1 A/m) and/or magnetic susceptibility (MS 10–4 SI) than sedimentary formations (NRM 
10–1 A/m; MS < 10–4 SI) for which the FMS was primarily developed.
For obvious economic reasons (exploration and exploitation of oil and gas), most logging tools, algorithms, and interpretative techniques were initially developed for sedimentary formations. In the framework of ocean drilling programs Deep Sea Drilling Project (DSD)) and Ocean Drilling Program (ODP), these technologies have been progressively applied to igneous and metamorphic ("hard rock") formations, tackling problems linked to oceanic crust internal structures (Goldberg, 1997; Ayadi et al., 1998; Brewer et al., 1998, 1999). In addition to traditional wireline logging tools (spectral gamma ray, lithodensity, electrical resistivity, porosity, acoustic velocities; see, for example, Rider, 1996), the Schlumberger Formation MicroScanner (FMS) and the Ultrasonic Borehole Imager (UBI) bring a new dimension to in situ characterization of penetrated formations (Serra, 1989; Pezard and Luthi, 1988; Pezard et al., 1988; Lovell et al., 1998). In the following discussion, we focus on the routinely used FMS-sonic tool string (www.ldeo.columbia.edu/BRG/), but an identical discussion is also valid for the UBI, which was recently deployed during ODP Leg 206 (Wilson, Teagle, Acton, et al., 2003).
The FMS is a probe with four orthogonal pads, each containing 16 microelectrodes (or buttons) of 6.5 mm radius. By measuring local microconductance of the formation, the electrodes provide a high-resolution resistivity image of the formation. Detection of resistivity contrasts allows identification of lithology changes, veins, fractures, and faults. When the image is "unrolled" and displayed from 0° to 360°, linear features intersecting the borehole appear as sinusoids (Rider, 1996). Assuming that FMS images are properly oriented to geographic north, the amplitude and minimum of the sinusoids can be related to the dip and azimuth of the associated feature, respectively, and, consequently, provide fundamental structural information regarding the encountered formation. Where the same fracture can be identified on both the images and the core, properly oriented borehole wall images also allow reorientation of cores, a key step in structural and paleomagnetic studies (MacLeod et al., 1995; Célerier et al., 1996; Goodall et al., 1998; Major et al., 1998; Haggas et al., 2001; Miller et al., 2003). Here, we investigate the validity of the orientation of FMS images in highly magnetized formations having potentially higher natural remanent magnetization (NRM 1 A/m) and/or magnetic susceptibility (MS 10–4 SI) than sedimentary formations (NRM 10–1 A/m; MS < 10–4 SI) for which the FMS was primarily developed.
Correction and orientation of FMS images are based on accelerometry and magnetometry measurements made with Schlumberger General-Purpose Inclinometry (GPI) modules such as the GPI Tool (GPIT) or the earlier GPI Capsule (GPIC) (see Robinson, Von Herzen, et al., 1989). The GPI modules use three-axis acceleration measurements to detect horizontal (shocks) and vertical (stick-slip and incomplete compensation of ship heave) movements. Three-axis magnetometers measure the intensities of the horizontal (Fx and Fy) and vertical (Fz) components of the local magnetic field (Fig. F1). From these components, intensity (F), declination (dec), and inclination (inc) of the total field in the tool frame are given by the following equations:
where Fh = the horizontal component of the total magnetic field measured in the tool frame and is defined as
Actually, outputs of the Schlumberger GPIT are the raw Fx, Fy, and Fz components and the smoothed with depth total field (FNOR, filtered version of F), tool inclination (FINC, filtered version of inclination), and the tool azimuth (P1AZ, deduced from the comparison of the filtered version of declination with the locally tabulated declination at the site [DEC]) (see Table T1 for details on notation). This smoothing (filtering) operation is referred hereafter as the "Schlumberger algorithm."
Thus, the GPI modules measure the properties of the local magnetic field, from which it is possible to deduce the orientation of the tool relative to north. When the field direction can be assumed to be constant, measured changes in the relative orientation of the field imply changes in tool rotation (Fig. F1). In particular, the rotation of the tool about its vertical axis can be measured relative to the component of the tabulated local geomagnetic field (Barton, 1997) in the plane perpendicular to the tool axis (Fig. F1C). In this case, the Schlumberger filtering operation is a data smoothing operation that reduces noise (high-frequency fluctuations). If the local magnetic field is not constant with depth but depends on the magnetic properties of the surrounding rocks, then this simple procedure may not work. In this case, the magnetometer's readings are not a simple linear function of the tool rotation but are a complex function integrating local magnetic direction variation with depth (Fig. F1D). In other words, the approximation that the total measured field is close to the international geomagnetic reference field (IGRF, see www.ngdc.noaa.gov/IAGA/vmod/igrf.html) is commonly valid for sedimentary formations but may break down and lead to incorrectly oriented FMS images for highly magnetized formations (e.g., oxide-rich layers) such as those drilled during ODP Legs 118, 176, and 197.
In connection with investigations of the physical and structural properties of formations containing alternating oxide-rich layers (Einaudi et al., in press), we present a methodological study checking, for the first time, the validity of the Schlumberger algorithm in such an environment. This algorithm was initially developed for sedimentary formations and is routinely used during ODP legs independent of formation type. First, based on data from subvertical ODP Hole 1203A (Leg 197), we investigate the frequency nature of the vertical and horizontal accelerations of the traditional ODP FMS-sonic logging tool string. These accelerations, respectively, reflect vertical and horizontal displacements of the tool. Based on the intrinsic low-frequency nature of the tool rotation, we propose to isolate the rotational component of the tool from the raw local magnetic field records using a low-frequency filtering algorithm. In order to prove the validity and efficiency of this simple filtering operation, we take advantage of the fact that an oriented borehole magnetometer, developed and designed by the Geophysical Institute of Göttingen, Germany (Göttingen Borehole Magnetometer, GBM) (Steveling et al., 1991), was deployed during Leg 197. This magnetometer is equipped with an optical gyro (Tarduno, Duncan, Scholl, et al., 2002) and provides direct and independent records of (1) the local total magnetic field and (2) the tool rotation by measuring its angular rotation rate with depth. Thus, with this data set, the tool orientation determined using a filtered version of the raw magnetic logs (magnetometer-based tool orientation) sensitive to potential magnetic anomalies can be compared with the true (optical-gyro based) tool orientation measured by the gyroscopic system. In the second step of our demonstration, we compare our previously validated algorithm with results obtained from the Schlumberger algorithm and validate the latter for highly magnetized (hard rock) environments. As additional proof of the validity of the latter algorithm for such formations, we finally compare the two discussed algorithms using data acquired from subvertical Hole 735B logged during Legs 118 and 176.
