How Handwriting Analysis Works


1956 Weinberger kidnapping: An FBI-conducted analysis made a match between the top two sign-offs ("Your baby sitter"), pulled from the ransom notes, and the bottom two, from the prime suspect. See more forensics pictures.

When there's a suspect in a crime and the evidence includes a handwritten note, investigators may call in handwriting experts to see if there's a match. In some cases, it might be the one piece of evidence that gets a suspect charged and eventually convicted. But what if it's a false match? How exactly do experts go about analyzing someone's handwriting?

In the world of forensic analysis, which includes crime scene investigation, DNA testing, fiber analysis, fingerprint analysis, voice identification and narcotics analysis, to name just a few of the disciplines, handwriting analysis fits into the area of questioned documents. Questioned document examiners (QDEs) analyze documents for signs of alteration, forgery and, when sample documents are available, handwriting or typing comparisons to determine or rule out authorship (and/or tie a document to a specific machine in the case of typing). Handwriting analysis is a tedious and methodical process that relies on extensive knowledge of the way people form letters, which characteristics of letter formation are unique and the physiological processes behind writing - the ways in which a person's fine-motor skills can affect his or her handwriting and leave clues about the author's identity.

The primary basis of handwriting analysis as a science is that every person in the world has a unique way of writing. When we were all kids in primary school, we learned to write based on a particular copybook - a style of writing. Which copybook our handwriting is based on depends on when and where we grew up (see Handwriting-L: Copybook Examples for good examples of copybooks from different countries and eras). So at first, we all probably wrote in a similar way to kids of our own age and location. But with the passing of time, those writing characteristics we learned in school - our style characteristics - became only the underlying method of our handwriting. We developed individual characteristics that are unique to us and distinguish our handwriting from someone else's. Most of us don't write the way we did in first or second grade. And while two or more people may share a couple of individual characteristics, the chance of those people sharing 20 or 30 individual characteristics is so unlikely that many handwriting analysts would say it's impossible.

First and foremost, handwriting analysts must be able to accurately distinguish between style characteristics and individual characteristics, which takes a lot of training. They can pretty much ignore the style characteristics, which are only useful for determining with a fair degree of certainty which copybook the writer learned from. The individual characteristics are what matter the most in determining authorship.­

So the process of handwriting analysis when comparing two documents - one by a known au­thor, one by an unknown author - starts not with checking for similarities, which any of us could do with a fair degree of accuracy, but instead with checking for differences. It's the differences that initially determine if it's possible that the same person wrote both pieces of text. If the­re are key differences in enough individual characteristics, and those differences do not appear to be the result of simulation (an attempt to disguise one's handwriting or copy someone else's), then the two documents were not written by the same person. Simulation has its own telltale characteristics, which we'll discuss in the next section. However, if the differences don't rule out a match, and there are significant similarities in the individual traits in the two documents, singular authorship becomes a possibility.

Moving from possibility to probability is where the heavy lifting comes in.

Analyzing a Sample

Partial results of handwriting analysis in Lindbergh kidnapping case
Partial results of handwriting analysis in Lindbergh kidnapping case
Photo courtesy FBI

­Analyzing handwriting is a long, careful process that takes a lot of time and, under ideal circumstances, a lot of comparison samples, or exemplars - documents that have a known author. It's not a matter of looking at two documents­ and saying "Hey, they both have a 'B' with a downstroke extending below the line - same author!" In the case of the Lindbergh kidnapping in 1932, the police had a slew of questioned documents- in all, the kidnapper sent 14 notes to Lindbergh with ransom instructions. Handwriting analysts had no problem determining that all of the ransom notes were written by the same person. But pre-existing exemplars from the main suspect, Richard Bruno Hauptmann, were scarce, so the police had to get samples from Hauptmann in the police station by way of dictation. From those requested exemplars, handwriting analysts determined a match.

­However, the police officers' methods of obtaining those samples has since been called into question - Hauptmann was forced to write for hours and hours until he nearly passed out from exhaustion. He was also told how to write, and he was shown a ransom note and told to copy the handwriting as best he could, to name just a few of the major no-nos. This of course means that the validity of the determined handwriting match is in question, and Hauptmann's execution makes a re-test impossible. There are now strict rules in place about how police should obtain a requested exemplar.

Good, untainted, numerous exemplars make handwriting analysis far more reliable than a simple one-to-one comparison. While every person's handwriting is unique, no one person writes exactly the same way twice. There are natural variations in a person's writing within a single document. Having, say, 10 questioned documents and 10 exemplars from a suspect assures that not only will the words and letters in the questioned documents show up in the exemplars, but also that almost all of that suspect's individual characteristics will appear in both sets of samples if the suspect is indeed the author of both.

The process of professional, forensic handwriting analysis is all about thoroughness. An analyst will use a magnifying glass and sometimes even a microscope in the comparison process. An analyst is looking for a wide array of individual traits:

  • Letter form - This includes curves, slants, the proportional size of letters (relationship between size of short and tall letters and between the height and width of a single letter), the slope of writing and the use and appearance of connecting lines (links) between letters. It's worth noting that a person may form a letter differently depending on where the letter falls in a word - beginning, middle or end. So an analyst will try to find examples of each letter in each placement.
  • Line form - This includes how smooth and dark the lines are, which indicates how much pressure the writer applies while writing and the speed of the writing.
  • Formatting - This includes the spacing between letters, the spacing between words, the placement of words on a line and the margins a writer leaves empty on a page. It also considers spacing between lines -- in other words, do strokes from words on one line intersect with strokes in words on the line below and above it?

With these traits in mind, we'll take a look at one common method of comparison in which the analyst begins with the first letter in the first word in the questioned document and starts building a table. To illustrate the process, we'll perform a simulated comparison using a questioned document and an exemplar each consisting of a single sentence - a nightmare for determining a definite match or non-match in the real world, but perfect for our purposes.

Questioned document
Questioned document
Exemplar provided by "suspect"
Exemplar provided by "suspect"

At first glance, the two samples do not appear to be different enough to automatically declare that they were written by two different people. And on closer examination, they actually appear to be quite similar. So we're going to build a table that catalogs each varied form of every letter that appears in the questioned version of this sentence. If we get to an "a" that looks exactly like an "a" we already have in our table, we skip it. We simply want each different formation of "a" in the document in our table, taking into consideration letter formation, linking strokes, spacing and other traits. In forensics, they would "copy" each letter form using a digital camera, but we're going to do it by hand. They would also make separate tables for uppercase and lowercase letters. But we're going to simplify, since this is only an example of the process of determining a match or a mismatch and not a professional or accurate comparison.

Next, we're going to make the same kind of table using the exemplar:

Finally, we're going to compare the tables and see if we have a match for each questioned-document letter form in the exemplar. Since our document consists of a single sentence, we don't have many instances to choose from. Under normal circumstances, we would have an array of potential matches for each letter form, and we'd want to find a good match in the exemplar for every letter form occurring in the questioned document. For simplification purposes, our third table is going to consist of a side-by-side comparison of our two initial tables, although an expert analyst would probably create a third table showing the exact words in each document that make up each letter-form match.

Questioned (left) and exemplar tables
Questioned (left) and exemplar tables

While this analysis would definitely not hold up in court due to its extremely limited scope (and sadly inaccurate letter copying), it nonetheless seems we have found a match in the exemplar for each letter in the questioned document. The same person probably wrote both sentences.

But what if the writer of the exemplar was trying to copy the handwriting in the questioned document? The problem of simulation is a difficult one in handwriting analysis. Simulation occurs when a person is either trying to disguise his handwriting to prevent the determination of a match or to copy someone else's handwriting to encourage the inaccurate determination of a match. While simulation makes an accurate analysis far more difficult (and sometimes impossible), there are certain traits that professional analysts look for to determine whether a handwriting sample is the result of simulation. These include shaky lines, dark and thick starts and finishes for words and a lot of pen lifts, all which come from carefully, slowly forming letters instead of writing quickly and naturally.

Mickey Mantle's known signature is on top; the FBI determined that the bottom two signatures are forgeries. Note the shaky line quality and variations in the starting and ending strokes.
Mickey Mantle's known signature is on top; the FBI determined that the bottom two signatures are forgeries. Note the shaky line quality and variations in the starting and ending strokes.

Simulation is just one of the factors that can foil an accurate handwriting analysis.

Shortcomings of Handwriting Analysis

­While an expert analyst can detect many instances of forgery, a good simulation can be undete­ctable. One example of a forgery the experts missed is the case of the "lost" Hitler diaries. (Although there's a good reason why they missed it.)

In the 1980s, a man named Konrad Kujau, a supposed collector of Nazi memorabilia, approached a German publishing company with 60 handwritten journals purported to be written by Adolf Hitler that had, according to Kujau, just been discovered in the wreckage of an airplane that had left Germany after World War II. The texts seemed to be genuine, and Kujau had an apparently good reputation, so the publishing company paid $2.3 million for the lot. The diaries were immediately published in installment form in a German newspaper owned by the same publishing company, and syndication rights were sold to several international publications, including The London Times. It was The Times that requested a professional handwriting analysis to ensure authenticity.

Three international experts in forensic handwriting analysis compared the diaries to exemplars that were apparently known to be written by Hitler. All agreed that the diaries were written by the same person who wrote the exemplars. The diaries were for real.

It was an analysis of the ink and paper used to write the diaries that revealed them as fakes. An ultraviolet-light examination revealed that the paper contained an ingredient that wasn't used in paper until 1954. Hitler died in 1945. Further forensic tests on the ink showed it had been applied to the paper within the last 12 months. As it turns out, though, the handwriting analysis was in fact correct - the person who'd written the diaries had also written the exemplars. Kujau, later found out to be an experienced con artist, had also forged the exemplars the police were using as comparison documents.

The Hitler diaries debacle is an extreme case of fraud and expert forgery that spanned every stage of the analysis. And while this level of expertise is seldom found in forgeries, the fact remains that if the investigation had relied on handwriting analysis alone, the "lost Hitler diaries" would now be part of the history books. Some other issues affecting the accuracy of handwriting analysis include:

  • You can't make a meaningful comparison between uppercase and lowercase letters.
  • Drugs, exhaustion or illness can significantly alter a person's handwriting.
  • The quality of the exemplars determines the quality of a comparison analysis, and good exemplars can be hard to come by.

­In the initial comparison work done in the case of John Mark Karr, who confessed in August 2006 to the 1996 murder of six-year-old JonBenet Ramsey in Colorado, the ransom note found in the Ramsey house was long enough to be useful as one side of the equation, but finding good exemplars was an issue. In a series of preliminary handwriting analyses, documents expert John Hargett, former head of document analysis at the U.S. Secret Service, compared the ransom note to two exemplars: a yearbook inscription written when Karr was in high school and a job application Karr filled out in Thailand. Hargett found no matches, although the results were inconclusive because the yearbook inscription was written more than 20 years ago and in an artistic writing style, and Karr filled out the Thailand job application in all uppercase letters, while the ransom note was written in both uppercase and lowercase letters. DNA testing later made further handwriting comparisons unnecessary, as Karr's DNA was not a match for the DNA found on JonBenet's body.

Certainly the most significant shortcoming of handwriting analysis as a science is the fact that it is ultimately subjective. This means that its acceptance in the scientific community and as evidence in court has historically been shaky. Only recently, as the training of analysts has become more standardized and certification procedures have been put in place, has handwriting analysis started to gain more acceptance as a reproducible, peer-reviewed scientific process. The results of a handwriting comparison are still not always accepted as evidence in a court case, partly because the science has a few more hurdles to clear, including determining a reliable error rate in analysis and setting standards for the comparison process. The addition of computerized handwriting analysis systems to the process, including the FISH (Forensic Information System for Handwriting) system, which allows examiners to scan in handwritten documents and digitize the comparison process, may speed up the process of general acceptance of handwriting analysis as a science and as expert evidence in court.

For more information on handwriting analysis and related topics, check out the links on the next page.

Related HowStuffWorks Articles

More Great Links

Sources

  • Cohen, Fritz. "Scientific Handwriting Analysis." National Handwriting Association. http://www.nha-handwriting.org.uk/Article_5.htm
  • Found, B., & Rogers, D. "Documentation of Forensic Handwriting Comparison and Identification Method: A Modular Approach." Journal of Forensic Document Examination. 1999. http://www.latrobe.edu.au/humanbio/forensic/method/intro.pdf#search=%22handwriting%20comparison%22
  • "Handwriting, Typewriting, Shoeprints, and Tire Treads: FBI Laboratory's Questioned Documents Unit." FBI.gov. http://www.fbi.gov/hq/lab/fsc/backissu/april2001/held.htm
  • "JonBenet suspect to return to U.S. on Sunday." CNN.com. Aug. 19, 2006. http://www.cnn.com/2006/LAW/08/18/karr.questions/index.html
  • "Is handwriting analysis legit science?" The Straight Dope. http://www.straightdope.com/columns/030418.html
  • "LIMA: Forensic Handwriting Analysis." Warwick Centre for the Study of the Rennaissance. http://www2.warwick.ac.uk/fac/arts/ren/publications/lima/handwriting/forensic/
  • Ramsland, Katherine. "Literary Forensics." CourtTV Crime Library. http://www.crimelibrary.com/criminal_mind/forensics/literary/2.html
  • "The Thornton Handwriting Examination Court Decision." Forensic-Evidence.com. http://forensic-evidence.com/site/ID/handwrtg_Thornt_ID.html