Sample preparation and heating experiment
Skeletal material was extracted from unembalmed human cadavers. The left and right radii, ulnae and humeri from two cadavers were used, one male (age at death 66 years) and one female (age at death 75 years). The cadaveric material was obtained through the body donation program of the Department of Anatomy, Embryology and Physiology of the Academic Medical Centre, Amsterdam, the Netherlands. The bones were manually defleshed and stored between 4 and 7 °C. Thin transverse cross sections, of approximately 4 mm, were sawn with a bone saw from the radial and ulnar diaphyses and from parts of the humeral diaphysis, until the epiphyses were reached or the required number of samples was obtained. The remaining diaphyses of the humeri were divided into sections of approximately 40 mm thick. The bone was kept wet during sawing to prevent unwanted heating due to friction of the saw.
Thermal stress was applied for varying durations in a preheated muffle oven (with an accuracy of ±2 °C) in porcelain cups, up to a temperature of 1100 °C with increment steps between 20 and 100 °C. Two surrounding media were used, air and porcine subcutaneous fat (Sus scrofa domesticus). The latter was chosen to mimic the presence of soft tissue. Heating in adipose tissue was limited to a temperature of 450 °C because of rapid autoignition. The thin transverse cross sections were heated in air to a maximum temperature of 900 °C, since this covers the temperatures generally reached during a house fire . The diaphyseal thick sections and epiphyses were heated up to 1100 °C, to enable a comparison with a modern crematory. The samples were heated and subsequently left to cool down to room temperature; details concerning the temperature, duration, medium and sample size are given in Online Resource 1 section A. During the entire process, the samples were handled with tweezers, and nitrile gloves were worn, to prevent contamination.
Samples collected from a modern crematory
Four unembalmed, undefleshed and unaltered (prior to cremation) human cadavers, which were donated to science but unsuited for preservation, were recovered after a modern cremation. The sample population consisted of two males (age at death 77 and 81 years) and two females (age at death 77 and 83 years). Three of the four cadavers were kept refrigerated between 4 and 7 °C before cremation, and one cadaver was kept frozen and thawed prior to cremation (male, age at death 77 years). The postmortem interval prior to cremation did not exceed 2 days for the refrigerated cadavers and was 78 days for the frozen cadaver. The remains were cremated at a temperature of ±1000 °C for a duration of 2.5 h and salvaged prior to pulverization.
The cremated remains were handled with nitrile gloves. The salvaged material was sieved, and metals and other non-osseous materials were removed. The cremated remains were then categorized as cranial bones, teeth, vertebrae, ribs, irregular bones, epiphyseal ends and diaphyseal fragments.
Visualization and imaging
The cortical surface of the thin transverse cross sections, the cortical, periosteal and articular surface of the diaphyseal ends and thick diaphyseal sections, and the remains collected from the modern crematory were illuminated with an ALS to induce luminescence. In total, five types of ALS were used: 350 to 380 nm (UV, peak at 365 nm), 400 to 430 nm (violet, peak at 410 nm), 420 to 470 nm (blue, peak at 445 nm), 445 to 510 nm (blue/green, peak at 475 nm) and 480 to 560 nm (green, peak at 520 nm) .
The samples were placed on a visually non-luminescent and strongly visible light-absorbing surface. The following long pass filter goggles were used to filter out the excitation light (1% transmission): 435 nm (pale yellow), 476 nm (yellow), 529 nm (orange) and 571 nm (orange). All combinations of ALS–long pass filter, higher than the excitation bandwidth of the ALS, were used in the experiment. A Nikon D700 with a 35-mm AF-D f2.8 lens was used for photographic documentation, in conjunction with long pass lens filters from Schott (1% transmission): GG455 435 ± 6 nm (pale yellow), GG496 476 ± 6 nm (yellow), OG550 529 ± 6 nm (orange) and OG590 571 ± 6 nm (orange). Digital images were taken in raw image format and postprocessed in Adobe Lightroom CC® (2015, Inc., San Jose, CA) for Mac. Contrast was enhanced by setting the levels appropriate to the image; the background surface was adjusted to black by manual selection in the majority of the images. No changes were made to the white balance, nor was the colour of the image enhanced.
Excitation and luminescence interference
Spectroscopic measurements were performed to determine the actual spectral bandwidth of the ALS. This served the purpose of determining whether any illumination light would pass through the used filters and add to the observed luminescence. Measurements were recorded using a spectrograph (USB4000 from Ocean Optics, Duiven, NL), a standard multi-mode fibre (FT400EMT-M28L01 from Thorlabs, NJ, USA) and different long pass filters (400 LP 232, 450 LP 9604, FEL0500 and FEL0600 from Thorlabs, NJ, USA). The spectral output of the ALS exceeded the respective nominal cut-off wavelengths provided by the manufacturer. The five spectra are included in Online Resource 1 section B.
In order to visually observe ALS output at wavelengths exceeding the spectral bandwidth specified by the manufacturer, and thus potential false positive luminescent observations, a mirror (PF10-03-P01 from Thorlabs) was used to inspect the reflectance. Several ALS–long pass filter combinations led to an observed reflection in the mirror. UV light reflected purple, although this was not observed in any of the photographs or when the sample was observed through the prism of the mirror reflex camera. A purple reflection was also observed when using the purple ALS with the pale yellow long pass filter. The blue ALS (420 to 470 nm) reflected blue-green in the mirror when observed through a yellow long pass filter (476 nm), as can be seen in Fig. 1. The blue-green ALS reflected green in the mirror, and the green ALS reflected yellow when observed through an orange-2 filter. The reflectance was relatively low in intensity, best described as a homogenous illumination, and disappeared when using the subsequent long pass filter.
Scoring and statistical analysis
To evaluate the effect of the thermal stress on the luminescent property of the bone, the luminescence was scored in a similar fashion as Ramsthaler et al.: present [strong] (4), present (3), present [weak] (2) or absent (1) . Present [strong] was scored when the sample luminesced as intense as a fresh sample; present was scored when luminescence was evident but not as intense as a fresh sample; present [weak] was scored when only a slight amount of luminescence was observed, and absent was scored when no luminescence was observed, see Fig. 2 for a series of samples corresponding with the scoring index. This ordinal scoring index was used for each specific ALS–long pass filter combination. The previously mentioned reflectance for specific ALS–long pass filter combinations was discarded as false positive and thus not scored as luminescence. Two observers (TK and KN) scored a total of 260 samples, all in duplicate, with 11 ALS–long pass filter combinations.
Intraobserver and interobserver agreement
The samples were scored at two different moments by two observers in randomized order, without prior knowledge on the experimental treatments. The first and second scores by the observers were statistically compared by means of a kappa test to determine the intraobserver agreement. A kappa test was conducted on the four possible pairings of the duplicate scores of each of the observers to determine and interobserver agreement . The kappa agreement scores were interpreted according to the suggested levels of agreement from McHugh . Both observers achieved an almost perfect agreement for the kappa analysis on the first versus the second score; the κ values for this intraobserver agreement were κ 0.961 (p < 0.001) and κ 0.949 (p < 0.001), respectively. The kappa analysis of agreement between the observers ranged between κ 0.870 and κ 0.892 (p < 0.001), implying an almost perfect agreement between the two observers. Details on the kappa analysis are given in Online Resource 1 section C. Further statistical analysis was, therefore, performed on the mean of the two observations of both observers, for 11 ALS–long pass filter combinations for 260 samples.
Statistical analysis of temperature-dependent and duration-dependent changes of luminescence of bone in different media
Statistical analyses were performed in Microsoft® Excel for Mac 2016 and SPSS statistics for Mac. The overall mean score (with 2σ) was calculated and plotted for the temperature groups of the transverse cross sections heated in air and adipose tissue for 10, 20 and 30 min and of the diaphyseal thick sections and epiphyses heated in air at various durations.
The intensity scores of both the transverse cross sections heated in air and adipose tissue were compared with the Mann–Whitney U test to determine the significance of the difference between the different media. In order to determine the most efficient ALS–long pass filter combination, the various ALS–long pass filter combinations were compared with a Kruskal–Wallis H test; if a significant difference was found, a multiple comparison of groups, based on the mean rank, was performed to determine which combinations differed from each other. For all tests, statistical significance was accepted at p < 0.05.