Technical Note: Intra-pane refractive index variability and adequate sample sizes in forensic glass analysis

A vital aspect for the forensic analysis of glass using refractive index (RI) is the determination of adequate sample size, or how many glass fragments are needed to sufficiently characterize the RI from a source of glass, such as a side window of an automobile. The number of fragments required is inversely related to the variance of the RI across the source of glass. Previous literature indicates decreased variability across tempered glass panes over time of manufacture; however, the most recent work is over two decades old and may not reflect potential increased homogeneity in more modern glass. A set of 218 tempered vehicle windows was constructed and 30 different edge RI measurements were gathered from ten bulk fragments of each window. Variability analysis was conducted using a linear mixed effects model. Within-source (between-fragments) and within-fragment variances were found to be similar (approximately 4.3e-5 and 4.7e-5) and relatively lower than previous studies have reported. Simulation studies were also conducted, estimating the error rates based on the comparison conclusions of the sample set. 21 RI measurements were taken from seven randomly selected glass fragments of one window to characterize the known source and either nine RI measurements (three fragments, each taken from separate tempered dice) or three RI measurements (one fragment) were used to characterize the recovered source. The conclusion of the comparisons where nine and 21 RI measurements were used yielded a lower false exclusion rate (approximately 1.97%) as compared to three and 21 RI measurements (6.73%), while the false inclusion rate remained mostly stable regardless of recovered glass sample size (approximately 4.05%).(c) 2023 Elsevier B.V. All rights reserved.

Automated summary

The determination of sample size is crucial in forensic analysis of glass using refractive index (RI), and it is inversely related to the variance of RI across the glass source. A study on 218 tempered vehicle windows found similar within-source and within-fragment variances, suggesting potential increased homogeneity in modern glass. Simulation studies showed that using nine RI measurements instead of three yielded a lower false exclusion rate, while the false inclusion rate remained relatively stable regardless of recovered glass sample size.