The material consisted of 2795 radiograms (Table I) of 1970 ethnic Finns (966 M and 1004 F) from the Helsinki area; ages varying between 0 and 25 years. The radiographs were taken in the period 1964 to 2005, 63% before 1980. The study sample comprised 2314 dental panoramic radiographs (PR) of non-patient subjects in two large materials, one cross-sectional, the other semi-longitudinal [2], [10], [13]. In addition, the material included 108 PRs taken of mothers of children participating in the semi-longitudinal study [10] and 214 PRs from the files of the Institute of Dentistry, University of Helsinki. The material also comprised 38 dental PRs taken at the Department of Forensic Medicine, University of Helsinki, during medicolegal autopsies of ethnically Finnish children from 0.0 to 2 years of age [14]. All had been considered healthy before death. The material also included 52 extra-oral oblique lateral radiographs taken of the jaws of healthy non-patients from 0.5 to 3 years of age [15]. Finally, the study material comprised 69 periapical radiographs of the premolar area in children from 2 and 5 years of age. These radiographs were from the files of the Institute of Dentistry, University of Helsinki. Table I.

 Distribution of radiographs by age, gender, and type of radiograph

The 8 left mandibular teeth were rated on an 8-stage scale from A to H using the Demirjian grading [4], which has clear descriptive criteria, line drawings, and radiographic illustrations of the stages. In addition, the crypt stage (O) was recorded, representing the period when the bone crypt is visible without a dental germ inside. For children in the semi-longitudinal study, we also recorded the stage with no sign of the tooth yet. This information was needed for calculating ages at attaining the crypt stage (Table II). If a tooth was missing or its image was unclear, the contralateral homologous tooth was used. The distribution of evaluated teeth by stage of development and gender is presented in Tables IV and V. The evaluator, the first author, calibrated herself regularly with the help of the Demirjian dental development computer program (Silver-Platter Multimedia Database, Silver Platter Information Inc., Norwood, Mass., USA). Table II.

 Attainment ages at 8 developmental stages of tooth formation (Demirjian) and the crypt stage (O) in 8 mandibular teeth

To study ages at attainment of a stage, those radiographs taken before 1.5 years of age were grouped into 0.25-year intervals and the other radiographs into 0.5-year intervals. However, when attainment ages for stages of short duration (stages O, A, and B) were calculated, the 0.25-year grouping was used also for teeth that reached those stages later.

As a rule, the age at attaining a given stage is presented as the median age of the youngest age group in which the prevalence of teeth at that stage, together with those at more developed stages, reached 50%. If the 50% limit was not in accord with age group, linear interpolation was provided for more precise age assessment. For example, the prevalence of stage H in first premolars (P1) of girls reached the 50% limit for the first time in the age group 12.50 to 12.99 years (median 12.72). In this age group the prevalence of H was 76%. In the earlier age group, 12.00 to 12.49 years (median 12.25), the prevalence of stage H was 43%. The interpolated age at attaining stage H was 12.35 years (Table II).

Mean, standard deviation of the mean and percentile distribution were employed to calculate age in a stage of tooth development. The Kappa index [16] was used in estimations of intra-examiner agreement. Percentage agreement of stages was also calculated.

Intra-examiner agreement in assessment of developmental stages of teeth concerning 2314 panoramic tomograms in this study has been presented previously [10]. Ratings differed in 6.3%. The Kappa index [16] was 0.91 (95% CI, 0.87–0.94). Other radiographs taken of children under 6 years of age (Table I) were re-evaluated about 4 months after the first evaluation. Of the 541 left permanent mandibular teeth, evaluations differed in 47 (8.7%) by one grade. In 22, the second evaluation was advanced and delayed in 25. As in the previous study [10], we expected the first and second ratings of an experienced observer to differ at most by 1 grade and, thus, expected the proportion of agreement by chance to be 1/3 [p(e) = 1/3], resulting in Kappa = 0.87 (95% CI).

For all teeth except 3rd molars (M3) and stages combined, attainment ages of girls were usually further advanced than those of boys (39 stages of 49, 80%). Boys were advanced at 7 stages (14%), and at 3 stages (6%) attainment ages were the same (Table II). Differences were greatest in canines, where female advancement increased from 0.64 years at the completion of the crown to 1.74 years at closure of the apex. M3 showed an opposite trend: 6 stages were advanced in males (67%) and 3 in females (33%).

For all teeth combined, in-stage age (IS age) medians (Tables IV and V) were often later than mid-attainment ages (MA age; Table III) (78/113; 69.0%). Stages attained within the first year of life or slightly later, i.e. stages A and B in incisors and canines, and stages O, A, and B in M1, were of short duration and observations of the stages were within a narrow age range. In these stages, differences between IS ages and MA ages were small.

The greatest differences between IS ages and MA ages (from 0.8 to 1.1 years) in boys were at stage C in the central (I1) and the lateral (I2) incisor and the canine and at stage D in I2. At all stages with attainment ages over 3.5 years in all teeth, the difference between IS age and MA age averages was mainly at most ±0.25 years (52/76; 68.4%) and never reached 0.5 years.

Even in radiographs of excellent quality, the earliest mineralization stages may not be visible because there is not enough mineral mass to register above the absorbance of the mandibular bone [17–19]. Furthermore, in small children, and especially in the incisor area, the tomographic layer is seldom ideal, and developing teeth may not conform well to the average sharply depicted plane of the tomographic unit [14], [20]. This is probably why crypts of anterior teeth were never visible, and early mineralization stages seldom visible. Thus, we did not quite reach our goal of providing complete gender-specific developmental schedules. Problems with the tomographic layer may also have been the main reason for the late attainment ages of stages D and E in I2 compared with I1. The fact that the image of I2 was often distorted may have affected assessments of these stages.

The great majority of the radiographs were PRs (95.7%). To increase the number of observations in small children, radiographs taken with two other techniques were also included. Radiographs obtained by various techniques have been used [3], [19], but they give slightly divergent ages for initial mineralization [14].

Age of subjects in a stage (IS age) is practical for age estimations, but truncated and uneven distributions of ages of subjects in reference samples reduce the value of data for age prediction and may result in serious bias [6], [21]. In our material extending from birth to age 25, truncation involved only the O-stage of M1, which was always present at birth indicating that the crypt could already be seen in radiographs prenatally. The greatest problem in observations was the uneven age distribution. The small number of children in age groups 1 and 2 years led to IS ages markedly too high, especially at stage C in incisors and canines, and also at some early mineralization stages of premolars and molars (Tables IV and V).

Knowledge about timing of initial mineralization in mandibular incisors is based on histological studies. Reported initiation ages [18], [22] are slightly earlier than in this Finnish radiographic study. Since radiographic studies give later mineralization schedules than do histological studies [17], [19], the present findings support earlier findings and also the view that individual variations in tooth formation in early infancy are small [1], [17]. The only radiographic study we found states rather generally that the earliest evidence of mandibular I1 is seen in the first half of the first year [19].

Mineralization of M1 was already evident at birth in a large American study with oblique lateral radiographs [1], but in Finnish PRs it was visible first at 2 months of age. Similarly, the initial mineralization of P1 was later in Finns than in North Americans of Caucasian origin [1], whereas most ages at initial mineralization in canines, P2 and 2nd molars (M2) closely resembled the North American findings [1]. Furthermore, timing of initial mineralization in M3 was in accord with several other populations [1], [11], [23].

True termination of enamel formation occurs on buccal and lingual surfaces. This can be observed in histological, but not in radiographic studies, which allow estimations of the end of crown development on proximal surfaces only. Some recent studies present histological results also in forms that facilitate comparisons with radiological studies [21], [24]. Finnish ages at radiographic crown completion of I1 (Table III) were in line with ages based on measurements of an English skeletal material [21]. Compared with recent histological findings of northern Europeans, Finnish ages resembled timing of crown completion in I1, I2, M1 and in canines, but were markedly later for M2 and M3 [24].

In comparisons between radiographic studies, it should be noted that stage D in the Demirjian grading [4] is closest to the stage “initial root formation” in the grading used by Moorrees et al. [1]. Finnish ages of incisors were earlier at that stage than those reported for both North American [11] and Japanese children [25], but, in M1, age at crown completion in Finns was close to that of the Japanese [26]. As a whole, timing of crown completion in canine, P1, P2, and M2 was close to the findings for two North American populations of Caucasian origin [1], [11], but earlier than those for children of Caucasian and African origin in the US mid-South [27]. Furthermore, timing in Finns was later than that of French Canadians [12]. On the other hand, Finnish ages for M3 were close to those of Canadians [11], [23], but both earlier [1] and later ages [3], [28] have been published. To conclude, inter-population differences in radiographic crown completion are inconsistent and may be obscured by methodological differences and problems.

The present ages at apical closure in 7 mandibular teeth resembled those reported for white children from the northeastern USA and eastern Canada [1], [11], [12]. Compared with the Japanese, ages at root completion in I1, I2, and M1 were earlier in Finns [25], [26].

In contrast to other teeth, in M3, female delay was obvious in the last stages of root development, which accords with previous observations [1], [3], [7], [23], [28–30]. In men, timing of apical closure in M3 was the same or slightly later than reported for some other populations [1], [3], [28]. In Finnish women, however, timing was the same as reported for Hispanics in the southern USA [4], but later than in many studies [1], [2], [7], [23], [28]. This finding may indicate developmental variability between populations, but it may also reflect different sizes of age groups in the studies.

The present schedules, with ages as midpoints between attainments of consecutive stages (MA; Table III) can serve for assessments of average ages in children and adolescents of all ages. The tables presenting age as average ages in all individuals at a given stage (IS; Tables IV and V) can serve the same purposes as MA tables, and they have the advantage of presenting variations. In individuals of known age, they show to what degree the development of a child's tooth is advanced or delayed and whether development falls within normal limits. However, we recommend use of our IS tables without reservations only during the first year of life and from about 5 years of age onwards. Finally, attainment schedules provide average ages for teeth when entering developmental stages. Although the two other types of schedules are often preferred in clinical work, only attainment schedules (Table II) can serve for assessments of apical closure.