Where tectonic deformation reorients rocks without penetrative strain, thei
r paleomagnetic vectors may be restored to their original attitudes by unti
lting. For strata, paleomagnetic inclination is readily restored but the ti
lt axis must be precisely known if paleodeclination is required. For dikes,
without the know-ledge of the rotation(s), neither declination nor inclina
tion of the paleomagnetic vector can be uniquely defined. Furthermore, back
-rotating dike orientations to an upright attitude assumes primary vertical
ity whereas primary dike dips are bimodal across the spreading axes (e.g. T
roodos ophiolite, Cyprus). In the Cyprus ophiolite, the dikes of the Limass
ol Forest Transform Zone are tilted due to uplift of the mantle-sequence ro
cks and deflected against the Arakapas Fault. Their paleomagnetic vectors m
ay be restored rotating about the two axes defined by the strike and the ve
rtical, or about a net axis that is possibly the actual tectonic rotation a
xis. This net axis is determined from the tectonic regional dispersion of t
he dike orientations. In this test case, the results of the restorations di
ffer slightly but underline the difficulty in selecting the best restoratio
n procedure and the greater difficulty of restoring the paleomagnetic data
from dikes vis g vis strata. For dikes, it is recommended that the paleomag
netic vectors are restored using average dike orientations to minimize the
inaccuracies due to the large primary variation in dike orientation. (C) 20
01 Elsevier Science B,V. All rights reserved.