Archaeomagnetic dating of a Medieval brick kiln from Tønsberg, Norway

Abstract

Archaeomagnetic dating of a Medieval brick kiln from Tønsberg, Norway

Medieval bricks and tile were never commonly used in Norway as they were in Denmark. Thus only 3 Medieval brick kilns have hitherto been found and described in Norway, and their products have not been investigated in greater detail. Dating of the bricks is therefore somewhat uncertain. In the case of the Tønsberg kiln, no tools or written sources providing a date have been found. When the kiln was excavated in 1980-81, magnetic dating was therefore planned (1, 2, 3, 4).

The kiln is situated on a small plateau at the northern edge of the Medieval town (fig. 1). Although only partly preserved, the kiln is seen to be of a rectangular shape common in Medieval Scandinavia, the flat floor and the front wall both being constructed in bricks. Five phases of the kiln were found, phase 1 being the oldest and phase 5 the youngest (fig. 2). The remains of the 5 floors are situated one above the other, except for a ca. 20 cm thick layer of sand separating phases 1 and 2.

Magnetic measurements

The magnetic dating method is a relative method based upon the principle that the direction of the earth's geomagnetic field is stored in the iron oxides of the fired clay during the last cooling, all previous magnetic memory being erased by the last heating above the Curie point of some 580°C (1, 4, 8, 9). As the direction of the geomagnetic field varies slowly with time, the secular variation (5) being typically of the order of 10 per 10 years, the direction of the measured remanent magnetization is diagnostic for the time of the last cooling, provided a mastercurve for the secular variation as a function of time is available.

A total of 27 orientated bricks still supposed to be in situ from the oldest and youngest kiln phases were collected (3), with the purpose of possibly establishing the age as well as the duration of the use of the kiln. In the laboratory, 4 to 8 cores were drilled from each brick, and the total number of 136 specimens thus obtained magnetically measured in a Digico spinner magnetometer after stepwise AF-demagnetizations up to 100 mT (1000 Oe) to eliminate possible viscous magnetic components. The rather large number of specimens was used to obtain a statistically more veliable mean direction of remanent magnetization.

Four bricks from kiln phase 1 showed anomalous directions and behaviour during the magnetic cleaning (fig. 5), attributed to disturbances after cooling in accordance with the very fragmentary preservation of this part of the kiln. The remaining specimens all showed only small directional changes during AF-cleaning of typically 2-3° in inclination and 6-8° in declination, the least directional scatter being obtained by 60 mT (6000 Oe) cleaning. The mean results and Fisher statistics are summarized in table 1.

The geomagnetic secular variation in Norway is not known further back than to about 1700 AD (5). For comparison, the british archaeomagnetic master curve (8, 9) has therefore been used, after calibration for Denmark (Rude Skov Magnetic Observatory) on the basis of the central dipole hypothesis (1, 4), see fig. 9. The Tønsberg kiln is situated at approximately the same longitude as Rude Skov, but the latitudes are 59.24°N and 55.84°N, respectively. To make the Tønsberg geomagnetic direction comparable to the direction at Rude Skov, the difference in inclination for the central axial dipole field between the two sites applies as an approximation, i.e. Δ 1₀ = 73.4° - 71.2° = 2.2°, as shown in table 1. To compare with the secular variaton at Rude Skov, we therefore apply the corrected inclinations for the kiln of 68.9° and 64.8° for kiln phase 1 and 5, respectively.

In fig. 9 these corrected mean directions for phase 1 and 5 are shown, together with their 95 % significance circles. The last cooling of kiln phase 1 appears to have taken place about 1520 ± 20 AD and of phase 5 about 1470 ± 30 AD. The sequence of these two magnetic datings is in conflict with the archaeological evidence; we may therefore conclude that the accuracies of the magnetic dates are not sufficient to distinguish between the age of the two kiln phases.

Combining both datasets, a mean direction (after correction for latitude) of (D, I)= (5.4°E, 68.3°) is obtained, with a 95 % circle of confidence of 1.8°. This suggests a mean magnetic age for the last heating of 1480 ± 20 AD. The indicated statistical error of ± 20 years, however, is most likely too optimistic, since systematic errors originating from the dipole correction and the mastercurve are not included in this; an error estimate of ± 50 years is likely to be more realistic.

Historic Evaluation and conclusion

Comparing finally with historic knowledge, we may remark that the first known use of bricks in Tønsberg was in 1276 AD, when Magnus Lagabøter's fortification was finished, which is a part of the King's castle (10). It has been shown (2) that the kiln most likely produced bricks and tiles for the castle, the king's yard, the church of St. Peter and the monastary of St. Olav in the town. These buildings are mainly stone buildings with later or secondary additions in bricks.

The castle fell into disrepair from 1503 AD, when it was burned by a Swedish army (11, 12, 13). The town was not fortified after this event, hence the King lost interest in rebuilding the castle.

A great town fire in 1536 AD caused most churches and monasteries to be relinquished, and only two churches, perhaps the king's yard and parts of the monastary of St. Olav appear to have been in use after (14, 12). People moved away from the town, and the King had to promise 6 years tax exemption to get people to return. It seems that there was no demand for ceramic building material in the town in the early 16th century.

The magnetic dating of the last use af the kiln to about 1480 ± 50 AD thus appears to be in excellent agreement with indirect information from written sources. This also supports the supposition that this kiln was the main source of tiles in Tønsberg as long as there was a major requirement for this material.

Niels Abrahamsen & Sæbjørg W. Nordeide