Does shape matter? A comparative study of the usage of calibrated sieves in the study of burned human bone from archeological and forensic contexts

In recent years, the use of a set of sieves became a crucial part of the methodology in studies of burned human bone from archeological and forensic contexts. A set of three sieves (10 mm, 5 mm, 2 mm) is widely used in studies of burned human remains, particularly those from the British Isles, in order to determine bone fragmentation, following methodology proposed by Jacqueline McKinley (1993). Usually, mesh opening sizes are reported in methods; however, their shape is rarely mentioned. This study aims to analyze the potential impact of the sieve mesh opening shape on the results achieved. It was tested by sieving the human burned bones from ten cremation burials from two Polish archeological sites by using two sets of sieves (10 mm, 5 mm, 2 mm). The first had round openings, while these of the second were square. Furthermore, the intra-observer and inter-observer errors were computed to assess the significance of differences between the tested sets of sieves. The analysis of values obtained by using a distinct set of sieves revealed noticeable differences. Specifically, when comparing the results of sieving through square versus round mesh openings, the paired t-test yielded lower p-values than intra-observer error measurement. This finding suggests that the observed differentiation between two sets of sieves is less likely to be a result of random chance. However, it is important to note that the inter-observer error also showed statistically significant differences. This indicates that the sieving techniques may vary between observers, potentially impacting the results obtained. The diagonal of square mesh openings is proposed as the most probable factor causing differentiation. However, it should also be considered that the differences in the distribution of mesh openings and total sieving surfaces between square and round openings might also alter the results. In addition, the sieving surface has higher values for sieves with square mesh openings, which follows the pattern of a higher diagonal of mesh openings for square sieves. In conclusion, the shape of the mesh openings and their distribution on the surface of the sieves might have a significant impact on the results concerning burned bone fragmentation. Consequently, information about the shape of mesh openings should be included in the “Materials and methods” section, along with the specifications of the apparatus and the dimensions, as well as the diameter of the mesh openings.


Introduction
The use of a set of sieves in forensic anthropology and bioarcheological studies of human cremains has become a standard practice, beginning in the 1990s (McKinley 1994a) and continuing to develop in the first two decades of the twenty-first century (e.g., Coulombeix and Schuilar 2017;Jaskulska 2018Jaskulska , 2020McKinley 2000McKinley , 2004Silva 2015;Thompson et al. 2016). Usually, they are used as an essential part of the recovery and excavation of cremation burials from archeological sites (Annaert et al. 2011;Borić et al. 2009;Brothwell 1981;Higgins et. al. 2020), forensic case scenes (Bontrager and Nawrocki 2008;Fairgrieve 2007), and laboratory pre-study material preparation (Cerezo-Román and Williams 2014;Fontijn and Cuijpers 2002;Gonçalves et al. 2010Gonçalves et al. , 2015Marado and Braga 2018;Warren 2008; 1 3 109 Page 2 of 9 Zana et al. 2017). However, sieves are also a crucial methodological element of the analytical protocol proposed by Jacqueline McKinley (1994a) applied in the bioarcheological study of human cremains. Weights and proportions of size fractions obtained by the sieving of cremated bone through them, along with the measurements of the maximum bone fragments sizes for each fraction, determine the bone fragmentation degree in cremation burials and allow for quantitative and comparative analysis between cremation burials (e.g. McKinley 1994aMcKinley , 2000McKinley , 2004Silva 2015;Squires 2012). The fragmentation itself is caused by multiple factors, such as conditions during the cremation process (maximum temperature of the pyre, duration of cremation, degree of combustion), the biological profile of the deceased, the collection of bones after the cremation, depositional conditions (urned vs. unurned, disturbed vs. undisturbed, type of soil, etc.), and process of the both field and laboratory excavation of the burial features containing human remains (McKinley 1994b(McKinley , 2000(McKinley , 2016. In fact, rather than reflecting the actual sizes of bone pieces after the cremation burial rite bone fragmentation is influenced by the post-excavation conditions (McKinley 2000).
Some authors have argued that sieves should not be used in the study of burned human remains. The two most common factors quoted as criticisms against the use of sieves are the potential mechanical damage to bone pieces and the separation of the burned human bones into fractions based only on two dimensions (Waterhouse 2013). In some cases, fragmentation is estimated by measuring the maximum bone fragment size and the weights of the bones in the burials (Waterhouse 2013;Watson et al. 2015). The degree of bone fragmentation in Scandinavia is commonly described by two variables: weight and volume, which generate the Fragmentation Index (Harvig and Lynnerup 2013;Harvig et al. 2014). However, in recent years, the fragmentation of burned human bones measured by sieves has remained one of the most popular methodological approaches in the study of human cremains.
The most commonly used set of sieves in Europe is the group of 10 mm, 5 mm, and 2 mm sieves (Dzierlińska 2019;McKinley 1994aMcKinley , 2000McKinley , 2004MacGregor et al. 2003;Squires 2011;Silva 2015;Thompson et al. 2016;Wysocki and Cummings 2007). This standardized apparatus selection is the result of the uniform and consequent recommendations in the British guidelines regarding the osteoarcheological study of cremains (Mitchel and Brickley 2017). According to Gonçalves and Pires (2017), 70% of the papers using 10 mm, 5 mm, and 2 mm sieves were published by researchers from British universities and research institutes. In addition to that, Portuguese (Silva 2015) and Polish (Budziszewski 2018; Jaskulska 2018; Kontny et al. 2019) publications are worth noting. Alternatively, sets of 10 mm and 3 mm sieves are commonly used by Dutch researchers (e.g., Van den Bos and Maat 2002;Veselka and Lemmers 2014). Bioarcheological analyses conducted by scientists using this set of sieves are based on the methodology of counting the rate of intactness developed by Maat (1997).
However, an important feature of laboratory equipment is often overlooked in the abovementioned studies. This aspect is the shape of the mesh openings. No previous studies have focused on the impact that the shape of mesh openings has on the measurements obtained. This article presents the results of the first study examining this issue and explores whether differences between the results obtained from two types of sieves make data comparison impossible without specifying the shape of the mesh openings (Fig. 1). Intraobserver and inter-observer errors were also calculated to observe the significance of these differences.

Materials and methods
The cremains used in this analysis were taken from two Iron Age cemeteries attributed to the Bogaczewo culture (Nowakowski 1982;1991) from northeastern Poland. The first consisted of burials from the Paprotki Kolonia I site 1 (Karczewska and Karczewski 2007;Jaskulska 2011) excavated in the years 1992 and 2003. The second sample consists of the burials excavated in 1994 from the Wyszembork IVa site 2 (Szymański 2005). The most critical factor for the selection of these two samples for this study was the visible distinction in the bone fragmentation rate. This criterion allows testing whether the differences in the measurements obtained will depend in any way on the degree of bone fragmentation. Five bone deposits weighing more than 700 g were randomly selected from each site (n=10). The samples were selected due to the macroscopically visible differentiation in the bone fragmentation (Fig. 2).
The burials of Paprotki Kolonia show smaller bone fragmentation, a bigger maximum bone fragment size, and greater variability of thermal alterations compared to the cremation burials from the Wyszembork sample, which exhibits the opposite with most of the bone fragments fully calcined.
The bone remains from each burial were dry sieved through two sets of calibrated sieves with similar mesh sizes but differing in shape (round versus square  (Fig. 1). Differences in the shape of the mesh openings result in a different hole diameter. For round holes, the diameters are 10 mm, 5 mm, and 2 mm. However, when it comes to square holes, the dimensions 10 mm, 5 mm, and 2 mm refer to the dimensions of the sides of the opening, not its diameter. Thus, the dimensions of the diagonals of Cremation burials from both archeological sites were excavated in the 1990s and 2000s. Burned human bones were cleaned and separated from the external material shortly after excavations and stored in the depot of the Faculty of Archaeology at the University of Warsaw. Therefore, the burials used in this experiment did not require further washing or cleaning. Before sieving, each bone deposit was weighted on the electric scale Steinberg System (SBS-LW-7500A) with a readout precision of 0.1 g. The methodical protocol was based on the analytical procedure developed by Jacqueline McKinley (1994aMcKinley ( , 2004McKinley ( , 2017) and a recent methodological publication by Elżbieta Jaskulska (2020). Each sieve separated the cremated bone pieces into three fractions (10 mm, 5 mm, and 2 mm), which were independently weighed. Subsequently, the total bone weight of each burial was measured again and prepared for the next sieving. The bone deposits were sieved three times for each type of sieve, producing six measurements for each object.
In order to test the potential impact of differences visible in the distribution of mesh openings between sieves with square and round mesh openings ( Fig. 1), the sieving surface for 10 mm and 5 mm sieves for each set of sieves was calculated. The calculation did not apply to the 2 mm sieves. Osteological material retained on this sieve consists of small, unidentifiable bone fragments, bone dust, and small fragments of tooth roots (Gonçalves et al. 2015;McKinley 1993McKinley , 2004. Therefore, any material that would pass through a 2 mm sieve will be very fragmentary, regardless of the shape and dimension of the openings in both sieves. Moreover, in the case of the equipment used in this study, a 2 mm sieve with square mesh openings consists of wire ropes evenly distributed over the entire surface of the sieve. In the case of round mesh, the openings are also evenly distributed, and potential differences resulting from the diameter and sieving surfaces between these screens are marginal. The sieving surface was determined by counting the number of mesh openings for each sieve and multiplying by the surface of one hole for each sieve (Table 1). In this way, the total surface through which smaller bone fragments than the mesh opening can pass during sieving was determined. The values for the sieves with square mesh openings were then subtracted from the values for the round ones to calculate the difference between them.
Due to the normal distribution of the data, the paired t-test was applied to observe the differences between the results obtained during sieving through two types of calibrated sieves. Moreover, to assess the technical error measurement (TEM) of both intra-observer and inter-observer variability, additional paired t-tests were conducted. The intra-observer TEM was measured with a 2-week separation period between surveys, while the inter-observer TEM was measured with a six-month separation between surveys (Ulijaszek and Kerr 1999;Ulijaszek and Lourie 1994). Statistical analyses were performed using MYSTAT software version 12.

Results
Regarding the disparities in the diameters of the mesh openings and sieving surface between the two types of sieves, it can be observed that the sieves with square mesh openings exhibit higher values for both the diameters and sieving surface in both the 10 mm and 5 mm sieves. The difference in diameters is 4.14 mm for 10 mm sieves and 2.07 mm for 5 mm sieves. In terms of the difference in sieving surface, the sieves with square mesh openings exceed the round ones by 6 cm 2 for 10 mm sieves and 16.05 cm 2 for 5 mm sieves.
The results of the paired t-test determining the intraobserver error are presented in Table 2. Statistically significant differences were observed in each of the measurement samples analyzed, except for the largest sieving fraction (>10 mm) and the smallest sieving fraction (>2 mm) obtained by sieving through sieves with square mesh openings (Fig. 2). In the case of the 5 mm sieves with round mesh opening measurements, significant differences were observed in the two pairs. In the remaining measurements, the p-value <0.05 was reported in one paired measurement.
The inter-observer error was also measured by the paired t-test (Table 3). Statistically significant differences were observed in all pairs for sieves with both square and round mesh openings in the >10 mm and >2 mm fractions. Interestingly, no statistically significant differences were noted in any of the pairs within the >5 mm fraction. For this comparison, we can see a trend in which the measurements for the first observer are higher only in the case of the >10 mm fraction for both sieves with round and square mesh openings. Consequently, the results obtained for the other two fractions were lower for the first observer.
The results of the paired t-test between the sieves with square and round mesh openings are presented in Table 4. Statistically significant differences have been reported in all compared measurement pairs. Figure 3 presents the visible pattern of the differences between the weights obtained from the two sets of sieves. The weights of the >10 mm fraction obtained from the sieves with round mesh openings are higher in each of the tested pairs, and the values of the two subsequent fractions exhibit the opposite pattern (Fig. 3). Additional paired t-tests were conducted to test the differences between the sieves with square and round mesh openings for the second observer (Table 5). The trend was similar, with the weights of the >10 mm fraction being higher in the case of round mesh openings and the remaining pairs (>5 mm and >2 mm) exhibiting higher values for square mesh openings. It is worth noting that the t-test values were higher in the case of measurements performed by the second observer.

Discussion and conclusions
This study presents the first survey of the impact that the shape of the mesh openings of the calibrated sieves has on the obtained results. Despite having the same mesh size (10 mm, 5 mm, and 2 mm), there are statistically significant differences in the weights of the sieved fractions obtained from the screening of the osteological material through two sets of sieves.
The observed differences do not imply that we cannot compare the raw data obtained from two different sets of sieves. However, we have to be aware of the potential impact these factors may have on further analysis. Therefore, it is advisable to use a standardized apparatus and methodology in studies to determine the likelihood of differences generated by various sets of sieves and various observers. The results of the intra-observer error measurement also exhibit statistically significant differences. They are visible mainly  in the >5 mm fraction, both in the case of sieves with square and round mesh openings. However, the intra-observer error statistical significance and t-value are lower than their values in the case of the differences caused by different sets of sieves. On the other hand, the results of the inter-observer error experiment showed statistically significant differences in the achieved measurements, which should also draw our attention as a potential factor affecting our results. Statistically significant differences occurred in the >10 mm fraction. In practice, it is the most important fraction in the analysis of archeological cremains, as it contains the largest bone fragments, which provide the most reliable information about the biological profile of individuals. In addition, the rate of anatomical identification is usually the highest in this particular fraction (Dzierlińska 2019;Gonçalves et al. 2010;Jaskulska 2018Jaskulska , 2020Kontny et al. 2019;Wysocki and Cummings 2007). Although these differences do not affect the holistic results of bioarcheological analysis of cremation burials (Borić et al. 2009;Gilmour et al. 2010;Gonçalves et al. 2015;Hart et al. 2010;Silva 2015), they may alter the analysis of bone fragmentation and further statistical analyses based on quantitative data. Statistically significant differences between results obtained by two different observers using the same type of sieves, both with square and round mesh openings, were also observed. While it is currently unclear which specific behaviors affect such significant differences in inter-observer results, it is important to be aware that comparative studies based on analyses conducted by different researchers may be biased because of statistically significant differences between observers. Therefore, the potential impact of inter-observer error should be considered in future studies involving fragmentation data. It is also worth noting that the differences in the results obtained by the same observer were not as statistically significant as the inter-observer error. To address this, it is essential to include the issue in the standards and recommendations for sieving methodology to reduce differences between observers.
In the case of the comparison of results obtained by sieving material through sieves with round and square mesh openings, we can observe the recurring trend (Tables 4, 5 and Fig. 3) where higher values are obtained by sieving through the sieves with square mesh openings, except for the 10 mm Fig. 3 The box and whiskers plot present the range of weight (g) within the three fractions obtained by sieves with round and square mesh openings. The box presents the range from the lower quartile (Q1) to the upper quartile (Q3). Whiskers denote the most extreme data points not considered outliers. The middle line shows the median (Q2) of observations. The arithmetic mean (x) is marked for each plot sieve. In the case of the largest fraction, the weight values are higher in the case of round meshes. This could be a result of the differences in the diameters, but also the differences in the sieving surface between the two types of sieves. Consequently, when sieving the material through the sieves with square mesh openings, more larger fragments (and thus heavier) might pass to the subsequent sieves than in the case of the round mesh openings, where the calibrated dimension of the mesh is equal to its diameter. As a consequence of this differentiation, the greatest gross differences in the weights among all of the fractions will be observed.
In addition, we should also note that the sieves with square mesh openings have less evenly placed holes. Square meshes do not have their openings in the most external portions of the sieves. In the case of the sieves with round mesh openings, the holes are located almost all over the surface. However, the total sieving surface is higher in the case of sieves with square mesh openings. This indicates that the factor of differences in opening diameters probably has a greater impact on the observed differences in the results of sieving, particularly considering that the square mesh openings are less numerous in both the 10 mm and 5 mm sieves (Table 1). However, it cannot be concluded that the differences in the distribution of holes on the sieve surfaces have any effect on these results. Accordingly, the diameters of the openings and their placement on the individual sieves are likely to have the greatest impact on the observed differences in results.
Additionally, it is important to keep in mind that the cremation contexts used in this experiment were dry-sieved multiple times, which could have potentially resulted in mechanical fragmentation of the burned bones. It will be difficult to find space for a thorough and in-depth study of this very essential issue in this paper. It will be worth devoting another publication to it. This would be an opportunity to utilize the data collected in this study and base the experiment on the data presented in the Supplementary materials. The matter of the index of mechanical fragmentation and agency caused by it cannot be ruled out; however, based on the observations and reasoning presented above, it can be concluded that the shape and distribution of the holes will nevertheless have a greater impact on the differences than mechanical fragmentation.
The precise description of the mesh size in the Methods section of the articles became the proper standard in bioarcheological studies over human cremains, which utilize a set of sieves and fragmentation methodology (e.g., Dzierlińska 2019; Gilmour et al. 2010;Jaskulska 2018Jaskulska , 2020Maradoa and Braga 2018;McKinley 1994;Silva 2015). Now I believe it is time to include in the Methods section also the shape of the mesh opening, a more detailed specification about the type of sieves used, introducing also information about the distribution of the mesh openings. Without this important information, any kind of comparative study will be impossible. The use of published raw weights calculated within the fractions achieved with the set of sieves will be valid only if the information about the used sieves is precisely described. Furthermore, in cases where multiple observers are involved in the study, it is critical to explicitly include this information in the Methods section. This inclusion serves as an additional measure to minimize differentiation caused by technical factors. Finally, in such studies, the inter-observer error should be calculated. Based on the results, we can argue that the information about the shape of mesh openings might also be crucial information, which should be shown in the method section of studies using the sieves. This study suggests that the use of different types of sieves can produce altered results. Above all, we should be aware that both of these factors can affect our results. Therefore, it is important to report the precise specifications of the apparatus used in the studies, including the shape of the mesh openings, to prevent incorrect comparative interpretations that might arise from this issue rather than from actual differences between the samples. However, in cases where the fragmentation analysis is carried out by multiple observers or when comparing data from different publications, it is essential to consider the problem of inter-observer error and the lack of standardization in measurements, which can impact the accuracy of the results. Consequently, the influence of the shape of mesh openings on the sieving results may obscure the inter-observer error. Concluding, based upon the results discussed in this article, we can argue that shape does matter. However, it is not the only technical factor that might affect our results. Variation in the distribution, number, and thus total surface of mesh openings on the sieves has also been observed and cannot be ignored as an element influencing the observed variances. Finally, it is important to note that the variety of sieving techniques used by each scientist using this methodology can also alter the results.
Acknowledgements I would like to thank Elżbieta Jaskulska for the inspiration for writing this article. I would like also to thank Wiesław Więckowski, Rafał Fetner, and Elżbieta Jaskulska for essential advice and suggestions with the manuscript. I would also like to express my immense gratitude to the anonymous reviewer for his tremendous contribution in improving the quality of the text in both editorial and substantive aspects, as well as for his extensive bibliographic support. I would also like to thank Iga Szwed for helping me measure the interobserver error.
Author contribution Adam Budziszewski designed the research, collected data, performed the statistical analysis, and wrote the manuscript.