Review: Forensic Comparison of Soil Samples Using Nondestructive Elemental Analysis

Emily C. Lennert





soil, non-destructive, elemental, x-ray fluorescence, XRF, spatial resolution, Canberra distance, CNB

Article Reviewed

  1. Uitdehaag, S.; Wiarda, W.; Donders, T.; Kuiper, I. Forensic comparison of soil samples using nondestructive elemental analysis. Journal of Forensic Sciences. 2017, 62(4), 861-868.


The opinions expressed in this review are an interpretation of the research presented in the article. These opinions are those of the summation author and do not necessarily represent the position of the University of Central Florida or of the authors of the original article.


Soil is a form of trace evidence that may be encountered in forensic cases, allowing a suspect or object to be linked to a crime scene. Soil is a complex matrix that contains organic and inorganic particles, including microfossils, pollen, heavy metals, and bacteria. Samples are often analyzed by x-ray fluorescence (XRF), which allows for the determination of the elemental composition of a sample. Preparation for analysis involves grinding the soil sample into a fine powder to homogenize it; however, powdering may be destructive to certain elements of the sample such as pollen or diatoms. Therefore, a sample preparation method that is non-destructive is desired, to allow for additional analysis, e.g. pollen analysis. The authors of this study evaluated a non-destructive sample preparation method to determine whether the method was comparable to traditional sample preparation. The spatial resolution between collection sites was evaluated using both preparation methods: powdered and sieving (filtering), as well as the effect of transfer, sampling, and storage which were evaluated as well.

Non-Destructive Analysis

Ten soil samples of different types were obtained: 3 clay, 4 sand, and 3 organic-rich. Samples were divided into two equal parts. One part was prepared for analysis by powdering. The second part was sieved through a 250 micron mesh sieve. Sieving allows for homogenization of the sample without destruction. Samples were then analyzed by XRF and compared to determine whether the non-destructive method (i.e. sieving) produced comparable results to the traditional powdering method. While some differences were observed in the elemental composition between sieved and powdered samples, the values of most of the 20 elements analyzed were similar.

Spatial Resolution

Spatial resolution was evaluated to determine whether, using the sieving method, two different sites could be consistently discriminated from one another. This means that two sampling sites that are within a crime scene should share elemental composition, while an unrelated site should differ in composition. Three sites were sampled to evaluate this: (1) a 10 m x 10 m garden containing 3 garden beds, (2) a 20 m circle around a campsite, and (3) a 50 m ditch next to a rural road. At site 1, two samples were taken from each garden bed, i.e. 6 total samples. At sites 2 and 3, 10 samples were taken at each site. Canberra (CNB) distances between the elemental data were then calculated. The authors state that Canberra distance “takes the absolute difference per element between samples, weighs it to the total sums of the element in question, and those weighed differences are summed.” A Canberra distance of 0.6 or less was indicative of two samples originating from the same source, while a distance of 1.1 or greater was indicative of two samples originating from different sources. Canberra distances between 0.6 and 1.1 were considered within an area of overlap, which was neither indicative of same or different source.
At site 1, CNB distances between samples from the same spot were larger than expected. The authors cite this as a lack of homogeneity within and between the garden beds; unmixed potting soil or compost increased differences in elemental composition, resulting in higher CNB values. At site 2, an increase in CNB distance was observed as physical distance between samples increased. At site 3, CNB distances varied depending on sampling site. Samples that were taken parallel to the road had lower distances when compared to samples that were taken perpendicular to the road, as seen below. This indicates that differences in elemental composition are due to factors such as depth, land use, or construction work, rather than physical distance.

The authors concluded that, regarding spatial resolution, the optimal sampling strategy will be based on local soil variation. However, a sufficient sampling strategy may be to sample once per four square meters plus forensically relevant areas.

Transfer, Sampling, and Storage

Students of the Police Academy of the Netherlands collected twelve test cases for this study. Students were asked to make a shoe print in soil, then collect a soil sample from the shoe sole, shoe print, and an unrelated soil sample. Each sample was analyzed three times, and CNB distances were calculated. The authors reported that all test cases showed lower CNB distances, i.e. higher similarity, between soil from the shoe and shoe print versus higher distances, i.e. less similarity, between soil from the shoe’s sole and the unrelated soil sample. There appeared to be no effect from sampling being done by several different individuals, i.e. the method is not sensitive to individual sampling variations.

In 7 of 12 cases, the soil from the shoe compared to the shoe print resulted in CNB distances below 0.6, i.e. same source, while 4 fell in the overlap region, i.e. area of uncertainty. Only one case resulted in CNB distance higher than 1.1 for the shoe sole compared to the shoe print, i.e. different source. For the comparison of soil from the shoe’s sole and the unrelated soil sample, no samples indicated same source, i.e. CNB distance below 0.6. Two samples fell in the overlap area, while the remaining 10 correctly indicated different sources, i.e. CNB distance over 1.1.

Scientific Highlights

  • Sieving samples allowed for non-destructive sample preparation with little effect on sample analysis.
  • Based on CNB distances, the method is capable of distinguishing samples from the same vs different site with a high degree of accuracy.
  • CNB distances below 0.6 indicated that samples originated from the same site, while distances above 1.1 indicated that samples were from different sites. Distances between 0.6 and 1.1 were in the overlap area of uncertainty.


Soil analysis may be used to link a suspect to a scene or object associated with a crime. However, current sample preparation methods are destructive to some components in the soil sample. The method presented addresses the challenges of traditional sample preparation, while maintaining the ability to discriminate same and different source samples.

Potential Conclusions

  • The presented method is non-destructive and was shown to allow for discrimination between soil samples originating from same and different sites.