The Ellsworth-Whitmore Mountains (EWM) are one of five terranes that form West Antarctica. Constraining the positions of these terranes in pre-break up Gondwana is crucial to understanding the history of the palaco-Pacific Gondwana margin. We report the results of a detailed palaeomagnetic investigation of the EWM, which comprises some 150 sites in six formations, ranging in age from Cambrian to Permian. Five of the studied units yield only viscous remnant magnetizations of recent age, or unstable natural remanent magnetizations. The remaining unit, the mid-late Cambrian Frazier Ridge Formation, yielded stable magnetizations at 16 of 35 sites. This component passes a fold test at the 95 per cent confidence level, indicating that it pre-dates Permian deformation, and we argue that it is of primary origin. The resulting palaeopole (9degreesN; 293degreesE; A(95) = 5.1degrees) is in good agreement with two previously published palaeopoles from similarly aged rocks in the EWM. Collectively these data indicate that the EWM were located in the Natal Embayment prior to Gondwana break-up, and underwent 90degrees of anticlockwise rotation during break-up. All three studies, however, yield inclinations that are slightly too shallow when compared with coeval Gondwana reference poles.

A paleomagnetic study of Paleocene, Cretaceous, and Jurassic units in the Precordillera of northern Chile (26°S to 27°S) demonstrate a complex pattern of crustal rotations. The region was selected to investigate the pattern of deformation associated with the Sierra Castillo‐Agua Amarga Fault and its associated structural subdomains, which form part of the Domeyko Fault System, the fault system that controls much of the structure of the Chilean Precordillera. Paleocene lavas from the center of the study area pass a fold test, indicating the primary nature of the remanence in these rocks. A second group of lavas from the south has a similar formation mean inclination after tilt correction for a uniform dip, which coupled with the presence of reversals, suggests that these lavas also carry a primary remanence. The data indicate clockwise rotations of ∼35° and 42° respectively, for the two sampling localities. Three geographically separate areas of Lower Cretaceous red beds also reveal primary remanences based on the presence of sites with reversed polarity at two of the localities and an inclination‐only fold test of the three locality mean directions for this formation. These sampling areas yield remanence directions that reveal 30° of counterclockwise and 31° of clockwise rotation as well as an area without statistically significant rotation. These data include the first well‐constrained counterclockwise rotation to be recovered from rocks in the southern central Andes, although mapping suggests the block involved is of limited geographic extent. Jurassic strata failed to yield any reliable results. Overall, the data indicate a greater variability in the rotation pattern than has previously been observed in most comparable areas south of the Arica Deflection. The variation cannot be explained by large‐scale rotation mechanisms, such as oroclinal bending or domino‐style block rotation by widely separated transandean faults. Instead, the rotations are consistent spatially, and temporally and in their sense and magnitude with the Eocene transpressional deformation associated with the Domeyko Fault System; the deformational event that generated most of the important structures in the studied area. The data emphasize the significance of local tectonics in controlling rotations in this part of the south central Andes and the importance of understanding the local structure of regions sampled for paleomagnetic study.

SUMMARY Palaeomagnetic poles for a stable continental block are typically defined from a combination of declination and inclination information from several temporally constrained studies. Poles from regions that have undergone vertical-axis rotation as a consequence of tectonics are excluded due to the absence of declination data. These poles, however, do contain useful information in their inclinations. We develop a simple but statistically rigorous technique allowing palaeomagnetic poles to be calculated from a mixture of declination and inclination data drawn from localities in the stable continental block and inclination data from the regions disturbed by vertical-axis rotation. Together this provides a larger data set of high-quality palaeomagnetic poles from which to calculate reference poles. The technique was used to define palaeomagnetic poles for South America for the Late Cretaceous‐Cenozoic (120‐5 Ma) period. Data from stable, cratonic South America, combined with data from Africa, rotated into a South America reference frame, and the Andean margin, yield reference poles, as well as mean poles for the Palaeogene and Neogene. Analysis of the data reveals systematic biases in the data set, and, in particular, the fit of the inclination data is poor for most time periods. In many cases, this situation is improved if the effect of inclination shallowing due to sedimentary depositional processes and subsequent compaction is removed. The best-fit poles define an apparent polar wander path for South America that is consistent with the global plate reconstruction parameters. Use of the new poles in studies of tectonic rotation should allow greater temporal and spatial resolution of vertical-axis rotation and offer the ability to identify smaller rotations in the Andean margin.

A major problem in palaeomagnetic studies of intrusive rocks lies in determining whether or not such rocks have been subjected to post-emplacement tilting. Structural analysis of dyke emplacement directions can be used to show the current attitude of the extension direction for the dyke swarm. If the original extension direction, at the time of emplacement, can be deduced from geological evidence, this then provides a field test for post-emplacement tilting of the dyke swarm and its host rocks. In the example given from northern Chile, we were able to make a palaeomagnetic study of and structurally analyse three successively younger dyke swarms that intrude similarly younging plutons. All three dyke swarms yielded extension directions close to horizontal but with markedly different azimuths. It is argued that the similarity in the plunge of the extension directions cannot be coincidental and that the dykes and their host plutons have not suffered significant post-emplacement tilting. This simple technique should be widely applicable in the assessment of post-emplacement tilting of dykes in palaeomagnetic studies.

The Ellsworth Mountains of West Antarctica represent part of a displaced terrane once situated along the palaeo-Pacific margin of Gondwana, prior to supercontinent break-up, adjacent to South Africa and the Weddell Sea coast of East Antarctica. Middle Cambrian sedimentary rocks of the southern Ellsworth Mountains host locally thick volcanic and subvolcanic rocks forming five igneous centres. Geochemically, most of the igneous samples are mafic, with a subordinate suite of evolved types. The mafic suite is geochemically varied, ranging from MORB (mid-ocean ridge basalt)-like compositions to shoshonitic and lamprophyric (e.g. LaN/YbN = 0.95 to 15.2), with εNdi values ranging from +5.2 to −2.0, correlating with Ti/Y. They are interpreted as representing melts derived from more than one mantle source, with the MORB-like rocks being derived from a depleted mantle source, and the more enriched compositions representing partial melting of lithospheric mantle. Silicic rocks contain melt contributions from Late Proterozoic crust, which is inferred to form the basement of the Ellsworth Mountains. We interpret these igneous rocks as having been formed in a continental rift environment, with MORB-like basalts erupted near the rift axis, and melts from lithospheric mantle emplaced on the rift shoulder. Such an interpretation is consistent with the sedimentary host-rock palaeogeography and contemporaneous structures. This Middle Cambrian rift event is correlated spatially and temporally with rift-related sedimentary rocks in South Africa. It is currently unclear what rifted off the southern African–Weddell Sea sector of the Gondwana palaeo-Pacific margin at that time.

In this paper we discuss the evolution and tectonic significance of the Mesozoic trench-parallel fault systems which affected the Coastal Cordillera and their relation to magmatism and crustal rotation. The oldest, extensional, fault system separates basement from rift-related Late Triassic and younger sedimentary units. This system [I] subsequently developed into a wider extensional fault system which acted as the locus of magma ascent and emplacement of the Coastal Batholith during much of the Jurassic to earliest Cretaceous period. This extensional fault system defined the forearc sliver during this period and was the consequence of a retreating subduction boundary. During the Early Cretaceous (c. 132–125 Ma) the kinematics of this fault system changed to transtension [II] and accommodated a major component of left-lateral strike-slip motion, the principal fault being the Atacama Fault Zone along which plutons continued to be emplaced. The final phase of pluton emplacement within the Coastal Cordillera appears to be c. 106 Ma, after which this magmatic arc and fault system was abandoned. An Late Cretaceous arc and fault system [III] developed some 20 Ma later and located some 50 km to the east in what is now the Central Valley of northern Chile. This paper seeks to show that the Coastal Cordillera was deformed as a whole by this Late Cretaceous fault system [III] which formed a crustal-scale left-lateral transpressional duplex. During this deformation the thermally weakened crust was dissected into a series of large-scale blocks bounded by NW-trending left-lateral strike-slip faults which merge into a NNE–SSW fault zone which forms the eastern boundary to the duplex. We term this eastern boundary zone the Central Valley Fault Zone (CVFZ) and this together with the NW-trending faults defines the duplex system which we refer to as a whole as the Coastal Cordillera Fault System (CCFS) [III]. We have traced the CCFS duplex between 25°S and 29°S and suspect that it continues northward. The timing of the deformation is constrained to be post 106 Ma, the age of Coastal Cordillera arc abandonment, and pre-Tertiary based on the deformation and pluton emplacement in and along the Central Valley Fault Zone. Palaeomagnetic data from the fault bound blocks within the CCFS duplex indicate 35°–45° of post-Early Cretaceous clockwise rotation with no substantial latitudinal motion. We suggest that the observed fault kinematics of the CCFS are consistent with this crustal-scale duplex model where rotations would have occurred in response to left-lateral transpression.

Field and earthquake data are presented which show that small faults can be important during block rotation. `Small' here means those faults which are at least an order of magnitude smaller than the largest faults at a particular scale of observation. The NW–SE-trending dextral faults of SW England have millimetres to kilometres of displacement. The largest of these faults, such as the Sticklepath–Lustleigh fault, do not appear to cause more than a few degrees of rotation from the general E–W strike of beds. Tens of degrees of rotation are often visible, however, on faults with millimetres of displacement. Similarly, the apparent rotation of beds in the Badajoz–Cordoba Shear Zone, Spain, increases as the resolution of the faults is increased. The power-law scaling relationship of earthquake magnitudes in the San Andreas fault zone illustrates that small faults can also be important in regions of active block rotation. `Small' faults can allow deformation within rotating blocks and can allow high displacement gradients to occur on the block-bounding faults. This would reduce the need for void creation, which is a requirement of rigid block rotation models. A tentative model is presented which incorporates the concept that fault behaviour is fractal, emphasising the contribution of small faults. It is suggested that future studies of block rotation should rigorously test the contribution of `small' faults, and that particular care is needed when comparing palaeomagnetic data with regional-scale structures.

The Jurassic to Recent palaeomagnetic data from the stable cratonic areas of South America and Africa are reviewed and evaluated. From the combined data set seven mean reference pole positions are calculated for the Early, Middle and Late Jurassic, Early and Late Cretaceous, Palaeogene and Neogene. Most of these new palaeopoles are well defined and indicate a complex course of apparent polar wander for the South American plate. Comparison of the palaeomagnetic data from the Andean margin with the reference poles indicates vertical axis rotations defining a broad pattern of anticlockwise rotations north of the Arica Deflection at approximately 19°S, and clockwise rotations to the south. Further analysis indicates, however, that the distribution and magnitude of rotations is more complicated than this broad division. Tectonic models proposed to explain the pattern of rotations, and the possible origins of the Arica Deflection, are discussed. It is suggested that the rotations probably result from localised in situ rotations controlled by strike-slip fault systems or thrust sheets rather than large-scale rotation mechanisms. The broad rotation pattern may result from the influence of pre-existing structures in the margin or variations in the flexural rigidity of the Brazilian Shield behind the deformation zone.

Paleomagnetic analyses of Mesozoic lavas and dike swarms from the northern Chilean Coastal Cordillera, between 25.4°S and 26.4°S, reveal a clockwise rotation of about 42°. Magnetizations from lava flows of andesitic‐basaltic composition of the Middle Jurassic La Negra Formation pass both fold and reversal tests and are interpreted as prefolding remanences. Five dike swarms of Middle Jurassic to Early Cretaceous age yield similar directions to that obtained from the La Negra Formation. Four of the five swarms have mixed polarity, suggesting that they too carry a primary or very early remanence. The structural setting of the dikes suggests that they have not suffered any substantial tilting about nonvertical axes since acquisition of the remanence. The clockwise rotation of the area is believed to have been the consequence of transpressional deformation of mid–Late Cretaceous age, post‐100 Ma, associated with abandonment of the Jurassic–Early Cretaceous magmatic arc in this region and its eastward migration to form a new mid–Late Cretaceous magmatic arc in the former back arc region. This younger arc is located east of the Coastal Cordillera and lies in the Central Valley region. The clockwise sense of rotation is consistent with other paleomagnetic data from northern Chile and southern Bolivia, south of the Arica Deflection in the Andean margin, although it is the largest yet reported. To the north of the Arica Deflection, paleomagnetic studies report counterclockwise rotations, and several large‐scale models have been proposed to explain the overall pattern of rotations. Models include oroclinal bending of an originally straight margin, differential shortening across the margin at a preexisting bend which is subsequently tightened by the passive rotation of the limbs of the bend, and distributed shear throughout the margin as a consequence of oblique convergence at a preexisting bend. In contrast to these models, several workers have argued that rotation is better explained in terms of localized in situ rotations. We review these models in light of our results and present a domino‐type model with blocks bounded by left‐lateral faults and rotating clockwise in response to mid‐Late Cretaceous transpression within a crustal scale shear zone. This is consistent with the observed strike‐slip fault systems identified in the Coastal Cordillera.