Chapter 3
Signature Examination

3.1 Introduction

Signatures are a special kind of handwriting for a number of very important reasons. A signature is a way for a person to endorse the content of a document and this therefore goes beyond the creating of the content itself and means that a person is agreeing with that content. The most obvious example would be a cheque where the date, payee and amounts can be written by a third party, but it is the signature that is the component that effectively authorises the transfer of funds from the account. It is for this reason that the signature on a document is of particular relevance in many forensic cases.

Signatures can come in a wide variety of forms ranging from simple to complex, from legible to stylised (with few or no recognisable letterforms) and from full name to very shortened. A person's signature is, for some writers, their most frequently produced writing and is perhaps their most practised and automatic writing. As a result, signatures are often fluent even if other writing is less so, to the point where those who would be regarded as unable to write can at least sign their name.

Because of the significance attached to signatures, they are the primary target of simulation (copying by forgery) – far more so than normal handwriting. For this reason, as we shall see in Section 3.6 of this chapter when we consider how to interpret findings, the handwriting expert will have several alternatives to consider including the possibility that it is a genuine signature, a simulated copy of a genuine signature or a non-simulated version.

The production of signatures is physiologically similar to handwriting production as described in Chapter 2. The fluency and speed with which most people are able to execute their signature suggests that the motor plan is highly learned and automated. Since some writers have no recognisable letterforms in their signature, and given that there is no language input as such (other than perhaps spelling), the cognitive paths that initiate the movements for signatures are likely to be different in some respects to those associated with normal text.

Signatures show natural variation just as handwriting does. For many people a signature will be their most frequently executed written product and so might be expected to show strong similarity from one occasion to another. However, this is not necessarily the case as people vary in the consistency that their signatures show from one occasion to another. No two pieces of handwriting or signatures will ever be identical, but might it be possible that two signatures are so similar that they could be misidentified as one having been copied from the other?

The copying of signatures, whether it be freehand or by a tracing process, is itself imperfect, depending to some extent on the skill of the person doing the copying and the materials available. The penmanship of forgers has been examined (Dewhurst et al., 2008) and shown to be a factor, with skilled writers (calligraphers) better able to produce simulated signatures that cause problems for document examiners.

Given that signatures may not contain any recognisable letters, how is it possible to examine them at all? The reason is that signature comparison (indeed any handwriting comparison) is in reality a sophisticated pattern comparison process of a product of the human mind. The key point is that the comparison must be made on a like-with-like basis. In normal handwriting, the letters have to be identifiable so as to ensure that a like-with-like comparison is being made – the fact that the letter is an A or B or C is essentially immaterial. Hence, it is possible to compare signatures that are unreadable or indeed in a foreign script providing it is possible to be sure that they are purporting to be the same thing.

As a result, a master signature will not be identical to the copied signature derived from it. The degree of difference between the copy signature and a coincidentally similar genuine signature therefore becomes a theoretical possibility. The degree of similarity of just the height and length of signatures from the same writer was assessed by Evett and Totty, and the results provide evidence for expecting discernible variation to occur between genuine signatures (Evett & Totty, 1985). Indeed the authors found that over time the dimensions of signatures could vary, reinforcing the need to seek contemporary specimens whenever possible. In addition, traced signatures are by their very nature drawn rather than fluently written and hence the fluency is likely to be a key feature in discriminating between two coincidentally similar genuine signatures or a genuine signature and a tracing from it. The experienced document examiner should be able to distinguish between the various possible scenarios, and if there is uncertainty this will be reflected in the confidence with which a conclusion is expressed.

The expertise of forensic examiners is probably most tested by signature examinations due to the small amounts of writing involved and the willingness of fraudsters to try to perfect a simulation for whatever gain they have in mind. The capability of experts to reach a correct opinion is therefore nowhere more sharply focused than in signature examinations, where a correct opinion is that which is not only factually correct but also has the correct strength of opinion in terms of the available evidence (including an inconclusive opinion where the evidence does not support any view reliably).

3.2 The development of signatures

The factors that writers incorporate into their signature and how the decisions are made as to its form have received virtually no academic research. For many people the initial impetus to develop a signature will arise in their late teens or early 20s as they need to make decisions and transactions away from their parents and guardians, perhaps in education or in various dealings relating to property or finance. It is likely that most people will experiment with alternative forms of signature and settle on one that they like, because it is easy to write or pleasing aesthetically or for some other reason. However, it is also likely that the initial form may undergo some changes over time, perhaps major revision to begin with, but eventually settling into a form that becomes the highly automated signature that will remain for much of adult life. Signing one's name is something that some people will do much more than others, often depending on one's job. Frequent signing may lead to a person simplifying and/or shortening their signature to facilitate speed of execution. This may affect at least obvious properties, such as the size of the signature which can vary over time.

One obvious cause of a signature change is marriage, if one or other person takes the surname of their spouse. This requires the re-invention of a signature in a process that is presumably not unlike that when first a signature is created in adolescence (unless, by some unlikely event, there is no change due to coincidentally having the same surname).

One special example of the signature is the autograph used by celebrities. It is again a matter of the individual devising an autograph that suits their purpose and there may not necessarily be an obvious connection between an autograph and course-of-business signature of a celebrity. Because such signatures have a commercial value, there is a market for the buying and selling of autographs, and for this reason simulated autographs may be encountered by the forensic document examiner.

For the most part signatures, once their form has been finalised to the writer's satisfaction, will remain reasonably consistent over long periods of time. Indeed, the ability to write a signature is particularly resistant to the effects of age and incapacity that may occur later in life. This may be for a variety of reasons, including the automated nature of the process and also the personal significance that people attach to being able to sign their name.

Thus, the dynamics of signature development are similar in general terms to handwriting, except that the ‘learning’ phase occurs much later, typically in the teens, when normal handwriting has usually been mastered as a skill.

Signature writing is for many people a highly learned and automated skill. It was noted in Chapter 2 that such highly learned skills are difficult to copy by others attempting to adopt the relevant pen movements at the appropriate speed. Conversely, altering a signature to disguise it is difficult as the automatic movements are hard to suppress without losing natural fluency and appearance.

While a developed signature will be used by someone for many years, with often relatively little change, there are circumstances in which the signature can be affected, producing changes that require the document examiner's expertise to interpret what has caused the differences.

3.2.1 External influences: alcohol, infirmity and old age

In Chapter 2 a number of factors that can affect handwriting in general were discussed and these, of course, equally apply to signatures. For example, alcohol is likely to be a frequent external influence on handwriting and signature production. These effects are another reason why computer recognition of signatures is a difficult area. However, if it is known that the true signatory is drunk then automated systems can detect changes in the signature, such as lighter pen pressure and faster writing speed, and infer the possibility that the changes are due to intoxication (Shin & Okuyama, 2014).

Many of the factors that influence handwriting production in the infirm and elderly will also of course have the potential to affect signature execution. Because of the social and personal significance of the signature, even for those who find handwriting difficult, the production of a signature remains a matter of personal pride and it is common to see elderly people retouching their signature to make it ‘look right’. The signature is nonetheless often the most highly practised handwriting movement and it may be particularly resistant by virtue of its automatic production to showing the effects that appear in everyday handwriting.

In general, the factors that affect handwriting and are discussed in detail in Chapter 2 will also have the potential to affect the appearance of the signature. However, it is possible that the interaction between the cognitive and motor aspects of normal handwriting are changed somewhat due to the minimal amount of thought that has to go into writing something as familiar as one's signature.

3.2.2 Guided hand signatures

If a person is incapacitated, for example due to ill health, then it may be allowable for them to be assisted by another person should they need to sign their name. It may be appropriate to query both their mental capacity to understand what they are signing and their physical capacity to execute the signature. The mental capacity to understand what they are doing is clearly outside the handwriting expert's remit. However, the effects on a signature that guidance from another person may lead to are legitimate areas for assessment by the handwriting expert.

Guided hand signature cases are rare. One of the key aspects to guided hand signatures is the extent of the assistance provided and the capacity of the true signatory to write their own signature (Sellers, 1962). In general, the greater the capacity of the true signatory to write, the less assistance that will need to be given (often just requiring support of the writing arm, for example, with no movement contributed by the assistant) and therefore the resulting guided hand signature will tend to look like a normal genuine signature. However, if the true signatory is severely incapacitated then the assistance will necessarily need to be greater and the true signatory may contribute little towards the movement of the pen, so that the formation of the signature will correspondingly be more influenced by the person doing the guiding. This will tend to produce a signature that has much less similarity to the true signatory's signature. If the input is fairly even, then the resulting signature may take on a mixture of the true signatory's handwriting features and those of the guide, an outcome that is extremely difficult to predict in advance. Guided hand signatures therefore display some often unexpectedly marked departures from the specimens due to the ‘one off’ interaction between the guider and the guided against a background of an otherwise fairly ‘normal’ looking signature. This tends to contrast with simulated signatures in which the forger attempts to minimise any differences with the genuine signature.

3.2.3 Signatures in blind people

Handwriting for blind or visually impaired people is extremely difficult, but fortunately the development of computer technology has enabled written communication. However, signing one's name is still a desirable skill to acquire and various devices have been created to assist with this, using specially modified pens that are held in a stand so that they maintain contact with the paper, and with tactile, kinaesthetic (sense of force on the body) and audio feedback in the learning process replacing visual feedback (Plimmer et al., 2011). The social acceptability of normal signatures may also be reflected on documents such as job applications, in which the inability to sign ‘conventionally’ may bias a potential employer's view of a candidate's capability.

3.3 Simulating signatures

There are a number of methods of simulating signatures. By far the most commonly encountered in casework is a freehand simulation, whereby the forger attempts to reproduce, without any aids, the signature of another person usually with reference to at least one sample of that person's signature. In many instances it is possible for the sample signature to be visible at the time that the simulation is written. Of course, this may not be possible in all circumstances, such as when signing in front of a witness in a bank or lawyer's office; in these situation forgers have to rely on their memory to recall what the signature looks like.

Copying the signature of another person is not easy as the forger will have their own handwriting features that will need to be suppressed, and at the same time the handwriting movements to produce the simulated signature have to be adopted while maintaining appropriate fluency throughout.

The ease with which a target signature can be copied will inevitably be determined in part by the complexity of that signature. A very simple signature is likely to be more accurately copied than a complex one. The complexity of the target signature is determined by a variety of factors such as its length, the overlapping pen-lines, the unusual formations of letters, the absence of recognisable letterforms and the presence of unusual shapes of curves and lines in their place and so on.

Whatever the nature of the target signature, the handwriting skill of the forger is another important aspect of the outcome of the simulation process. A person that cannot write skilfully will not be able to produce a fluent copy of a skilful writer's signature. It is not surprising that skilful forgers are often skilful penmen that also have a good eye for detail (Dewhurst et al., 2008).

Simulated signatures need to reproduce all elements of the target signature as closely as possible from the form to its proportions and its fluency (Figure 3.1). Simulations tend to fail more for reasons of lack of fluency, reflecting the forger's desire to make the signature appear as pictorially similar as possible. This has been confirmed as being true for other handwriting systems (for example, see Al-Musa Alkahtani and Platt (2010) for Arabic signatures).

c02f001

Figure 3.1 The upper signature has a smooth pen line with variation in pen pressure. The lower signature is a freehand simulation showing a shaky pen line and a ‘drawn’ appearance.

Tracing signatures by various means are occasionally encountered in casework. A slight indentation in the paper surface is used as a guideline (for example, using tracing paper as an intermediate) and can be detected by close examination revealing the indented guideline running alongside the inked signature at the points where the inking in does not follow exactly the guideline (Figure 3.2).

c02f001

Figure 3.2 The tracing lines are close to, but not exactly overwritten by, the ink line.

A genuine signature can be scanned and the signature printed out onto a document. Microscopy will show that a computer printer rather than a pen has been used. If the computer-printed signature is then inked in to mask this, the result is often very clumsy (Figure 3.3).

c02f001

Figure 3.3 A scanned and printed signature overwritten in ink.

A so-called window tracing can be effected by placing the document that is to bear the simulation on top of a document bearing a genuine signature and illuminating them together from behind. The forger then produces the simulation by following the genuine signature that shows through. The simulation produced is usually of poor fluency and may well be incorrectly constructed since this process may only produce a pictorially similar simulation.

Any simulation based on a tracing or image-copying process can potentially be linked to the genuine (master) signature from which it is derived by overlaying the two signatures and finding them to be so similar that they cannot have been independently produced. (See Evett and Totty (1985) for a discussion of how similar two genuine signatures can be.)

One type of signature that is encountered in some parts of the world is the seal. This typically shows a signature that has been etched onto a wooden, plastic or metal substrate (see Box 3.1)

3.4 Computer-based recognition of signatures

Identifying a person has many important functions in society. There are many ways in which a person's identity can be confirmed, ranging from those who know the person to other means such as documentation (a passport for example). In this context, identification is the process of determining who someone is (comparing one from many, for example by examining photographs of a set of people) whereas verification confirms a person's identity (comparing one against one, for example comparing someone with the passport photograph) and such comparisons often involve the use of biometrics. Biometrics are, as their name suggests, ways of differentiating between people using various measures. These measures can be of two types. Physiological measures, such as fingerprints or iris eye patterns, measure aspects of the body as a physical object. Behavioural biometrics measure features caused by the body moving, such as how one walks (gait pattern), speech patterns or signature writing. As people will vary in almost any behavioural activity, biometrics that can capture this variability have the potential to assist in identification with all the ramifications that this has in a modern society.

The use of signatures as biometrics has been, and continues to be, extensively researched. Automatic signature analysis has been used in the banking industry and its use in other aspects of everyday security is under constant review. The frequency with which signatures are used in all kinds of social and financial transactions serves only to emphasise their importance. Crucially, it is almost always the case that the person who may be ‘accepting’ the signature (such as a shop keeper) is not trained sufficiently to be able to reliably assess a signature's authenticity. For this reason, a considerable amount of effort has gone into using computers to assess whether a signature is genuine or not.

There are two types of feature that machines can analyse. Some features can be derived from the static image of the completed signature, such as its size and the geometrical conformations of various elements such as loop shapes. The second kind are dynamic features such as the speed of pen movement and pen pressure. Dynamic features can only be measured accurately using the appropriate technology, namely a digitising tablet (see Box 2.2 in Chapter 2). Technological improvements to these devices have been mirrored to some extent by the many different ways in which the information attained is processed using various algorithms that extract from the data key comparison parameters. For example, the simple maximum dimensions could be measured or the velocity of pen movement could be measured or the number of acceleration and decelerations could be counted. Deciding which parameters to use and how to incorporate them into an algorithm that minimises errors is itself a key factor in the success of these systems (Wilkin & Ooi Shih Yin, 2011).

Such automated systems of signature comparison can produce two types of error. Type I errors are false rejections (where a signature is assessed as being not genuine when in fact it is genuine) and type II errors are false acceptances (where a signature is deemed to be genuine when in fact it is not genuine). Verification devices (note that to verify a person's signature requires that they have previously recorded their signature one or more times for comparison purposes) will show both error types and such errors are more or less likely to occur depending upon the nature of the signatures being studied. For example, if specimen (previously recorded) and questioned (the signature being verified) signatures are all naturally written then the error rates are lower, but if a questioned signature is disguised it may be wrongly deemed to be not genuine when in fact it is (type I error) or a skilful simulation may be deemed to be genuine when it is not (type II error). Genuine signatures that are fairly simple in formation and consistent may lead to many correct verifications with few false rejections. Similarly, if forged signatures are very crude these may be readily rejected with few false acceptances. However, this assumes a fairly naïve level of expectation from real-world situations where some people naturally have very variable signatures and some forgers do make considerable efforts to perfect simulated signatures. Such situations are a much greater challenge to automated systems and even a relatively low level of improvement and motivation by the forger can cause false acceptance of forgeries as genuine signatures (Ballard et al., 2007).

Signature variability must be accounted for in automated systems of signature verification, but variability itself varies from person to person. In addition, increased variability is often found in the signatures of older people due to general loss of motor control together with any relevant medical conditions. Thus age and health are particular factors that such systems may need to consider (Guest, 2006), factors that the forensic expert is trained to routinely consider when relevant.

Not only can such devices measure what is happening during the writing process, they are also capable of measuring the movement and time that the pen is not writing, in other words when the pen is in the air. It has been shown that the writer's habits also apply to the non-writing component as the pen moves from one part of the writing to the next. This will have particular value in text production where there are more such non-writing movements, but if a signature has breaks in it these too may be a characteristic (Sesa-Nogueras et al., 2012).

Both the dynamic and static dimensions of signature production are potentially legitimate sources of (imperfect) information about a person's identity. The amount of information available from the static image is significantly less and this inevitably leads to a much larger error rate in assessing signature authenticity, whereas the dynamic information leads to a smaller error rate (Kovari & Charaf, 2013).

However, global error rates do not show the level of difficulty in any particular comparison. Signature complexity, the ability of writers to disguise their own signatures or copy the signatures of others may all contribute to greater error rates in verification systems. Computers cannot readily account for the human dimensions of cunning and skill, which is where the forensic document examiner's experience of casework is of greatest value. Nonetheless, the evaluation of output from an automated (biometric) system with associated false rejection and false acceptance rates can be mapped onto a Bayesian approach (see Section 2.9.2 of Chapter 2) of evidence assessment (based on alternative hypotheses) with the use of mathematical manipulation (Gonzalez-Rodriguez et al., 2005).

Given the desire to minimise errors, one possibility is to use more than one biometric approach. For example, it is possible to record and measure pen grip, which could give yet more information about the writer. The way that a person holds the writing implement itself may be fairly characteristic (although very unlikely to be unique). There are a number of general ways of holding a pen (Schneck & Henderson, 1990). Measuring the relative amounts of contact between the writer and the writing implement (using a pen specially adapted and fitted with pressure sensors) adds another layer of information that can be derived from the signing process (Ghali et al., 2013). This grip pattern will be a reflection of how much pressure the fingers and hand place at various points around and along the barrel of the implement, a pattern that will show some variability depending on the physical properties of the implement, such as its diameter and shape.

A second possibility would be to use a signature together with speech recognition. Research has suggested that the signature and spoken name together provide more information than either on their own (Humm et al., 2009). To make an effective forgery would thus require imitation of the voice as well as the handwriting by the impostor. The advantage of speech verification is that it can occur at essentially the same timescale as the signing process.

The advanced devices used in handwriting research should not be confused with the devices used by delivery services, for example, where receipt of goods is endorsed by signing at the front door. Such devices are not intended to be used in the research laboratory but are a convenient way of recording signatures, albeit the forensic information available from such devices is limited.

In summary, the investigation into the mechanics of handwriting and the use of signatures as a biometric have reinforced the considerable scope that handwriting and signatures have for distinguishing between people or, alternatively, confirming a person's identity. However, the hardware and software (algorithms) used have yet to replace the forensic document examiner, whose experience at interpreting the evidence has not been improved upon (see Box 3.2).

3.5 The forensic examination of signatures

The basis of the forensic comparison process is essentially the same as that described for handwriting comparisons. Generally speaking, obtaining specimen signatures is not a problem as few people are unable to produce some kind of signature. Even those who have not learned to write, for whatever reason, can generally write their signature, although such signatures are often relatively simple in their formation and may not be very fluently written.

The fact that signatures can be modified over time due to a variety of reasons makes it important that specimens are as contemporary as possible to the disputed signature. The number of specimen signatures needed varies from case to case but typically about 12 is a good starting point from documents signed in the course of everyday life. Some people do not have cause to sign their names very often and it may be necessary to ask for specimen signatures to be supplied at request. The limitation of request specimen signatures is similar to that for normal handwriting, because requested signatures can only provide a snapshot of the range of variation to be found in a person's signature. With request specimens, the possibility of deliberate disguise always has to be considered, and this is even more likely to happen with signatures that can be changed relatively easily. However, disguise is not always very subtle and is often readily apparent due to loss of fluency, loss of consistency and, if available, a lack of similarity to any non-request signatures. Disguise of signatures, while relatively simple to do at the gross level, is much more difficult to do at a subtle level since the signing process is so automated that making small changes is more difficult than a complete transformation. When asked to make relatively subtle adjustments to their signature with the intention of later denying authorship, most people find it very difficult to achieve because of the automatic nature with which signatures are usually written. Any interruption caused by the intended changes lead to disruption of the motor pattern which is reflected in poor execution.

Making a comparison between the questioned signatures and specimens is very similar to that described for handwriting in the previous chapter. All relevant observations must be recorded, such as the structure, the variation, the proportions and the fluency, and these are compared between the questioned and specimen signatures. Because signing is just as prone to natural variation as handwriting is, there will always be some features that can be matched and others that cannot. It is the task of the expert to interpret the meaning of the findings and reach a safe and justified conclusion based on interpretation of all of the observations.

3.6 Interpreting findings in signature cases

Most of the features that the handwriting expert examines in signature cases are static features of shape, proportion and slope. Measurements of pen speed are not possible from the static image of the signature, but some dynamic aspects of signature production can be inferred. Just as with handwriting examinations, the speed of pen movement and the pressure applied to the writing implement can be inferred from the pen line. Pen pressure and pen speed are important diagnostic features in all signature cases since a forger may not write at the same speed or with the same pressure as the true signatory. Likewise, a person attempting to alter their signature deliberately may change the speed or pressure as they concentrate on making the changes.

The experience of the expert is crucial in assessing which explanation best fits the observations in a particular signature case. As with handwriting examinations, alternative explanations must be considered. A detailed Bayesian approach to a signature case is reported by Biedermann et al. (2012).

The various explanations for findings in relation to the authorship of a questioned signature are as follows, although not all possibilities may be relevant in all cases (Figure 3.4).

  1. A genuine signature written in normal circumstances – typically while seated at a table. Normal, genuine signatures usually lie within the range of variation to be found in the specimen signatures and will show a similar fluency when compared to the specimens.
  2. A genuine signature written at a different time to the specimens with differences attributable to changes in the signature over time. If there is a known significant time gap between the specimen and questioned signatures, especially when it involves young, typically teenage, or older people, then due caution must be used to take account of the possibility of experimental and developmental changes in the young and loss of pen control in the elderly.
  3. A genuine signature written in unusual circumstances but with no intention to alter it (for example, written while drunk, or using an unusual writing implement or standing and supporting paper on an infirm surface). If it is known that a signature was written in unusual circumstances, then it may be possible to use this information to anticipate what effect it might have on a signature. For example, as noted in Section 3.2.1, alcohol tends to lead to a relaxation of movement and signatures reflect this by being larger and tending towards a more scribbled appearance, but the fundamental structure usually is kept.
  4. A genuine but disguised signature that has been intentionally changed or a simulation of a genuine signature. In casework, this distinction is often the most crucial. Deliberate disguise where the writer intends to deny their signature at a later time (for instance due to not having sufficient funds to cover a financial commitment) can vary in extent from just a minor alteration, such as using a middle initial when it is usually absent, writing a first name in full rather than using initials or adding an underline, to changing some letterforms. Changes can be to individual letters or can be more global, such as a change of slope. For most writers for whom writing a signature is automatic, such changes lead to significant disruption to the fluency and most people can only modify a small number of features of their signature – changing the appearance of each component of the signature is extremely difficult to achieve in most cases. The nature of the disguise is also an important factor; counter-intuitively, the more subtle changes are more difficult to achieve than the gross, obvious changes since the former require more careful thought whereas the latter can involve a single major change such as writing in capital letters. The need for subtle change is sometimes referred to as autoforgery to distinguish it from very obvious differences where the resemblance to a usual signature is minimal. The need for subtle change is greater if the person accepting the signature has any basis for a comparison with a genuine signature (such as on a cheque guarantee card) such that a gross deviation would be obvious and call the authenticity of the signature into immediate doubt and non-acceptance.

    If the questioned signature has been disguised but the specimens are natural, then typically there is at least some degree of similarity between the questioned and specimen signatures but also differences. The problem for the expert is that this general description also applies when the questioned signature is a simulated copy written by some other person. The nature of each individual difference and similarity must then be assessed in terms of how closely it matches the specimens, not only in terms of form and structure but also fluency.

    Bird and colleagues showed that experts do indeed have most difficulty distinguishing between, on the one hand, genuine but disguised signatures written by the true signatory and, on the other hand, simulated signatures written by some other person attempting to reproduce the appearance of the true signatory's signature (Bird et al., 2010). The reason is that the resulting differences may appear similar in that in both cases there is usually a loss of fluency and also some difference in form or structure which are minimised to improve acceptability. The key point is that a forger is attempting to minimise differences whereas the disguiser is attempting to have as much difference from their normal signature as they believe will be accepted and not rejected. In addition, if the signature is complex, the forger may simply construct it incorrectly, whereas as noted above, the pen movements are so highly automated for the true writer that these remain similar even when introducing changes to the appearance of characters.

    The authorship of a simulated, forged signature is almost always impossible to determine (Hilton, 1952). The reason is that the natural handwriting characteristics of the forger are suppressed as they attempt to adopt handwriting features of the person whose signature they are copying. One exception is where at least parts of the forger's name and that of the target signature are shared, for example where the surname is shared. In such circumstances, there may be some evidence that parts of the questioned signature are more similar to the forger's natural handwriting characteristics and differ from the natural signature of the person whose signature has been copied.

  5. A memory copy. The ability of someone to copy a signature based on recall will depend in part on the circumstances. Husbands and wives may become very familiar with each other's signature, for example, and indeed may regularly ‘sign for each other’ which goes unchallenged unless and until there is a marital dispute! However, the recall is not usually as good as believed by those concerned and distinguishing the true signature from that of the frequent copier is often not difficult.
  6. A genuine signature of someone using more than one style of signature. It may be that the different forms share certain features or they may be completely unrelated.
  7. An accidental caused by a one off disturbance to the normal process of signing. Such signatures usually appear fluent for the most part and similar to the specimen signatures, but where the disruption to the signature occurs it is temporary as the writer regains the pattern of movement needed to complete the signature.
  8. A guided hand signature. For a detailed discussion see Section 3.2.2.
  9. A made up signature written by someone else with no knowledge of, or attempt to simulate, what a genuine signature looks like. This is the one situation where it may be possible to indicate who wrote the non-genuine signature since the writer's own handwriting features may appear in the made up signature (Hilton, 1952).
c02f001

Figure 3.4 Five naturally written specimen signatures compared to (i) a genuine signature written at about the same time; (ii) a genuine signature written many years earlier; (iii) a disguised signature written by the true signatory; (iv) a memory copy; (v) a freehand simulation.

Many documents are copied and the originals destroyed and the expert may have to do what they can based on often less than ideal quality copy documents. Physical evidence of tracing may well not be discernible from such documents and the fluency and structure of any complex overlapping pen lines may also be impossible to decipher. This can impose a severe limitation on what the expert can say in such cases and considerable caution may be needed before expressing an opinion on the authorship of a questioned signature shown on a copy document. Indeed, with the scanning and printing capabilities of modern technology, a signature that makes no attempt to be original but is shown by a copy document may well be a genuine signature that has been copied from elsewhere and added to a bogus document.

3.7 Note taking in signature cases

The notes that need to be made are inevitably very similar to those described in Chapter 2 for handwriting cases. The main difference is the need to consider very specifically the evidence for any relevant alternatives with the main points being to determine whether a questioned signature is genuine or not and, in particular, to distinguish between disguised genuine signatures and simulations. Any relevant information about the health or age of the person concerned should be explicitly mentioned and the influence this has on the final conclusion noted.

The adequacy, or otherwise, of the specimen signatures should be noted – are there enough of them and are they reasonably contemporary to the questioned signatures – crucial if the person is very ill, for example. The quality of copy documents supplied should also be noted and any details that cannot be discerned from the copy should be indicated as an area of uncertainty so that if an opinion has to be qualified it is clear why this is the case.

3.8 Reports in signature cases

The manner in which reports are written in cases involving signatures is similar to that described at the end of Chapter 2 for handwriting. For example, the limitations in a case need to be described so that reasons for a qualified opinion are made clear for the reader.

In handwriting cases the question of authorship usually revolves around the central question: was this piece of disputed handwriting written by the same person that wrote the specimen handwriting – yes or no. However, in signature cases this question is significantly modified in most instances as there is usually a need to specifically consider whether a disputed signature is either genuine or a simulation written by some other person attempting to copy a genuine signature of the true signatory. It is not so common for a questioned signature to be simply written by another person making no attempt whatsoever to copy some other person's signature.

For these reasons, the specific alternatives must be clearly stated in the report so that the findings of similarity and difference between a questioned and specimen signature can be considered in relation to them. If the conclusion is that a signature is genuine then that requires straightforward justification in the report, such as similar fluency, significant similarity in features and complex structure, all of which make simulation unlikely. However, if there is evidence of free-hand simulation (not by tracing of some kind), then identifying the author of the questioned signature is usually difficult or impossible and the reasons for this conclusion need to be briefly stated in the report by, for example, saying that when one person attempts to copy some else's signature they have to suppress their own handwriting habits and try to adopt those of the person whose signature they are copying – which inevitably changes their natural handwriting.

If a questioned signature is considered by the expert to be genuine but deliberately disguised, then the reasons for this need to be made clear especially in comparison to the possibility of simulation. Here, the nature of the differences will usually be the key issue and this needs to be clearly described so that the justification for the opinion is comprehensible to the non-expert.

 

References

  1. Al-Musa Alkahtani, A., & Platt, A. W. G. (2010). A statistical study of the relative difficulty of freehand simulation of form, proportion, and line quality in Arabic signatures. Science & Justice, 50(2), 72–76.
  2. Anonymous. (2011). Expert Evidence in Criminal Proceedings in England and Wales. (No. LC325). London: Law Commission.
  3. Ballard, L., Lopresti, D., & Monrose, F. (2007). Forgery quality and its implications for behavioral biometric security. Ieee Transactions on Systems Man and Cybernetics Part B-Cybernetics, 37(5), 1107–1118. doi:10.1109/TSMCB.2007.903539
  4. Biedermann, A., Voisard, R., & Taroni, F. (2012). Learning about Bayesian networks for forensic interpretation: An example based on the ‘the problem of multiple propositions’. Science & Justice, 52(3), 191–198. doi:10.1016/j.scijus.2012.05.004
  5. Bird, C., Found, B., Ballantyne, K., & Rogers, D. (2010). Forensic handwriting examiners' opinions on the process of production of disguised and simulated signatures. Forensic Science International, 195(1–3), 103–107. doi: 10.1016/j.forsciint.2009.12.001
  6. Dewhurst, T. N., Found, B., Ballantyne, K. N., & Rogers, D. (2014). The effects of extrinsic motivation on signature authorship opinions in forensic signature blind trials. Forensic Science International, 236, 127–132. doi:10.1016/j.forsciint.2013.12.025
  7. Dewhurst, T., Found, B., & Rogers, D. (2008). Are expert penmen better than lay people at producing simulations of a model signature? Forensic Science International, 180(1), 50–53. doi: 10.1016/j.forsciint.2008.06.009
  8. Dyer, A. G., Found, B., & Rogers, D. (2006). Visual attention and expertise for forensic signature analysis. Journal of Forensic Sciences, 51(6), 1397–1404. doi:10.1111/j.1556-4029.2006.00269.x ER
  9. Evett, I. W., & Totty, R. N. (1985). A study of the variation in the dimensions of genuine signatures. Journal of the Forensic Science Society, 25(3), 207–215. doi:10.1016/S0015-7368(85)72393-6
  10. Ghali, B., Thalanki Anantha, N., Chan, J., & Chau, T. (2013). Variability of grip kinetics during adult signature writing. Plos One, 8(5), e63216. Retrieved from 10.1371%2Fjournal.pone.0063216
  11. Gonzalez-Rodriguez, J., Fierrez-Aguilar, J., Ramos-Castro, D., & Ortega-Garcia, J. (2005). Bayesian analysis of fingerprint, face and signature evidences with automatic biometric systems. Forensic Science International, 155(2–3), 126–140. doi:10.1016/j.forsciint.2004.11.007
  12. Guest, R. (2006). Age dependency in handwritten dynamic signature verification systems. Pattern Recognition Letters, 27(10), 1098–1104. doi:10.1016/j.patrec.2005.12.008
  13. Hilton, O. (1952). Can the forger be identified from his handwriting? The Journal of Criminal Law, Criminology, and Police Science, 43(4), 547–555. Retrieved from http://www.jstor.org/stable/1139367
  14. Humm, A., Hennebert, J., & Ingold, R. (2009). Combined handwriting and speech modalities for user authentication. Ieee Transactions on Systems Man and Cybernetics Part A-Systems and Humans, 39(1), 25–35. doi:10.1109/TSMCA.2008.2007978
  15. Kam, M., Gummadidala, K., Fielding, G., & Conn, R. (2001). Signature authentication by forensic document examiners. Journal of Forensic Sciences, 46(4), 884–888.
  16. Kovari, B., & Charaf, H. (2013). A study on the consistency and significance of local features in off-line signature verification. Pattern Recognition Letters, 34(3), 247–255. doi:10.1016/j.patrec.2012.10.011
  17. Lee, J., Kong, S. G., Lee, Y., et al. (2012). Forged seal detection based on the seal overlay metric. Forensic Science International, 214(1–3), 200–206. doi:10.1016/j.forsciint.2011.08.009
  18. National Research Council. (2009). Strengthening Forensic Science in the United States: A Path Forward. Washington DC: The National Academies Press.
  19. Plimmer, B., Reid, P., Blagojevic, R., et al. (2011). Signing on the tactile line: A multimodal system for teaching handwriting to blind children. ACM Transactions on Computer-Human Interaction, 18(3), 17. doi:10.1145/1993060.1993067
  20. Schneck, C. M., & Henderson, A. (1990). Descriptive analysis of the developmental progression of grip position for pencil and crayon control in nondysfunctional children. The American Journal of Occupational Therapy, 44(10), 893–900.
  21. Sellers, C. (1962). Assisted and guided signatures. The Journal of Criminal Law, Criminology, and Police Science, 53(2), 245–248. Retrieved from http://www.jstor.org/stable/1141089
  22. Sesa-Nogueras, E., Faundez-Zanuy, M., & Mekyska, J. (2012). An information analysis of in-air and on-surface trajectories in online handwriting. Cognitive Computation, 4(2), 195–205. doi:10.1007/s12559-011-9119-y
  23. Shin, J., & Okuyama, T. (2014). Detection of alcohol intoxication via online handwritten signature verification. Pattern Recognition Letters, 35:101–104. doi:10.1016/j.patrec.2012.07.016
  24. Wilkin, T., & Ooi Shih Yin. (2011). State of the art: Signature verification system. Paper presented at the Information Assurance and Security (IAS), 2011 7th International Conference, 110–115. doi:10.1109/ISIAS.2011.6122804