I wanted to expand on that mention of Twain, and it seems it will take more than one blog post to do it. So here's Part 1!
These are two essential characteristics.
Let's say there is a bank robbery. Witnesses say the robber has long, red hair and was wearing dark clothing.
Police know that millions of people have red hair and a good number of those may also own dark clothing. A smaller percentage of those millions may have been able to be at the bank at that time, but this will still be a very large group. While the witnesses’ descriptions may be of help in immediately following a suspect from a scene, or prove valuable if the robber is caught some other way, they don't narrow the field much.
One other important element of determining identification is mutability. Can this trait be changed, or appear to be changed? In our example above, hair can be dyed and cut, clothing can be easily changed. So again, the witnesses description quickly loses value.
At the time Twain wrote "A Thumb-Print," if you talked to the average person on the street and said that if a criminal touched something with an ungloved hand, he might leave a uniquely identifying mark that could lead to his conviction, you'd have probably been met with disbelief. This would have sounded like voodoo to most of his contemporaries.
Until very late in the 19th century, even those who were studying scientific approaches to catching criminals were largely unaware of the potential of fingerprints, and no system of classification of fingerprints had been established in law enforcement organizations.
The first scientific treatise on the subject to be written in English and to become widely known was not published until 1892 — Sir Francis Galton’s Finger Prints.
How did Twain know?
More on the history of the forensic use of fingerprints.
And links for those who want to know more.