SUMMARY
An archaeomagnetic, rock magnetic and magnetic fabric study has been carried out on seven anthropogenic ash horizons in the Middle Palaeolithic sedimentary level XXIV at the rock shelter of Crvena Stijena (‘Red Rock’), Montenegro. The study has multiple goals, including the identification of iron bearing minerals formed during combustion, assessment of the suitability of these combustion features for recording the Earth´s magnetic field direction, revelation of the magnetic fabric and its significance in the characterization of cave (rock shelter) burnt facies, and identification of post-burning alteration processes. Magnetite has been identified as the main ferromagnetic component of the ash. The ash layers exhibit a high thermomagnetic reversibility in contrast to the irreversible behaviour of their subjacent burnt black layers which is related to the different temperatures attained. Seven mean archaeomagnetic directions were obtained with acceptable statistical values indicating that these features recorded the field direction at the time of burning. However, some of them are out of the expected range of secular variation for mid-latitude regions suggesting post-burning alterations. The magnetic fabric of the ash was characterized by anisotropy of low field magnetic susceptibility measurements. Statistical analysis (box and whisker plot) of the basic anisotropy parameters, such as foliation, lineation, degree of anisotropy and the shape parameter, along with the alignment of the principal susceptibilities on stereoplots, revealed variation among the ash units. The diverse, oblate to prolate, lineated or strongly foliated, quasi-horizontally and vertically oriented fabrics of the units may indicate different slope processes, such as orientation by gravity, solifluction, run-off water, quasi-vertical migration of groundwater and post-burning/post-depositional alteration of the fabric by rockfall impact. In sum, the magnetic characterization of the ash layers has shown the occurrence of different post-burning alteration processes previously not identified at the site. Alteration processes in prehistoric combustion features are often identified from macroscopic observations but our study demonstrates that multiple processes can affect them and are usually unnoted because they take place on a microscopic scale. Their identification is critical for a correct chronological and cultural interpretation of a site (e.g. collection of samples for dating, stratigraphic displacement of remains), especially if significant alterations are involved. Magnetic methods are therefore a powerful but underutilized tool in palaeolithic research for the identification and evaluation of taphonomic processes affecting prehistoric fires.