The Practice of Handwriting Identification

Introduction

There are two kinds of interest in the practice of handwriting identification. One is that of the scholar and librarian: literary and historical research, for instance, frequently encounters manuscripts that have no obvious author, or whose authorship is disputed, and sometimes there is a need or wish to settle the matter by a comparison of handwriting characteristics. This branch of scholarship is usually considered to be part of the area of study known as palaeography. The other domain is that of forensic science: around the world there are forensic document laboratories, expensively equipped and staffed by experts, most of whose full-time job consists of analysing documents for evidence of authorship. Their discipline is called forensic document analysis.¹ Palaeographers and document analysts have curiously little in common;² the former usually have arts degrees, and are either academics in university humanities departments, or librarians, or archivists, or enthusiastic amateurs of various kinds. Document analysts, in contrast, almost invariably have science degrees, are not much interested in old documents, and have a heavy case-load dealing with contemporary material whose defining characteristic is that it is, in some way, disputed, and subject to actual or potential legal action. The work of palaeographers is open and publishable, and the material they work with is enigmatic only because it was written in the past and information about it is lacking; the work of document analysts is highly confidential, and their material is enigmatic because there is normally some allegation of deliberate deception. Palaeographers and forensic scientists read and write for different journals, attend different conferences, work in different environments, and do not communicate with each other.

This paper is an attempt to bring the two together. My own position is unusual: I am an arts academic, working in an English department, specializing inter alia in bibliography and textual criticism. I have also worked as a forensic document analyst since 1974, have examined thousands of cases, and given evidence in court many times. Recently I have become involved in three very different palaeographical projects: the Cuneiform Digital Palaeography Project,³ which takes as its purview the entirety of three thousand years of cuneiform writing on clay tablets; the International Dunhuang Project at the British Library,⁴ which studies tenth-century Tibetan paper manuscripts from a miraculously preserved archive on the Great Silk Road; and the Johnson Dictionary Project,⁵ which is interested in the manuscripts produced by Samuel Johnson and the professional scribes who were his amanuenses in the compilation and revision of his great Dictionary of the English Language of 1755. In each of these projects the scholars concerned came to me with a question: how do we analyse the handwriting in our documents in order to determine authorship?⁶

These questions arose because the scholars in each of those very different disciplines were aware of interesting similarities in the handwriting that they studied. These similarities suggested to them that different documents had the same author. But nowhere in the literature of their subjects was there an agreed and established methodology for testing the hypothesis of common authorship by examining handwriting. In each of these disciplines scholars frequently notice instances of similarity suggesting identity, and frequently conjecture about authorship based on these instances, but there is no coherent published account of how to do it.⁷ The researchers in each of the projects assumed that in forensic science, where expert evidence about handwriting identification is offered in courts on a daily basis, such a methodology would be available: could it, they wondered, be applied to their own research?

This is an interesting question. Any attempt to answer it reaches to the heart of the subject, because what it asks is: can one communicate expertise? Expertise requires both explicit knowledge, from published material, and implicit knowledge, from experience. There is now much published research on forensic handwriting analysis,⁸ but one cannot read that material and derive from it a programme, an explicit algorithm that leads in an orderly fashion to an established conclusion. Experts consult their experience; it is as if each expert has a large, and...