From Variation to Verification: A Basic Forensic Method for Identifying Single vs. Multiple Authors of Anonymous Letters

The case study involves two corpora of anonymous threatening letters sent to high-profile political figures. The first corpus (Q set) consists of 15 typed letters, and the research aimed to determine whether the Q corpus was the work of a single author or multiple contributors within a shared community of practice. Initial qualitative and quantitative analysis of the Q corpus revealed internal consistency but with apparent idiolectal variation, suggesting the possibility of multiple authors, although such variation may also arise from shared norms within a sociolinguistic group.

The second corpus (C set) includes 13 handwritten letters previously confirmed through forensic handwriting examination to have been authored by a single individual. Both corpora were matched for genre, topic, period, and ideological context.

Style markers were divided into two categories: universal markers, such as sentence length or type-token ratio, which were measurable across both corpora, and distinctive markers, referring to letter-specific syntactic and morphological patterns. Universal markers were numerically encoded and analyzed using principal component analysis (PCA) and hierarchical clustering. While the C set displayed measurable intra-authorial variation, all Q documents fell within the stylistic variation range of the C set when projected into the same PCA space. Additional Euclidean distance analysis supported these findings: intra-group distances were 3.06 for the C set and 2.05 for the Q set, while the inter-group distance averaged 3.25.

These results indicate that the observed differences within the Q corpus are more likely attributable to stylistic flexibility than to multiple authorship. Binary tracking of distinctive markers further reinforced the internal consistency of each corpus in favor of single authorship. The convergence of linguistic and handwriting examination evidence, cross-validated with ISO standards and microscope-supported techniques, strongly supports the single authorship hypothesis for the Q set.

This case illustrates the forensic value of authorial benchmarks based on intra-authorial variation and demonstrates how such benchmarks enhance the reliability and feasibility of both qualitative analysis and statistically defined thresholds. It also underscores the necessity of defensible and replicable methodologies and protocols in forensic authorship analysis.