Comments on Handwriting Analysis and the Mar Saba Letter

[Peter Kirby]

I made an earlier post in a similar thread here

To paraphrase: Tselikas did not strictly speaking establish that the handwriting is later than the 18th century, What Tselikas did claim to establish is that although the handwriting is in one sense an 18th century Greek cursive, the scribe is not writing cursive because this is how they learnt to write rapidly and legibly. Instead the scribe is carefully and deliberately writing in a script that does not come naturally. This sort of conclusion is standard palaeography (analyzing the ductus of the script).

Such a conclusion is strictly speaking entirely compatible with an 18th century writer. However it is IMO more plausibly explained by a later writer imitating 18th century handwriting,

Andrew Criddle

One note here - regarding analyzing the ductus of the script - Tselikas writes:

The first impression created by the first glance at the manuscript of the Clement’s letter is that the writing belongs to the late 17th century until the late 18th. Also that the scribe is experienced, and keeps constant ductus in the design of the letters.

One of the features that would betray that it isn't actually an 18th century hand would be inconsistency in the ductus of the design of the letters, so it is natural that Tselikas would be looking to see if he could find evidence of that kind of inconsistency. His report says he didn't find it but instead found that the scribe maintains a "constant ductus in the design of the letters," indicating that "the scribe is experienced" in this form of writing, allowing this consistency to be maintained for the three pages of writing.

Then the next point that Tselikas makes is:

A big number of lines of the letters and links are not continuous, fact which means that the hand of the scribe was not moving spontaneously, but carefully and tentatively to maintain the correct shape of the letter.

The point on continuity indicates that the writer is picking up their pen from the paper before setting down again. Describing the consistency of the ductus requires standard paleographic expertise, as you point out. Describing the amount of discontinuity in the writing is also objectively a feature of the text.

Something not done here by Tselikas -- such as providing comparison with other writing with respect to the amount of discontinuity in the writing -- can also be done at the descriptive level, in a verifiable way.

The inferences made from the facts mentioned -- whether the writer was "careful," "deliberate," "tentative," and "not moving spontaneously" -- those are all inferences that can be offered and discussed. But it's correct to make a distinction here, that these claims are of a different nature. The opposite claims -- that the writer was "careless," "natural," "confident," or "moving spontaneously" -- would also be of a different nature. They would be inferences that are at one further remove from the facts themselves, i.e., the verifiable aspects of the text that can therefore be confirmed by analysis by any other paleographic expert.

Thanks again, Andrew, for your comment.

[Peter Kirby]

At the time that Tselikas was asked for his expert opinion as a paleographer, and by the same people (BAR), the professional forensic document examiner Venetia Anastasopoulou was simultaneously being asked to publish.

https://www.biblicalarchaeology.org/wp- ... -brown.pdf

... Ron N. Morris:

It cannot be over-emphasized that even the completion of a graduate degree program in forensic sciences does not qualify the individual as an expert in any of them. The graduate must still take part in a trainee/apprenticeship program before he is eligible to qualify as a competent, qualified, forensic expert in any forensic science, especially that of a FDE [forensic document examiner]. At the conclusion of his trainee program, the new FDE should continue to work daily with competent, qualified examiners for approximately two or more years before being considered senior enough to work independently.

After quoting these words I recommended that “one of our societies for biblical scholars . . . take on the task of arranging for some highly qualified and suitable professionals to examine the photographs in consultation with experts in eighteenth-century Greek handwriting.” Naturally, I was very pleased when I learned that Hershel Shanks had enlisted a Greek paleographer (an expert in old handwriting and its development) and a Greek questioned document examiner to study Mar Saba 65 and consider whether Morton Smith could have forged it. Shanks asked the two experts to work independently, in order not to influence each other’s opinions. Both received high-resolution scans made directly from both the original black-and-white photographs and the newer color photographs taken by a librarian at the Jerusalem Patriarchate in 1976. Although the paleographer has not yet submitted a report, the document examiner, Venetia Anastasopoulou, has done so, and in April 2010 BAR posted her thirty-nine-page report on their website.

It was asked that what was stated as a conclusion would not be limited to a comparison with Smith's writing:

Although this conclusion generated some reflection among Smith’s accusers, some felt that they could reconcile this finding with the theory that Smith forged the letter by supposing that Smith was responsible for its contents and some unknown person conspired with him to forge the manuscript itself. In response to this position, Shanks asked Anastasopoulou to clarify whether the manuscript shows signs that it is a forgery.

Anastasopoulou clarified additionally about what could be concluded on the basis of the study:

Shanks asked Anastasopoulou to clarify whether the manuscript shows signs that it is a forgery. Her one-page supplementary report lists six indications of natural writing that examiners look for and eleven indications of unnatural writing, and concluded, “The Secret Mark letter, as written in detail in my analysis report, is written in a natural and spontaneous way and in my opinion, does not have such indications so to make us think of a suspicious writing.”

The inferences were, to say the least, different, including:

It is written in high speed and although [it] is a connected writing, there are letters written one by one in a word, but this does not deter the good writing rhythm. ...

The whole writing shows freedom, spontaneity and artistic flair. ...

The writing in the disputed document is smooth, continuous, free ...

In our questioned document it is clear that the writing is automatically written in the paper and more over the letters are written calligraphically and not rigidly and copy booked.

Directly relevant expertise in forensic document examination came to bear on these inferences.

[Peter Kirby]

Forensic document examiners have a genuine area of expertise.

For example, here is a 450 page monograph based on empirical studies, involving collecting data about handwriting, that documents characteristics of disguised writing:

https://core.ac.uk/download/pdf/20535754.pdf

As an area of expertise, it's developed beyond just the question of whether two samples of handwriting are from the same person. While not infallible, forensic document examiners are able to investigate questions of whether handwriting is in a disguised hand, using empirically validated methods.

At least some techniques of recognizing a disguised hand can be regarded as transferable skills for a Greek forensic document examiner, even though the present subject involves what looks prima facie like ca. 18th century handwriting.

Here, based on relevant subject matter expertise, "six indications of natural writing that examiners look for and eleven indications of unnatural writing" were noted.

[Peter Kirby]

https://tuhat.helsinki.fi/ws/files/1252 ... _check.pdf

Tom Davis formulates the methodological basis of handwriting comparison in the following way: “a given writer will tend to produce writing that is idiographic … a given piece of writing can have characteristics that are ascertainable by expert analysis, constant between different writings by the same individual, and unique to that individual.” Idiographic refers to an individual’s writing characteristics, idiosyncratic details that differentiate one particular writer from another even when utilizing a different script. Though we learn our handwriting from exemplars, every one of us will adopt distinctive quirks of our own. Through continuous practice of our handwriting, we develop an internalized model hand, i.e. the ideal execution of our unique handwriting. The programming language of our internalized model hand is embedded into our kinesthetic memory as units of movement that constitute the basic building blocks of letter formation, i.e. the internalized model hand can be thought of as a modular repository of distinctive twists and turns of the hand, which remain constant in their output of strokes. The end result, the actual lines on the page that form the letters, words and sentences, however, comes about from the interplay between our internalized model hand and the particulars of that specific writing situation ...

From these considerations forensic document examiners have established that some aspects of our handwriting are always produced with more conscious effort than other aspects and that the less conscious aspects are the ones more difficult to disguise; hence, they are also the best place to discover idiographic features of the author. Furthermore, writers attempting to disguise their handwriting or simulate someone else’s will eventually switch (i.e. fall back, or lapse) into their unconscious habits and make use of those units of movement that form their unique internalized model hand. The fundamental difference between writing with one’s own handwriting and imitating someone else’s is that the former uses proprioceptive feedback (i.e. internal feedback that allows the body to keep track of the relative positions of its parts) while the latter is essentially drawing, and relies heavily on visual feedback to maintain its desired form. The continuous attention to form involved in disguise or imitation produces mental fatigue, which results in unconscious lapses into one’s personal idiosyncrasies, especially when the disguised or simulated handwriting is of considerable length. These lapses tend to become more frequent towards the end of the document (or the later stages in any uninterrupted period of imitating). These facts about how a forger’s idiographic traits enter into a forgery lead to one other observation: that a document, especially one that is of considerable length and written in a difficult script, is probably not simulated or disguised if the writing is consistent from start to finish.

In general, we could question:

(a) the analysis of the document and/or the competency of the examiner
(b) the validity of methods that could be used to detect a disguised hand

Most criticism has implicitly focused on (b), the validity of methods that could be used to detect a disguised hand, usually by presentation of the facile claim that, whatever was described in the analysis, it could have been produced by a sufficiently skilled imitator since that is their purpose. This is perhaps understandable because this subject is approached in an interdisciplinary way by the interested scholars. Most commenters offering criticism are well outside of their own disciplines.

[Peter Kirby]

Commenting outside of his own discipline, and writing after Anastasopoulou published, Tselikas offered criticism in category (b), stating that “the scribe of the letter would not use the own personal style” and implicitly rejected the validity of the methods used in handwriting examination when considering whether writing is in a disguised hand:

As for Tselikas’s observation; while we agree that a forger would certainly try to avoid his own personal style while forging, the unconscious and the deeply engrained aspects of handwriting are rarely fully suppressed.

This is not particular to Tselikas but characteristic of most responses that have been made. They have tended to reject the idea of such a method itself -- a sort of casual, amateur, and yet readily adopted "forgery finds a way" philosophy of handwriting examination -- and thus haven't needed to be overly concerned with the details beyond that.

[Peter Kirby]

Something not done here by Tselikas -- such as providing comparison with other writing with respect to the amount of discontinuity in the writing -- can also be done at the descriptive level, in a verifiable way.

This has been done by Timo S. Paananen and Roger Viklund, who (like most of us) also are not forensic document examiners.

Specifically looking at the kind of feature mentioned here by Tselikas, they have compared it to samples of other writing.

https://tuhat.helsinki.fi/ws/files/1252 ... _check.pdf

How many times does the scribe of Clement’s Letter to Theodore make a pen lift? How many ligatures and abbreviations are there in the text? How many of the glyphs (i.e. individual, distinct elements of writing such as a letter or a ligature) are written without the pen being lifted in-between; i.e. how many clusters of two, three, four and more glyphs are there? All of these questions are easily (albeit laboriously) answered, and their results are suggestive of how controlled the handwriting in Clement’s Letter to Theodore is. Moreover, to provide a necessary context for the numbers, a simple comparison with other eighteenth-century manuscripts will suffice. And since the question of forgery is relentlessly upon us, we need only to recall that the process of forgery is essentially one of conscious handwriting control in which internal and external constraints affect every movement of the pen.Hence, as Sirat informs us, “One of the best ways to recognize a forgery is to look at the hints of control, such as an abundance of fresh starts.”

For comparison with the handwriting in Clement’s Letter to Theodore we studied two eighteenthcentury manuscripts—a copy of the writings of Gregory of Nazianzus (Add MS 8240) and a personal letter of Konstantinos Dapontes (Add MS 8237)—both of which are quite similar to the Clementine letter in their execution of the Greek script. We took a random sampling of 1000 letters or glyphs (excluding abbreviations and most ligatures), beginning from line fourteen of the first page of all three manuscripts. The results are presented in the table below.

	Theodore MS 65	Dapontes MS 8237	Gregory MS 8240
Total number of letters or glyphs	1000	1000	1000
Individual glyphs	352	460	520
Glyphs in groups of two	382	422	286
Glyphs in groups of three	216	106	153
Glyphs in groups of four or more	50	12	41
Ligatures	107	104	51
Nom. Sac. and other abbreviations	3	3	1

While all three manuscripts have many ligatures and other markings of cursive writing in common—e.g. the circular ligature combining omicron and upsilon, the stigma for marking the combination of sigma and tau, and the word καί written more often than not in one continuous stroke—it is notable that Clement’s Letter to Theodore and Dapontes use them twice as often as the copy of Gregory does, while their use of nomina sacra and other abbreviated words is thrice as frequent. The signs of control are most evident in the copy of Gregory, in which more than half of the glyphs are written individually, compared to only 352 individual glyphs (c. 35 %) in Clement’s Letter to Theodore. A similar trend can be seen in the other clusters of glyphs as well. Dapontes’s personal letter stands in the middle in this regard, with slightly less than half of the glyphs written individually, but with the largest number of two-glyph clusters. Of these three eighteenth-century manuscripts, Clement’s Letter to Theodore exhibits the least signs of control, judged solely by the amount of pen lifts.

While none of the manuscripts could be characterized by Sirat’s “abundance of fresh starts,” the scribe who copied the text of Gregory seems the most interested in preserving the readability of the writing. Compared to Clement’s Letter to Theodore, it is generally easy to distinguish between his letters epsilon and cursive pi, for instance, and the letterforms are less simplified, which implies lesser velocity in executing the strokes. These numbers will hopefully render the expert opinion of Anastasopoulou more intuitive. Her observation that the letters in Clement’s Letter to Theodore are written “unconsciously,” to take but one example, corresponds well with the number of pen lifts we have counted; i.e. the handwriting in the Clementine letter leans towards personal writing, which is the realm of our own internalized model hand, unconsciously and automatically executed—an observation we base on the relative numbers of pen lifts, ligatures and abbreviations, and on the generally poor readability of the script in Clement’s Letter to Theodore compared to other examples of eighteenth-century handwriting. The implications of our study are clear enough: as the general quality of appearing unconscious and inconspicuous is difficult for forgers to imitate, it follows that the more characteristics of personal writing a script contains, the less likely it is to be a forgery. In the amount of signs of control Clement’s Letter to Theodore exhibits, it is indistinguishable from genuine eighteenth-century manuscripts.

[Peter Kirby]

Anastasopoulou commented:

https://www.biblicalarchaeology.org/dai ... a-forgery/

When a large document is consistent, we have a first indication of genuineness and this applies to the Secret Mark letter.

The Secret Mark letter, as written in detail in my analysis report, is written in a natural and spontaneous way and in my opinion, does not have such indications so to make us think of a suspicious writing.

With reference to the handwriting examination report:

https://www.biblicalarchaeology.org/wp- ... alysis.pdf

Scott Brown has contextualized some of the observations made by Anastasopoulou in the context of the question of whether it is a later, laboriously produced (but not natural) imitation of ca. 18th century handwriting.

https://www.biblicalarchaeology.org/wp- ... -brown.pdf

Although I am not trained in QDE myself but only an interested participant in the debate, I believe that Anastasopoulou’s two reports provide us with the information we need in order to assess the probability that Mar Saba 65 could be a forgery of an eighteenth-century Greek monastic hand. That is because many of the characteristics of Mar Saba 65 that she identifies in her first report are precisely the qualities that experts in QDE consider most valuable in distinguishing between authentic and forged writing. My intention therefore is to assess the plausibility of the modern forgery theory by surveying what the experts have to say about these qualities.

These comments are helpful in moving beyond the mere idea that, in an imitation, all things are possible, i.e., that this kind of study is not effective at a fundamental level because an imitator's explicit goal is to avoid detection of their artifice.

But it is artifice itself, the use of handwriting technique that does not come naturally, that can be revealed by handwriting examination of a lengthy composition that requires skillful execution.

[Peter Kirby]

As Brown notes, with reference to Anastasopoulou's report on the subject, a document like the Mar Saba fragment is particularly susceptible to a successful demonstration of its artificiality, if that were true about its handwriting:

Forging a document is not a simple matter of picking up a pen and imitating someone else’s handwriting. In the judgment of Albert S. Osborn, “the successful forgery of a whole document is a task of extraordinary difficulty and requires intelligent attention to many particulars and details that do not enter into the task of fabricating only a signature.”

The degree of difficulty varies with the skill and complexity of the hand being imitated. Unskilled handwriting (like that of children or the semiliterate) is the least challenging writing to forge, because it is produced by slow and uncertain movements, which are similar to the drawing movements of forgery. By contrast, skilled writing is very difficult to imitate, because it contains features that cannot be replicated except through equally quick and skilful pen movements. Osborn put it this way: “That writing is imitated with the greatest difficulty which is strong, smooth, free and skilful and that cannot be correctly reproduced by a slow, copying movement.” The four qualities that Osborn highlights in this statement are characteristics that Anastasopoulou ascribes to Mar Saba 65. Thus we are dealing with a whole document in a kind of handwriting that is especially difficult to imitate ...

Katherine Koppenhaver states the problem boldly: “It is impossible to maintain a successful simulation of a lengthy document. The longer the text, the more the writing reverts to the natural characteristics of the writer as his concentration wanes.” Anastasopoulou’s supplementary report describes the style of Mar Saba 65 to be consistent: “When a large document is [internally] consistent, we have a first indication of genuineness and this applies to the Secret Mark letter.”

[Peter Kirby]

Recall that:

The fundamental difference between writing with one’s own handwriting and imitating someone else’s is that the former uses proprioceptive feedback (i.e. internal feedback that allows the body to keep track of the relative positions of its parts) while the latter is essentially drawing, and relies heavily on visual feedback to maintain its desired form.

When imitating handwriting that requires a high level of skill to execute, the same kind of skill isn't even drawn upon. Instead of acquiring the same set of handwriting skills that someone had through repetition, the imitator would activate a different set of skills in the attempt to produce something similar in appearance. The effort itself affects smoothness and line quality.

As Brown points out:

The features that Anastasopoulou underscores (through repetition) as characteristic of Mar Saba 65 are artful and unconscious qualities that result from quick, habituated pen movements. Through years of repetition, a person’s natural handwriting will possess those qualities, but the same person’s imitation of another’s handwriting will not. Instead, imitations of skilful handwriting lack smoothness and certainty of movement; they display a quality that document examiners refer to as tremor and bad line quality. Koppenhaver and Tom Davis offer helpful explanations:

Tremor is the most easily detected and the most frequently found sign of forgery. Tremor results from slow writing or drawing. Writing is a little like riding a bicycle. When the bike rider slows down, the wheels will wobble. If the rider leaves a path in the dirt, it will resemble a wiggly line. The same is true of the writing line. When the writer slows down to copy or trace a line, the line will waiver.

Line quality refers to the degree of smoothness of the pen line. Good line quality is smooth, confident, with regular uninterrupted curves. Bad line quality is irregular; the curves are interrupted, tending to degenerate into a series of straight lines, showing pausing, or even penlifts, where the pen is taken completely off the paper. It is a matter of degree of skill, and speed. If our expertise in writing is inhibited in some way, the pen will move slowly, and the line quality will degenerate accordingly. In the act of forgery, since we are necessarily not as expert at producing the graphic shapes of someone else’s writing as the writer will be, then the pen will move more slowly and the line quality of our imitation will be poorer than that of the original.

These explanations bring together three of the qualities that Anastasopoulou identifies in Mar Saba 65—smoothness, continuousness, and confidence—under the rubric of good line quality, attributing them to skill and speed and opposing them to the overall bad line quality that characterizes freehand imitation, the shakiness and uncertainty of direction sometimes called forger’s tremor.

It is important to realize that forger’s tremor is not something that occurs sporadically within writing that generally looks normal. As Anastasopoulou stresses in her supplementary report, “tremor of fraud shows a painstaking and unnatural care at every point that indicates an effort to follow an unfamiliar copy.” Forger’s tremor is the slow, hesitating pen movement involved in simulating another person’s writing, so one cannot talk about forger’s tremor unless the writing generally exhibits poor line quality and uncertain movement.

The conclusion drawn here is that the handwriting isn't an imitation but rather someone's own natural handwriting:

Anastasopoulou was clearly interested in whether the questioned document displayed the degeneration in line quality that results from imitation and concluded that it does not. Rather, based on the overall fluency of this complex hand she deduced that it was written with speed and skill by “a hand used to writing in this manner.”

[Peter Kirby]

Another aspect of handwriting examination looks for natural variation within a writer's individual master pattern:

When engaged in repetitive activities such as writing, humans lack the precision of machines; they therefore do not write words and individual letters exactly the same way each time. Writers do develop one (sometimes more than one) typical shape for each letter, which examiners call its master pattern, but individual renditions of this pattern vary, with some letters (usually the larger or more complicated) varying more than others. The overall amount of natural variation also varies among individuals. Because this variation occurs naturally, the presence of unnatural consistencies in writing implies a conscious attempt either to write homogenously (e.g., to practice calligraphy or formal penmanship in a particular script) or to replicate whole words used as models.

The presence of natural variation has important implications in the detection of forgery: The forger doesn’t recognize the need for natural variation and makes the words and letters as close to the known writing as possible. This is obvious when he or she attempts to copy extended writing. The forger will exactly duplicate the letter form, crossing the “t” at the same angle and in the same place, dotting the i in the same location, forming the design of the letter exactly like the model, giving the writing a rubber stamp look.

Forgers normally try to make their freehand imitations of another person’s handwriting as close as possible in form to the letters, syllables, and whole words that they use as models, supposing that the success of their deception depends on their ability to exactly replicate what they see. When their sources contain few examples of the words or letter combinations required for the forgery, the resulting imitation contains less variation and more exact duplication than does the authentic writing.

Brown writes:

Anastasopoulou also uses the terms “freedom” and “spontaneity” to describe the questioned handwriting, claiming that it is written “automatically” or “unconsciously.” These ideas are related to natural variation. When the variation of letterforms within a handwriting is noticeable yet the words remain legible, the hand is said to exhibit freedom. When the variation increases to the point of illegibility, the hand is said to exhibit carelessness. Freedom and carelessness result from the normal inattention to the writing process that occurs when a person is paying more attention to what she or he is writing than to the act of writing itself. Hence document examiners infer that visibly free and fluid handwriting was also spontaneous, unconscious, or automatic, “the result of a habit and not the conscious following of a copy.”

Out of misunderstanding but picking up on words such as "following of a copy," some might then suggest that an imitation could be produced successfully in this respect if it didn't have to follow any particular handwriting examples. The misunderstanding is based on the idea that the problem is the attempt to follow an exemplar. Freed from that constraint, a suggestion might be made that an "imitation" could be produced with the same level of natural variation used in someone's own handwriting. The idea here betrays the mistaken assumption that an imitation would be produced with the same kind of skill that would produce someone's own handwriting.

What is in fact is being tested is this very thing: is this writing a demonstration of the skill learned from the natural production of one's own handwriting, its own habits, and the way one has learned to write, without the traits shown when artificially attempting to modify it? In short, is it simply "the result of a habit," or is it a product of artifice?

After a review of sufficient evidence and the application of relevant skill in handwriting examination, Anastasopoulou concluded that the Mar Saba fragment is "the result of a habit" of someone's own natural handwriting.