We compared the magnetic fabric with the macroscopic surface lineation direction (previously assumed to be parallel with magma flow direction) for 35 dikes of the Koolau complex, Oahu. Anisotropy of magnetic susceptibility (AMS) was determined for 530 samples from 59 dikes (including a sill and a pluglike intrusive body) and two filled lava tubes, and statistically significant AMS clusters were found in all but five dikes. For the 35 lineated dikes, the mean maximum AMS direction (&khgr;1) coincides with the macroscopic lineations to within 25¿ in 18 and to within 15¿ in eight.

The AMS ellipsoid shape in about half of the samples is prolate (cigar shaped) and in about half is oblate (pancake shaped). We interpret the AMS to be determined by shape anisotropy of ferromagnetic grains. The mean total anisotropy is 2.2%, and total anisotropy (H) ranges from 0.24% to as high as 14.12%. The principal AMS axes for samples from a dike are tightly grouped on equal--area plots, and for most of the dikes the minimum AMS axes plot near to the pole of the dike. Our AMS measurements enable us to infer absolute magma flow directions for half the dikes studied; the maximum AMS axes of samples from one side of the dike form a tight cluster on an equal--area plot, and those from the other side of the dike from another tight cluster. The two clusters are close together (separated by 10--30¿) for narrow dikes (<100 cm wide), and are farther apart (30--60¿) for wide dikes and are symmetrically disposed on either side of the plane of the dike. We infer that these paired clusters are caused by imbrication of nonequant ferromagnetic crystals against each margin of the dike during deposition in a velocity gradient, and hence indicate the flow azimuth, i.e., the absolute magma flow direction. Nearly half of the dikes measured (25) gave paired AMS plots; in 12 the magma flowed up at low oblique angles 5--30¿ from the horizontal, in five it flowed down at 5--20¿, in four it flowed horizontally (<5¿), in three it flowed up at a steep angle (50--75¿), and in one it flowed down at a steep angle. The magma flow in general was rarely vertical or horizontal. ¿ American Geophysical Union 1988