**3. Error Evaluation**

To avoid interobserver error, all measurements were measured by the same author Y.Z. Also, to assess intraobserver error, three specimens were chosen and measured six times repeatedly on six different days. The coefficient of variation (CV) is calculated with the repeated measurements (as illustrated in Table 3). The intraobserver error is well controlled within most of the measurements except for the horizontal angle. All of the measurements were included in the formal analysis, although the results of the horizontal angle should be taken with grea<sup>t</sup> caution.


**Table 3.** CV calculated from repeated measurements of chosen specimens.
