) is equal to or greater than some threshold requirement.

The assumption that two nearby impressions are from the same individual and have been deposited contemporaneously can be addressed only on a case-by-case basis. Some might say that for some scenarios two prints found together may have been deposited at different times and thus may not be from the same source. Alternatively, if an item could only be held in a certain manner, then the only way of explaining the evidence is that the multiple prints are from a single person. In some cases, identifying simultaneous prints may infer, for example, the manner in which a knife was held. It may be better to define simultaneous prints as “cluster impressions or prints” so as not to infer the timing of the deposition of the multiple prints. However, before proceeding, more explicit guidelines on when it is appropriate to assume that prints are simultaneously deposited need to be created, and it should be required that any assumptions made be stated in the examiner’s report.

VII. Exclusions

The general practice in the field of latent prints is that of “making an identification.” Simply because no latent print of sufficient quality and quantity was found with features similar to the suspect does not mean that the suspect did not handle the evidence. Someone can handle an object and leave no latent print(s); therefore, practitioners espouse that no one can ever be excluded as having touched the evidence. In keeping with this philosophy, a latent print examiner tends to approach the comparison to “make an ident,” rather than to attempt to exclude. This concept is similar to any other forensic analysis in that a lack of evidence does not necessarily exclude a suspect. However, it contrasts slightly with the doctrine of other forensic science disciplines. In forensic science examinations, regardless of the discipline, a pattern or profile (or some other data) is generated from the evidence, and it is compared with that obtained from a reference sample(s) in an attempt to exclude the two samples as having originated from the same source. When an examiner fails to exclude, then some significance is placed on that observation or finding. The more powerful or resolving the analysis, the more likely it is that wrongly associated samples will be excluded. The tremendous variability observed in friction ridge skin makes analysis of latent prints one of the most powerful exculpatory tools available to the forensic scientist. In fairness, an examiner does look for discrepancies in ridge detail that would result in an interpretation of exclusion. However, this approach is implemented only for prints deemed suitable for comparison.

In the first step of ACE-V, the examiner analyzes a latent print to determine if it is suitable for comparison. However, some prints may not meet this criterion, but they may provide exculpatory information. This can depend on how one frames the focus for exclusion. As stated above, no one can ever be excluded as handling the evidence because a person can touch an object and not leave a latent print. Alternatively, a defendant may desire to know if there are latent prints on the object demonstrating that someone else did handle the object.

An example may illustrate the point. Consider the recovery of a latent print on a glass found at a crime scene. The print is degraded such that the quality of Level II and Level III features is too poor to proceed to the comparison phase of the examination. The print is therefore declared “of no value” and discarded. No further work will be performed on this print. In this scenario, the Level I features clearly present the pattern of a whorl. Now assume that a suspect is apprehended and his or her fingerprints (and possibly other areas of friction ridge skin) have no whorls. In this case, even though it lacked sufficient quality or detail for an identification, the pattern would have excluded the suspect as the source of any prints found on the evidence. The defense and prosecution may want to know if the evidence revealed that someone other than the defendant handled the evidence. Exclusions are a very useful investigative tool and are currently underutilized.

The issue of exculpatory power of evidence is complex but needs further investigation. Some interviewees suggested to the committee that patterns insufficient for an identification could be artifactual. Thus false-positive and false-negative results could be obtained. This needs to be further studied and documented. It would not be wise to recommend a procedure that may have an inherently high error rate. There also is a tremendous resource consideration. If it were deemed reliable to proceed with the exculpatory model, then a substantially larger workforce (and concomitant resources) would be needed, case backlogs would increase, and more storage facilities would be required.

VIII. IAFIS Searches

An examiner encodes the minutiae on a latent print and then enters the data into the Integrated Automated Fingerprint Identification System (IAFIS) to search for possible matches. When an examiner uses IAFIS, false-positive and false-negative (missed) minutiae from reference prints may be encoded. In addition, IAFIS is not as good as the examiner for interpreting distorted patterns and uses only partial detail. Thus the list of matches constitutes candidates, not absolute identity. Based on FBI LPU experience, about 82 percent of the time that the true matching reference pattern (i.e., mate) is on the list, it is in the top candidate position. The other 18 percent of the time, approximately one-third of the true matching reference patterns reside at position number 2 (Table 1). It would be desirable to improve the functionality so that the true mate is the top candidate significantly more than 82 percent of the time (and raise the ranking of those farther down the list). Some interviewees also suggested that if a second examiner encoded the same print, he or she may not select the same minutiae. Thus the pattern used by the second examiner for searching in IAFIS is different than that used by the first examiner. In turn, the candidate list could be different to some degree. Perhaps if two examiners independently encoded minutiae on the latent print, the results of each of their searches combined may increase the success rate of identifying the true mate, provided, of course, that the mate is in the database. A study should be considered to test the effects of two examiners’ encoding a print and launching a search on IAFIS. A cost-benefit analysis should be considered here as well.

Table 1: Position of the True Mate on an IAFIS Candidate List (Data from IAFIS Searches Conducted at the FBI Laboratory Through October 2004)

IX. Sourcebook

In the course of this review, the committee found that papers, studies, and other data were not all collated to facilitate an analysis or to be useful for training. The existence of a sourcebook would have facilitated this review. Furthermore, the committee may have found that some of its recommendations had been addressed to a greater degree than was apparent. Thus the committee recommends the development of a sourcebook and formalization of a notebook of the data collected operationally by the FBI LPUs. The sourcebook should be coordinated with members of the Scientific Working Group on Friction Ridge Analysis, Study and Technology (SWGFAST), which also recommends the development of a sourcebook.

X. Conclusions

This committee’s review found overwhelming evidence that latent print examinations can be carried out and that reliable identifications can be made. However, there are scientific areas where improvements in the practice can be made, particularly regarding validation, more objective criteria for certain aspects of the ACE-V process, and data collection. The main benefit would be to better ensure the consistency of interpretation practices across the field. The recommended projects are summarized in Table 2 and are divided into two categories (High Priority and Priority).

Table 2: List of Recommendations Sorted into High-Priority and Priority Categories

Note: No ranking of priority is made within each category.

1. High-Priority Projects

1. Quality
   
   
1. Develop guidelines for describing quality metric features.
   
   
2. Test whether 12-point system is correlated with total number of points and clarity.

1. Quantity
   
   
1. Test hypothesis of independence of features.
   
   
2. Test hypothesis that there is no scientific basis for minimum point threshold.
   
   
3. Establish a quantitative model for identification.
   
   
4. Survey Latent Print Units (and community) to determine if unwritten minimum threshold of seven detectable points is applied routinely.
   
   
5. If seven-point minimum threshold (or whatever is used by majority) is generally accepted, test with statistical models and by black-box approach.

1. Performance
   
   
1. Establish minimum number of features that can be evaluated pragmatically in friction skin ridge casework comparisons.
   
   
2. Test performance of examiner as a black box rigorously in a controlled manner.
   
   
3. If a minimum threshold for an identification can be developed, test a selected panel of latent print examiners.

1. Exclusions
   
   
1. Review value and reliability of exculpatory power of evidence.

1. Priority Projects

1. Permanence Test
   
   
1. Test hypothesis of permanence of Level III features.
   
   
2. Test hypothesis of permanence of features on the lower joints, soles, and palms.

1. Data Collection
   
   
1. Test existing algorithms and collect existing data for review.
   
   
2. Develop well-defined protocol(s) describing the process for recording, collating, evaluating, and editing research materials.
   
   
3. Develop a sourcebook and collate existing data within the Latent Print Units (and with members of the Scientific Working Group on Friction Ridge Analysis, Study and Technology).

1. Cluster (Formerly Simultaneous) Impressions
   
   
1. Develop more explicit definitions on cluster prints and guidelines on when it is appropriate to assume that cluster prints are deposited simultaneously.
   
   
2. Test hypothesis of independence of features across fingerprints and lower joints on the same hand (simultaneous impression interpretations).

1. Additional Validation Studies
   
   
1. For quality testing, develop method to artificially generate patterns and test degree of variation at which incorrect matches are made.
   
   
2. Assess accuracy of representation of the friction ridge detail on the finger when using the image-capture systems that record reference prints.
   
   
3. Test impact of two examiners’ independently encoding a print and launching a search on the Integrated Automated Fingerprint Identification System.

In the committee’s opinion, all of the recommendations in Table 2 are considered priorities for research efforts. Projects within each category are listed in no particular order. Additionally, several of these proposed efforts are best performed via a hierarchical, rather than a parallel, approach. That is, the results of one project will assist in the research design and methodology for subsequent projects. Because performing all of the proposed research projects and implementing the recommendations require substantial monetary and personnel resources, final decisions regarding prioritization of projects will need to be made. The committee concluded that although the use of friction ridge skin impression pattern analysis is fundamentally sound, additional studies could improve confidence in the results obtained, provide guidelines for more consistent practices throughout the latent print community, and provide metrics for assessing performance.

Acknowledgments

In the course of this review, the following experts outside the FBI were interviewed: Mr. Rockne Harmon (Deputy District Attorney, Oakland, California); Dr. David Kaye (Professor, Arizona State University College of Law); Ms. Susan Narveson (Chief, Investigative and Forensic Sciences Division, National Institute of Justice, Washington, D.C.), Mr. Barry Scheck (Professor, Yeshiva University, Benjamin N. Cardozo School of Law, New York, New York); and Dr. David Stoney (Chief Scientist, Stoney Forensic, Inc., Clifton, Virginia). The committee greatly appreciates their candor, advice, and time. The committee also thanks the members of the FBI Laboratory Division Latent Print Units for their time and willingness to honestly answer probing questions about their work practices and the history of their profession.

The opinions expressed are those of the authors and may not necessarily be ascribed to any of the individuals who were interviewed.

This is publication number 05-12 of the Laboratory Division of the Federal Bureau of Investigation. Names of commercial manufacturers are provided for identification only, and inclusion does not imply endorsement by the FBI.

Suggested Readings

Amy, L. Valeur de la preuve en dactyloscopie I, Journal de la Societe de Statistique de Paris (1946) 88:80–87.

Amy, L. Valeur de la preuve en dactyloscopie II, Journal de la Societe de Statistique de Paris (1946) 88:189–195.

Ashbaugh, D. Ridgeology: Modern evaluation friction ridge identification, Journal of Forensic Identification (1991) 41:16–64.

Ashbaugh, D. Quantitative-Qualitative Friction Ridge Analysis. CRC Press, Boca Raton, Florida, 1999.

Ashbaugh, D. Incipient ridges and the clarity spectrum, Journal of Forensic Identification (1992) 42:106–114.

Ashbaugh, D. Premises of friction ridge identification, clarity, and the identification process, Journal of Forensic Identification (1994) 44:499–513.

Babler, W. Quantitative differences in morphogenesis of human epidermal ridges. In: Dermatoglyphics—Fifty Years Later. Birth Defects Original Articles Series. March of Dimes, Washington D.C. (1979) 15(6):199–208.

Babler, W. Embryologic development of epidermal ridges and their configurations. In: Dermatoglyphics: Science in Transition. Birth Defects Original Article Series. March of Dimes, Washington D.C. (1991) 27(2):95–112.

Babler, W. Prenatal development of dermatoglyphic patterns: Associations with epidermal ridge, volar pad and bone morphology, Collegium Antropologicum (1987) 11(2):297–303.

Babler, W. Prenatal communalities in epidermal ridge development. In: Trends in Dermatoglyphic Research. Kluwer Academic Press, Dordrecht, Netherlands, 1990, pp. 54–68.

Babler, W. Prenatal selection and dermatoglyphic patterns, American Journal of Physical Anthropology (1978) 48:21–28.

Bonnivie, K. Studies on papillary patterns of human fingers, Journal of Genetics (1925) 15:1–112.

Bridges, B. C. No duplicate finger prints, Finger Print Magazine (March 1946) 5–6.

Chacko, L. and Vaidya, M. The dermal papillae and ridge patterns in human volar skin, Acta Anatomica (1968) 70(1):99–108.

Chakraborty, R. The role of heredity and environment on dermatoglyphic traits. In: Dermatoglyphics: Science in Transition. Birth Defects Original Article Series. March of Dimes, Washington D.C. (1991) 27(2):151–191.

Cummins, H. Epidermal ridge configurations in developmental defects, with particular reference to the ontogenetic factors which condition ridge direction, The American Journal of Anatomy (1926) 38(1):89–151.

Cummins, H. and Midlo, C. Fingerprints, Palms and Soles: An Introduction to Dematoglyphics. Dover Publications, Inc., New York, 1961.

Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 US, 579 (1993).

Dell, D. and Munger, B. The early embryogenesis of papillary (sweat duct) ridges in primate glabrous skin: The dermatotopic map of cutaneous mechanoreceptors and dermatoglyphics, The Journal of Comparative Neurology (1986) 244:511–532.

Evett, I. W. and Williams, R. L. A review of the sixteen points fingerprint standard in England and Wales, Journal of Forensic Identification (1996) 46:49–73.

Hale, A. R. Morphogenesis of volar skin in the human fetus, The American Journal of Anatomy (1952) 91(1):147–173.

Holbrook, K. A. Structure and function of the developing human skin. In: Biochemistry and Physiology of the Skin. L.A. Goldsmith, ed. Oxford University Press, New York, 1991, pp. 64–101.

Langenburg, G.M. Pilot study: A statistical analysis of the ACE-V methodology—Analysis stage, Journal of Forensic Identification (2004) 54:64–79.

Loesch, D. Quantitative Dermatoglyphics: Classification, Genetics, and Pathology. Oxford University Press, New York, 1983.

Meier, R. Sequential developmental components of digital dermatoglyphics, Human Biology (1981) 53(4):557–573.

Misumi, Y. and Akiyoshi, T. Scanning electron microscopic structure of the finger print as related to the dermal surface, The Anatomical Record (1984) 208:49–55.

Montagna, W. and Parakkal, P. The Structure and Function of Skin. 3rd ed., Academic Press, Orlando, Florida, 1974.

Okajima, M. A methodological approach to the development of the epidermal ridges viewed on the dermal surface of fetuses. In: Progress in Dermatoglyphic Research. Alan R. Liss, Inc., New York, 1982, pp. 175–188.

Okajima, M. Development of dermal ridges in the fetus, Journal of Medical Genetics (1975) 12:243–250.

Osterburg, J. An inquiry into the nature of proof: The identity of fingerprints, Journal of Forensic Sciences (1964) 9:413–426.

Osterburg, J., Parthasarathy, T., Raghavan, T., and Sclove, S. Development of a mathematical formula for the calculation of fingerprint probabilities based on individual characteristics, Journal of the American Statistical Association (1977) 72(360):772–778.

Pankanti, S., Prabhakar, S., and Jain, A. K. On the individuality of fingerprints. In: IEEE Transactions in Pattern Analysis and Machine Intelligence. IEEE Computer Society, Washington, D.C., 2002, Vol. 24, No. 8, pp. 1010–1025.

Penrose, L. Dermatoglyphics, Scientific American (December 1969) 72–84.

Penrose, L. and O’Hara, P. The development of epidermal ridges, Journal of Medical Genetics (1973) 10:201–208.

Pollitzer, W. S. and Plato, C. C. Anthroplogy and dermatoglyphics, Birth Defects (1979) 15(6):211–223.

Roberts, D. Population variation in dermatoglyphics: Field theory. In: Progress in Dermatoglyphics Research. Alan R. Liss, New York, 1982, pp. 79–91.

Roxburgh, T. On evidential value of fingerprints, Indian Journal of Statistics (1933) 1:89–214.

Schneickert, J. H. Finger prints of the mono-ovum twins, Finger Print Magazine (January 1938) 16–31.

Sclove, S. The occurrence of fingerprint characteristics as a two dimensional process, Journal of the American Statistical Association (1979) 74:588–595.

Siervogel, R., Roche, A., and Roche, E. Developmental fields for digital dermatoglyphic traits as revealed by multivariate analysis, Human Biology (1978) 50:541–556.

Slatis, H., Katznelson, M., and Bonne-Tamir, B. The inheritance of fingerprint patterns, American Journal of Human Genetics (1976) 28:280–289.

Smith, L. T. and Holbrook, K. Embryogenesis of the human skin, Pediatric Dermatology (1986) 3:271–280.

Stoney, D. Distribution of epidermal ridge minutiae, American Journal of Physical Anthropology (1988) 77:367–376.

Stoney, D. A. Measurement of fingerprint individuality. In: Advances in Fingerprint Technology. 2nd ed., CRC Press, Boca Raton, Florida, 2001, pp. 327–387.

Stoney, D. and Thornton, J. A critical analysis of quantitative fingerprint individuality models, Journal of Forensic Sciences (1986) 31:1187–1216.

Stoney, D. and Thornton, J. A method for the description of minutia pairs in epidermal ridge patterns, Journal of Forensic Sciences (1986) 31:1217–1234.

Stoney, D. and Thornton, J. A systematic study of epidermal ridge minutiae, Journal of Forensic Sciences (1987) 32:1182–1203.

Stucker, M., Geil, M., Kyeck, S., Hoffman, K., Rochling, A., Memmel, U., and Altmeyer, P. Interpapillary lines—The variable part of the human fingerprint, Journal of Forensic Sciences (2001) 46:857–861.

Taroni, F., Champod, C., and Margot, P. A. Forerunners of Bayesianism in early forensic science, Journal of Forensic Identification (1999) 49:285–305.

Trauring, M. Automatic comparison of finger-ridge patterns, Nature (1963) 197:938–940.

Tu, P. and Hartley, R. Statistical significance as an aid to systems performance evaluation. In: ECCV 2000. Vol II, 2000, pp. 366–378.

Vanderkolk, J.R. Forensic individualization of images using quality and quantity of information, Journal of Forensic Identification (1999) 49:246–255.

Weninger, M., Aue-Hauser, G., and Schieber, V. Total finger ridge-count and the polygenic hypothesis: A critique, Human Biology (1976) 48:713–725.