The Spences Bridge Group of late Albian age (palynology and U/Pb zircon data of 104.5 Ma) is situated in the Intermontane Belt of the southern Canadian Cordillera. Zeolite-grade andesite lavas and interbedded volcaniclastic rocks have been sampled through a thickness of about 2900 m. Sampling has been confined to localities whose bedding attitudes are well defined and to rocks that have not undergone internal deformation. Of the 27 sites studied, 25 gave single-component magnetizations with well-grouped directions, high coercivities and high unblocking temperatures. Applying corrections for geological tilt to all 25 sites yielded minimum dispersion after 65% untilting. Analysis indicates that this is a consequence of the interaction of two modes of behavior, one (exhibited by Y magnetization at 8 sites) in which minimum dispersion occurs after 35% untilting, and one (X magnetizations at 17 sites) in which minimum dispersion occurs after 100% correction. (Actually the dispersion is essentially constant in the range 88 to 100% correction). The Y magnetizations are uninterpretable in terms of the paleofield because no estimate of paleohorizontal can be made. Because of their positive response to the tilt test, the X magnetizations were acquired before tilting, and it is argued that geologically determined bedding planes are good estimators of paleohorizontal at the time of remanence acquisition.

There may be errors because of paleoslope, but arguments are presented for believing that these are small. On this basis, the best estimate of the mean direction of the paleofield in Spences Bridge time is estimated to be D, I (declination, inclination)=38.7¿, 63.9¿ (17 sites, k=44 compared with k=11 in situ, α95=5.4) with a paleopole at 64.4 ¿N, 321.0 ¿E (K=24 compared with K=5 in situ, A95=7.5¿). When compared with the cratonic reference field for mid-Cretaceous time, this pretilting magnetization indicates a clockwise rotation of the study area of 66.3¿¿11.8 (P=0.05), and an apparent displacement from the south of 15.5¿¿7.1 (1725¿790 km). When these, and other data previously obtained from Cretaceous rocks of the Pacific Northwest, are compiled, they show an increase in apparent displacement from east to west across the Cordillera. This result is discussed in terms of three hypotheses: (1) no translation but systematic tilts of about 30¿ to the southwest; (2) translation of a substantial part of the Pacific Northwest (referred to as BajaBritish Columbia and is defined more exactly in the text) by variable amounts up to more than 2000 km from the south together with clockwise rotations; and (3) lesser translation of Baja British Columbia by about 1500 km combined with variable tilting predominantly to the south. It is concluded that paleomagnetic data, as they are presently known, are consistent with hypotheses 2 and 3, but unless serious sources of error are present, are not consistent with hypothesis 1. The most likely candidate for such errors is the uncertainty in estimating initial dip of beds (paleoslope) in volcanic rock units. ¿ American Geophysical Union 1990