The below known non-matches display some of the largest and best amounts of corresponding ridge formations ever recorded in a non-match.  Images 1-6 are look-alikes displayed on Kasey Wertheim's web site www.clpex.com:

 Image #1

Close-up of non-match #1

 Image #2

Overview of non-match #1 

 Image #3

A non-match displayed on Kasey Wertheim's web site:  www.clpex.com 

 Image #4

A non-match displayed on Kasey Wertheim's web site:  www.clpex.com 

 Image #5

Overview of non-match #6  

 Image #6

A non-match displayed on Kasey Wertheim's web site:  www.clpex.com 

The following look-alike (Images #7 and #8) was presented by Pat Wertheim during fingerprint examiner training [20]:

 Image #7

Latent look-alike presented by Pat Wertheim. 

 Image #8

Exemplar look-alike presented by Pat Wertheim. 

The Shirley McKie Case

What is considered by many to be the 2nd most notable erroneous identification made in the history of fingerprint identification was attributed to the Scottish Criminal Records Police (SCRP)  "identification" of a latent fingerprint found at a Marion Ross homicide crime scene to police woman Shirley McKie.  The following images (#9 and #10) show the latent fingerprint, enlarged and charted, that the SCRP identified with "absolute confidence" to the exemplar thumb print, also enlarged and charted, belonging to Shirley McKie (the images are from Ed German's web site [22]):
 

 Image #9

Charted exemplar fingerprint of Shirley McKie 

 Image #10

Images #9 and #10 display more or less look-alike ridge features.  Similar to the Madrid Error, the fingerprints display a quality, albeit poor, of being a double, a twin.  Also similar to the Madrid error, the SCRP only used professional judgment to establish basis for sufficiency to individualize.  They had no tool other than human decision-making to weigh amounts of corresponding ridge formations present in two impressions and to determine whether or not the two impressions could be attributed as having come from the same source with a degree of probability that borders on certainty.

Basis for Invalid Individualization 

Based on a conservative upper-bound total human population for Scotland, e.g., roughly 5 million people, and application of T-Model formulae and reduction factors for reduced levels of ridge feature clarity, reliability and quality of agreement, the estimated T-Value for the small, distorted amount of "corresponding" ridge features present in the charted enlargements, clearly fails to establish an estimate for a number of fingerprint close matches or look-alikes as less than 1.  Therefore, based on T-Model theory, there is no valid basis to establish inference for positive identification*.

* It is significant to point out that human interpretation and assessment of ridge feature types, clarity and quality of agreement values are unavoidably subjective.  Nevertheless, when T-Model general guidelines for these areas are strictly followed and conservative lower bound values are applied, the final conclusion regarding the number of close matches or look-alikes estimated to be present in the 50 million fingerprints is repeatedly more than 1, and therefore the amount of "matching" ridge features present in the two impressions is insufficient to infer identification. 

Note:  Latent print examiner's are encouraged to not blindly accept the opinion of the author but to interpret the "corresponding" ridge features in the two impressions for themselves, assess the clarity, reliability and quality of agreement for each individual ridge feature based on guidelines defined by the T-Model, apply conservative lower bound values for ambiguity, and independently use T-Molde formulae or the T-Model fingerprint calculator to estimate the number of close matches or look-alikes likely to exist.   When personal interpretations and assessments for each individual ridge feature made by the examiner is fixed and documented, then it is easily subject to critical scrutiny.   Only then can incorrect interpretation and/or assessment by an untrained examiner be ferreted out.   

NEXT PAGE >>> 

If the T Model, or equivalent tool, had been used by the SCRP, then the Shirley McKie error may have never occurred.