Volcanic glass is often considered an ideal recording material for paleointensities. Experiments to determine the ancient field intensity are time consuming and mostly have low success rates. Studies have shown that the usage of glassy samples can increase success rates very much as the remanence carriers are in or close to the single domain range. Further, effects like magnetic anisotropy and cooling rate correction can be corrected for. The aim of this thesis is to clarify whether an ideal behavior can be expected when working on volcanic glass. Studies were done on samples of different compositions (phonolite, pantellerite and rhyolite) and varying degrees of devitrification/hydration. Rock magnetic measurements were done to determine the remanence carriers of remelted glass samples of phonolitic composition. Single domain (SD) titanomagnetites were identied to carry the stable remanence and in the course of paleointensity experiments the validity of a cooling rate correction method that makes use of the natural cooling rate of the samples as determined

from relaxation geospeedometry was shown. After correction the samples reproduce the intensity of the previously applied field. For the unhydrated samples from Montana

Blanca, Tenerife (phonolitic composition, 2 ka old) and from Mayor Island, New Zealand (pantelleritic composition, 8 ka old) remanence carriers in or close to the SD range were found and high quality well defined paleointensities were obtained. Success rates of the paleointensity experiments were 70%. The results compare very well with other paleointensities from close-by studies at the respective times. These data support the ideal

recording behavior of volcanic glass. Yet, rock magnetic and paleointensity experiments on devitrified and hydrated samples give contrary results: It was found that hydration and devitrification result in a loss of magnetic stability and remanence carriers, an increase in grain size of magnetic particles and a decrease of the "apparent" paleointensity. A possible explanation for these obsevations is a partial or full overprint of the original thermoremanent magnetization by a chemical remanence. It is therefore obvious that suchlike altered glasses are far from being ideal recorders. Great care has thus to be taken when sampling volcanic glass. If pristine glass is sampled, an ideal behavior during paleointensity experiments is probable and - as the here presented data suggest - a good estimate of the ancient field intensity is likely to be gained. If, however, the glass is unknowingly altered

the determined paleointensities are prone to underestimate the true field value.