A case study is presented for applications of paleomagnetism in the volcanic field. First, the importance of magnetic cleaning was demonstrated for the present-day pyroclastic flow. Some blocks contained a very large secondary component which was removed only after heating to temperatures over 400• C. Second, progressive thermal demagnetization was used to determine the cooling history of volcanic products. In the case of a blockand-ash flow found at a depth of 450 m in a drill core, the emplacement temperature varied from block to block, giving inconsistent results. Some blocks, however, might have undergone rotations during cooling because one or two sharp kinks were often recognized in the orthogonal plots of thermal demagnetization. In another case of a Holocene block-and-ash flow, the remanence directions for block samples were completely scattered. This contradictory observation is interpreted by an eruptive sequence that the blocks had already cooled down below the blocking temperature at the summit when the lava dome was at the stage of endogenous formation. Finally, case studies of lava identification by remanence directions are given. For some lava flows which are coeval in terms of volcano-stratigraphy, remanence directions are different beyond the error, suggesting a short time gap between their extrusion.

A paleomagnetic study was carried out on volcanic rocks from Unzen volcano: samples were collected from a total of 69 sites with 19 sites in pyroclastic flows and 50 sites in lava flows. Ages for the flows were determined either by K-Ar methods or detailed field surveys, and indicate that all of the flows were deposited during the Brunhes chron. After demagnetization 10 pyroclastic and 48 lava flows had stable site-mean directions. One lava flow in the Senbongi area with a K-Ar age of 197±17 ka had an intermediate virtual geomagnetic pole (VGP) at 8.3• N, 21.6• E. The age and VGP position apparently correlate with the Iceland Basin event, suggesting that it is a record of the event on volcanic rocks. Our study added 53 Brunhes-aged VGPs to the overall collection for Japan, increasing it by 40% to a total of 175 VGPs. After excluding all site-mean directions having α 95 > 10• and VGP latitude < 50• , the remaining 148 VGPs have a mean pole at 89.7• N, 40.9showing no significant deviation from the geographic pole. The angular standard deviation (ASD) was calculated as 15.2• ± 1.2 • (N = 148), which is compatible with paleosecular variation models from the literature. However, the data set was found to deviate from a Fisher distribution. The actual meaning of the ASD value after removing the intermediate VGPs needs to be reconsidered.