The aim of this study was to collect statistical measures for the temporal progression of third molar mineralization in a northern Chinese population. Tables 3 and 4 show the minima, maxima, means with standard deviations, and medians with lower and upper quartiles for mineralization stages D to H, separated by sex. When applying these measures in age assessment practice, it should be noted that the mean values and standard deviations for stage H depend on the upper age limit of the sample examined and are therefore not suitable for age assessments.

Several studies are available on third molar mineralization in Chinese populations. Zeng et al. (2010) studied 3153 OPGs of southern Chinese individuals aged 4–26 years [24]. Li et al. (2012) examined 2078 OPGs from a western Chinese population [25]. The age of the individuals was between 5 and 23 years. Qing et al. (2014) studied OPGs of 2192 southern Chinese subjects aged between 8 and 25 years [26]. Guo et al. (2015) examined 3212 OPGs from a northern Chinese population [27]. The age of the individuals was between 5 and 25 years. Liu et al. (2018) studied 2519 OPGs of central southern Chinese individuals between the ages of 8 and 23 years [28]. Table 5 shows the mean ages with standard deviations from these studies for stages D-H of tooth 48. For comparison, the values for a German and a South African population are also given. Olze et al. (2004) examined OPGs from 1,430 Germans and 584 South Africans in the 12 to 26 age group [23].

An analysis of the mean ages in these studies does not reveal a consistent picture. In the studies by Li et al. (2012), Guo et al. (2015), and Liu et al. (2018), the mean ages in the low and high stages are lower and in the middle stages higher than in the present study [25, 27, 28]. Noteworthy are differences of more than two years in stage H in the studies by Li (2012) and Liu et al. (2018) [25, 28]. In the study by Zeng et al. (2010), the mean ages in all stages are lower than in our study [24]. In the South African population, the mean ages are predominantly lower than in our study [23]. In the study by Quing et al. (2014), the mean ages in the lower stages tend to be higher and in the high stages tend to be lower than in our population [26]. In the German population, except for stage H, the mean ages are higher than in our study [23].

Knell et al. (2009) and Gelbrich et al. (2010) have shown that the age range of the study population influences the mean ages if the youngest and oldest individuals who can exhibit the respective stages are not included [29, 30]. Truncation of the upper age segment leads to a reduction in the mean ages, while truncation of the lower age segment leads to an increase in the mean ages. This effect may partly explain the differences in mean ages in the studies compared. In our study population, the youngest individuals were 12 years old. In the study groups of Zeng et al. (2010), Li et al. (2012), Quing et al. (2014), Guo et al. (2015), and Liu et al. (2018), the youngest individuals were between 4 and 8 years old [24,25,26,27,28]. Since stage D can occur before the age of 12 [24, 27, 28], the lower mean ages for stage D in the studies by Zeng et al. (2010), Li et al. (2012), Guo et al. (2015) and Liu et al. (2018) can be explained by the lower age limits of their samples [24, 25, 27, 28]. However, the mean ages for stage D in the study by Qing et al. (2014), which also has a lower age limit than our sample, are slightly higher than in our study [26]. The fact that the mean ages for stages G and H in our study are significantly higher than in almost all other studies compared cannot be adequately explained by the upper age limit of the samples.

Gelbrich et al. (2010) pointed out that the mean ages depend not only on the age range of the sample, but also on the age distribution within the sample [30]. A predominance of younger subjects leads to a decrease in the mean ages, and a predominance of older subjects leads to an increase in the mean ages. This effect could be another reason for the different mean ages.

The studies compared do not allow to deduce any influence of ethnicity or different dietary habits on the mean ages. This becomes particularly clear when comparing our data with the data from Guo et al. (2015) [27]. Both studies examined a northern Chinese population from the same catchment area. However, the mean ages differ by up to 1.7 years, which cannot be explained by the influence of ethnicity or different dietary habits.

If exceeding legally relevant age limits must be proven with the highest degree of certainty, the minimum ages of the mineralization stages are decisive. For the most forensically significant age limit of 18 years, the minimum age of stage H is of interest. In our study, this minimum age was 18.8 years for males and 18.5 years for females, and thus above 18.0 years for both sexes.

Table 6 shows the minimum ages of stage H of tooth 48 in various Chinese populations. In the study by Bai et al. (2008), the minimum ages are higher, while in the other studies they are lower than in our study [31]. It can generally be assumed that the number of cases influences the minimum age. The higher the number of cases, the higher the probability that the youngest individuals in the respective stage were included in the sample. In the studies by Zeng et al. (2010), Guo et al. (2015), and Liu et al. (2018), significantly more individuals were examined in stage H than in our study [24, 27, 28]. In the studies by Guo et al. (2015) and Liu et al. (2018), the minimum ages for both sexes were below 18.0 years, while in the study by Zeng et al. (2010), the minimum age for females was below 18.0 years [24, 27, 28]. The minimum ages we have reported are therefore not suitable for age assessment practice.

In the study by Liu et al. (2018), the minimum ages are surprisingly low, at 16.5 years for males and 16.1 years for females [28]. In addition to differences in case numbers, incorrect stage determinations may also be a cause of differing minimum ages. These are particularly problematic when applying the minimum age principle. In most previous studies, all stage determinations were made by a single examiner. To determine intra-observer agreement, some or all of the cases were re-evaluated by the same examiner. To determine inter-observer agreement, some or all of the cases were evaluated by a second examiner. The agreement was never 100%, which means that there must have been misclassifications. The first determinations made by the first examiner were generally used to calculate the statistical measures. It can be assumed that this approach resulted in misclassifications being included in the evaluations. To minimize such misclassifications, stage determination in our study was performed by consensus.