Identification of osseous materials is generally established on gross anatomical factors; however, highly fragmented or taphonomically altered materials are often problematic and alternative methods, such as biological, histological, or chemical analysis, must be utilized. Recently, chemical methods have been proposed to sort unknown materials according to their Ca/P ratios. Ubelaker and colleagues (2002) proposed using SEM/EDX to achieve this distinction and Christensen and colleagues (2012) have validated X-ray Fluorescence Spectrometry (XRF) for this application. An alternative method of analysis involves performing principal component analysis (PCA) on element spectra to classify unknown materials based on their trace element composition. Zimmerman (2013) proposed the validity of this method with data obtained using hand held XRF. Subsequently, performing PCA on elemental data obtained using SEM/EDX demonstrates potential for material differentiation. Elemental weight percent data were collected using SEM/EDX then processed in R, version 3.0.1, by the R Foundation for Statistical Computing using PCA and Fisher Linear Discriminant Analysis. A two-tiered analysis was undertaken to improve discrimination between sample groups. The first tier involved distinguishing between osseous and non-osseous materials. After outliers were removed overall correct classification was 98.02% with one of 1504 osseous and 39 of 520 non-osseous spectra misclassifying. Since forty spectra were collected for each sample, the single misclassifying spectra would not affect the overall classification of the sample, resulting in 100% correct classification with a 0% error rate for the osseous samples. The second tier assessed differentiation of human and non-human osseous materials but demonstrated a poor correct classification rate of 72.41%. Finally, a blind study was conducted using 20 samples to assess the applicability for using this method to classify unknown materials as osseous or non-osseous. All of the samples were correctly classified resulting in 100% correct classification, further demonstrating the efficiency of SEM/EDX and statistical analysis for differentiation of osseous and non-osseous materials. Due to its high specificity, small sample requirements, and relative non-destructive testing protocol, as well as its presence in most modern crime laboratories, SEM/EDX has been proposed as a laboratory method for chemical differentiation of osseous and non-osseous materials. Additionally, the proposed method does not require advanced training or knowledge of analytical chemistry as the SEM/EDX provides clear results that can be processed using publically available statistical analysis software. By assessing and improving chemical analysis methodologies used for material differentiation, forensic anthropologists might be able to identify osseous and non-osseous samples as a preemptive step in forensic investigations involving fragmentary and taphonomically modified materials, reducing time and cost investments spent on forensically insignificant samples.
| Identifer | oai:union.ndltd.org:ucf.edu/oai:stars.library.ucf.edu:etd-2285 |
| Date | 01 January 2014 |
| Creators | Meizel-Lambert, Cayli |
| Publisher | STARS |
| Source Sets | University of Central Florida |
| Language | English |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Electronic Theses and Dissertations |
Page generated in 0.0025 seconds