Quasi-continuous depth profiles of rock magnetization from magnetic logs in the HSDP-2 borehole, Island of Hawaii



[1] A quasi-continuous magnetic log has been obtained in the Hawaii Scientific Drilling Project 2 (HSDP-2) between 600 m and 1800 m, which corresponds to a time interval of approximately 350 ka to 480 ka. A tri-axial borehole magnetometer was employed to measure the horizontal and vertical magnetic fields. Measurements were taken in downhole and uphole runs, with a good correlation between the two. In a first step the logs were corrected for the transfer function of the employed low-pass filter and then for the logging depths. To calculate rock magnetizations from magnetic components, we used a multidisk cylindrical model for the penetrated rocks. The disk thickness corresponds with 0.1 m to the logging sampling rate. Magnetic borehole logging in the HSDP-2 hole has established the following: Massive lava flows can be distinguished from those with prevailing hyaloclastites and enables us to supplement the lithology, especially in depth intervals with poor core recovery. The inclinations of rock magnetization derived from the magnetic log agree well with those measured in core samples from HSDP-2 hole. The same applies to magnitudes of magnetizations from logging, as the sum of induced and remanent magnetizations, with laboratory determinations of the remanent magnetizations of core samples. We observe a distinct discrepancy between the local present-day geocentric axial dipole (GAD) inclination of 35.6° and the mean logging inclination of 22.7°. Furthermore, a systematic inclination decrease with depth is observed. Logging and core inclinations in HSDP-2 can be brought into agreement with core inclinations in the HSDP-1 pilot hole by shifting depths of HSDP-1 100 m downward. The correlation of inclination data between the two boreholes and the known age-depth relation of HSDP-1 is used to reexamine the “age versus depth” model curve for Mauna Kea of Sharp et al. (manuscript in preparation, 2003). We identify tentatively two logged inclination minima with excursion of the geomagnetic field in Brunhes, namely, “Levantine” at 360–370 ka and “Unknown” at 400–420 ka.

1. Introduction

[2] Oceanic volcanic islands play a prominent role in many geological and geophysical hypotheses. The Hawaii Scientific Drilling Project (HSDP) at Hilo on the Island of Hawaii was conceived in order to study the nature and evolution of a mantle plume hot spot by drilling through the flank of an active ocean island volcano [DePaolo et al., 2001]. Five shield volcanoes form the Island of Hawaii. It has currently two highly active volcanoes - Mauna Loa and Kilauea - and one volcano (Hualalai), that is much less active. The Mauna Kea volcano, which makes up the Northeastern part of the Island of Hawaii, was chosen for study by the HSDP because it has very recently become extinct. By drilling into the volcano, the record of its activity can be extended back in time. As these volcanoes grew by accumulations of lava from summit and rift eruptions, their associated structures have been buried by flows. The current location of the plume core is postulated to be a little southwest of the summit of Kilauea volcano. The drillsite at the city of Hilo is on the east flank of the Mauna Kea volcano, which is the youngest Hawaiian volcano that has completed its life cycle.

[3] Paleomagnetic studies on the Island of Hawaii have established that Hawaiian basalts record the ambient field direction extremely well and that all flows exposed on the surface were deposited during the Brunhes normal polarity chron [Doell and Cox, 1963]. Lava ages vary from about 1 ka at the surface to about 630 ka at 3100 m depth (W. Sharp et al., Dating the growth of Mauna Kea, manuscript in preparation, 2003) (hereinafter referred to as Sharp et al., manuscript in preparation, 2003). Assuming that the original thermal remanent magnetization is preserved, the remanent parts of rock magnetization as obtained from core samples and extracted from our magnetic logs provide a detailed record of magnetic field changes versus time. But volcanic records are not as continuous as sedimentary records and may miss excursions of the geomagnetic field, which are often less than a few thousand years in duration.

[4] Published inclinations of the paleomagnetic field are derived mostly from core analyses in the laboratory. Magnetic borehole logging yields a vertical profile of the magnetic field in short time and, with a particular set of assumptions, magnitudes and inclinations of rock magnetization, as they are stated at the end of section 2. Thus the HSDP project facilitates an efficient comparison of results from laboratory and logging, respectively.

[5] A deficiency of magnetic logging in its present state is the lack of knowledge about the orientation of the horizontal components because the magnetometer usually spins round its vertical axis during the log run. Therefore only horizontal and vertical magnetic field components can be derived. Especially in a crust with strong magnetization it is impossible to determine the orientation of the spinning magnetic sensors from measuring the magnetic main field components, because magnetic anomalies prevent their orientation with respect to true north. Therefore, the following magnetic logs yield information about variations of the magnetization magnitude and inclination but, no information about variations of its declination with depth. A gyro upgrading of our borehole magnetometer will overcome this deficiency in the near future.

2. Tool Description, Logging Procedures, and Data Processing

[6] The Goettingen Borehole Magnetometer (GBM) was originally designed for geomagnetic induction studies, applying the vertical gradient method in the German Continental Deep Drilling Program (KTB) to investigate the electrical conductivity of crust and upper mantle [Steveling et al., 1991; Spitzer, 1993]. The GBM can be employed in two modes: First, the tool, locked in a particular depth, is optimized in recording time series of geomagnetic variations with high sensitivity when the main field is compensated. Second, the GBM can also be employed without compensation to log the static magnetic field in a borehole, as in the now to be described HSDP-2 campaign.

[7] The tool (Figure 1) contains a three-axial fluxgate magnetometer, type Foerster IDA 1.752, with ±100 μT range and 12.2 nT resolution. This wide range was selected to prevent overrange in huge magnetic anomalies. A possible range bisection leads to 6.1 nT resolution. Two pendular inclinometers yield the tilt angle in the borehole. The magnetic field data are low-pass filtered with a cutoff at 0.35 Hz. All analog signals are A/D converted and transmitted to the logger at the surface using a 1 Hz sampling rate. Sensors and electronics are inside a non-magnetic housing (Monel, Schlumberger ECH-MSA). The total length and diameter are 3.25 m and 85.7 mm (3 3/8 inch), respectively. The tool fits to a 31 pin Schlumberger cable head and can be operated up to 70 MPa pressure and 100°C temperature. It was run in combination with two non-magnetic bronze centralizers to keep the magnetic sensors aligned with the borehole axis.

Figure 1.

The Goettingen Borehole Magnetometer (GBM).

[8] Measurements were carried out in the HSDP-2 borehole on July 5th, 1999. As preparation the GBM was set up vertically and oriented against geomagnetic north at the drilling site to check the equipment and to determine the components of the main geomagnetic field. In Table 1 the measured fields (GBM) are compared with the International Geomagnetic Reference Field IGRF95 (National Data Centers, NGDC). The site position was taken from a GPS receiver and estimated to 19° 42′ 40″ latitude and −155° 3′ 20″ longitude. The magnetic field values are extrapolated for 7/5/1999 and calculated for the surface and 1 km depth. The site's geocentric axial dipole (GAD) inclination amounts in this epoch to 35.6°.

Table 1. Geomagnetic Main Field Components From IGRF95, at the Surface and 1 km Below the Surface, and From GBM at HSDP-2 Site
 IGRF95, 0 kmIGRF95, −1 kmGBM, 0 km
D0 (Declination)10° 03′10° 03′ 
I0 (Inclination)36° 39′36° 40′36° 24′
H0 (Horizontal)28157 nT28169 nT28720 nT
X0 (Geogr. North)27726 nT27738 nT 
Y0 (Geogr. East)4911 nT4913 nT 
Z0 (Vertical)20955 nT20965 nT21140 nT
F0 (Total)35099 nT35115 nT35660 nT

[9] The distinct difference between the IGRF and the GBM should not be due to uncertain scale factors of the GBM, since its sensors were calibrated carefully in the Goettingen geomagnetic observatory before use in the HSDP. The difference may be explained plausibly by anomalous magnetic fields within the strongly magnetized island itself, but we are unaware of any local geomagnetic surveys prior to drilling. An aeromagnetic map of Hawaii [Hildenbrand et al., 1993] shows a weak 100–300 nT (positive) local anomaly in the site's vicinity.

[10] A two day logging program in HSDP-2 was carried out in the open hole following the cessation of reaming to a caliber of 8 equation image inch. The hole was open down to 1810 mbsl (meters below sea level) and the GBM was put down to this depth. At that time the drillhole had a 9 5/8 inch steel casing extending from the surface down to 592 mbsl. The recordings within the cased section are partially out of range, due to the magnetization of the casing, and are not displayed.

[11] About 45 kilometers away from HSDP-2 site near the village “Volcano” a temporary magnetic observatory was set up to check for time dependent geomagnetic variations. This allows for a separation of temporal and spatial variations in case of high geomagnetic activity during logging. However, the geomagnetic activities remained low and there was no need for any corrections.

[12] Due to its former task our GBM system embodies no depthtometer, but throughout the experiment we run two different data loggers simultaneously. First, the GBM system logged the magnetic field, the borehole deviation from vertical and the time. Second, a logging system of the GeoForschungsZentrum GFZ in Potsdam recorded the depth information supplied by the depth measuring reel and the time. The two system timestamps enabled a correlation of GFZ depth information with our magnetic field observations and with the borehole deviation. The logging speed was 18 m/min within the cased section and 6 m/min in the open hole, which yielded a vertical depth resolution of 0.3 m and 0.1 m, respectively, for the fixed 1 Hz sampling rate.

[13] Despite using centralizers during downhole logging the tool rotated about 20 times around its axis. We measured the magnetic field components X, Y, Z in this rotating system, yielding in equation image the horizontal intensity, provided that the borehole axis is vertical. Otherwise a correction has to be applied, as will be discussed later. The tool's inclinometers recorded deviations of less than 1° within 600–1500 mbsl but they increased further down to 2.8° at 1800 mbsl.

[14] We acquired one continuous downhole log from 69–1808 mbsl and two single uphole logs from 587–998 mbsl and from 1501–1808 mbsl, respectively. In Figure 2 a raw data plot from uphole and downhole exemplifies the very satisfying reproducibility of magnetic logging. The obvious 0.6 m depth offset is the maximum observed offset while logging. All following results are derived from the downhole log only.

Figure 2.

Horizontal (H) and vertical (Z) magnetic flux density components from logging. Raw data from uphole (red) and downhole (green) logs are compared without any depth correction. A removal of a 0.6 m depth offset between uphole and downhole logs yields a nearly perfect match in both components.

[15] Prior to the conversion of magnetic logging data to magnitudes and inclinations of rock magnetization, some data processing is needed. First, the magnetic log is corrected for the effect of the phase shift and amplitude of the transfer function of the low-pass filter as a function of frequency. Especially, the phase affects the depth information in the order of 0.2 m. The correction is achieved by deconvolving the magnetic log with the response function of the filter transfer function.

[16] A general well-logging problem is the discrepancy between driller's depths, such as for lithology, and logging depths. In a first step we plotted lithological columns and intervals, where susceptibility values were available with high depth resolution from determinations on cores [Kontny et al., 2003], versus our magnetic log. In many cases, especially in the submarine sequence, lithology correlates with susceptibility and furthermore susceptibility correlates with magnetic logging. Core related depths are driller's depths and the above outlined correlations allows fitting of logging to driller's depth. Smoothed over the total logging interval we detected a shift between the two different depth logs, which increases with depth and may be due to stretching of our logging cable. In order to fit the logging depth to the driller's depth, Δd = 4d2 · 10−7m was substracted from our original logging depth d. The maximum correction is Δd = 1.4 m at 1800 mbsl depth.

[17] The thus corrected magnetic components H, Z and the total intensity equation image are presented in the left part of Figure 3. They are shown in comparison with the IGRF95 elements H0, Z0 and F0 (see Table 1). In general, the logs can be subdivided into quiet and highly fluctuating sections. In quiet sections the field strength is close to the surface values. In fluctuating sections the horizontal field usually is greater (H > H0) and the vertical field smaller (Z < Z0) than the IGRF95 surface values. This is as it should be in vertical holes drilled through rocks with positive downward components of magnetization.

Figure 3.

Horizontal (H), vertical (Z) and total (F) magnetic field components from logging with the IGR95 reference field H0, Z0, F0, as vertical lines. The magnetization shown on the right is calculated by conversion of the anomalous fields ΔH = HH0 and ΔZ = ZZ0 (see section 3.2). The determination of the inclination in quiet sections (e.g. 1670–1730 mbsl) is poor. The most right column displays the magnetic field vector in magnitude and inclination.

[18] The boundary between subaerial/submarine (SSB) lava flows was encountered at 1079 mbsl [Hawaii Scientific Drilling Project 2, 2000]. In our logs the SSB separates high amplitude fluctuations in the subaerial part from small amplitude fluctuations in the submarine section. The strongest deviation from the IGRF95 field is observed close to the SSB at 1072.5 mbsl with ZZ0 = −18500 nT, while very small deviations were found within submarine depth intervals, e.g. 1415–1470 mbsl and 1675–1725 mbsl, respectively.

[19] In the following, we denote with ΔH = HH0 and ΔZ = ZZ0 the deviations of our magnetic logging fields H and Z from the IGRF95 fields H0 = 28157 nT and Z0 = 20955 nT, respectively. We shall refer to them as components of the anomalous field vector equation image as caused by magnetization of the penetrated rock formations.

[20] In view of the clear influence which the chosen zero reference field has upon their definition, we shall restrict our analyses to those depth ranges, in which the anomalous field exceeds 500 nT. With the approximations in the following section, connecting anomalous fields to magnetizations, this restriction implies that, in the case of ΔH, depth ranges with magnetizations below 2 · ΔH0 = 1000/(4π · 100) ≈ 0.8 A/m are excluded, even though the stated resolution of 12 nT of our measurements would allow the detection of much weaker magnetization down to 0.01 A/m. But the uncertainty about the zero reference field, together with the not yet accountable effect of deviations of the borehole axis from the vertical, limits our investigations under given circumstances to relatively strongly magnetized rocks.

3. Modeling and Conversion

3.1. Field of a Hollow Magnetized Cylinder

[21] To model the anomalous magnetic field within the borehole, it is useful to define a cylindrical coordinate system with the vertical axis z parallel to the axis of the borehole, positive down. The model consists of a uniformly magnetized cylinder with radius r1 and thickness 2h. The top of the rock cylinder is at z = −h and the base at z = +h (Figure 4). Such models are described, for example, by Bosum et al. [1988]. The cylinder has a hole of radius r0 = 0.15 m, representing the borehole, and we assume that r1 is very large in comparison to r0.

Figure 4.

Cylinder model with a vertical hole.

[22] The formulas given by Bosum et al. [1988] are converted from cgs units to SI units by introducing the factor μ0/4π. The following relations are for the horizontal and vertical component of the magnetic flux density, in T = Vs/m2, at any point z on the cylinder axis:

display math
display math


display math



magnetization (A/m)


its horizontal component (A/m) in direction of ΔH


its vertical component (A/m)


its inclination (deg)

μ0 = 4π10−7Vs/Am

free space magnetic permeability.

The negative sign in equation (2) arises from an inner cylinder with a magnetization in opposite direction, thus generating zero magnetization within the hole. For a sufficiently long cylinder (h ≫ r0), C = 2 and equations (1) and (2) reduce to ΔH = μ0Mh/2 and ΔZ = −μ0Mz and I = arctan(Mz/Mh) = −arctan(ΔZ/2 · ΔH). These useful approximations allow to estimate the magnetization and inclination directly from the magnetic log.

3.2. Conversion of Magnetic Logging Data into Magnetizations of the Surrounding Rocks

[28] As seen from equations (1) and (2), a linear relation exists between the anomalous horizontal and the vertical magnetic field components, and those of the magnetization. For the stated conversion problem we use a model, which consists of a sequence of N + 1 (N even) cylindrical disks of very large outer radius r1(r1r0), constant thickness 2h throughout, and a small central hole of radius r0. The horizontal and vertical magnetizations in the ith disk, i = −N/2, −N/2 + 1,…−1,…N/2, are Mh;i = Mi cos Ii and Mz;i = Mi sin Ii, respectively.

[29] Hence, our underlying assumptions are: (i) laterally uniform magnetizations to distances which are large in comparison to the borehole radius, (ii) horizontal interfaces where these magnetizations change with depth, (iii) measurements taken in the center of a circular drilling hole. Obviously, dipping layers affect the interpretation of magnetic logs [Gallet and Courtillot, 1989]. A uniform dip causes a shift of the magnetic components only, whereas a depth dependent dip creates variations of the ΔZH ratio and thereby variations of the inclination. But despite of some general information about slopes in the vicinity of Mauna Kea, which are shortly reviewed in section 5.2, there is too little known about the dip of the penetrated lava flows, and thus we made our model as simple as possible.

[30] The magnetic field measured at depth zi on the borehole axis is the superposition of the fields caused by the central disk at zi and a sequence of (N/2) disks above and below zi. Because of the quadratic decrease of the field with vertical distance, the influence of adjoining disks with their centers at depth zi+j diminishes rapidly with distance ∣zizi+j∣.

[31] The adapted thickness h = 0.1 m of each disk equals the sample rate during logging. With our choice N = 50 the calculations are performed in modeling sections of 5.1 m length. Hence, for equal magnetizations, the contribution of a disk at the ultimate positions zi±N/2 in relation to the disk at depth zi is r02: (N/2 · h)2 or 0.0036:1, that is even an outermost disk with 300 times stronger magnetization contributes only about one percent to the field at zi. Since magnetizations were found to lie within two orders of magnitude at the most, the chosen length of the modeling section appears to be adequate, as it has been confirmed by numerical tests with variable numbers N.

[32] At an axial point in the center of the ith disk, the field arising from the disk at zi+j, has the components

display math


display math


display math

For each component, a separate set of N + 1 linear equations

display math

is then formulated with equation image = (gij) as a quadratic (N + 1) · (N + 1) coefficient matrix of geometric factors, with equation image as model vector for (N + 1) unknown magnetizations, and with equation image as data vector for (N + 1) given magnetic fields. The conversion problem is solved in the following way:

[33] The total log is subdivided into 5.1 m long overlapping sections and within each section the depth zi of measurement is moved downward from top to bottom, yielding N + 1 measurements ΔHi or ΔZi. The first section, for which the magnetizations are found from equation image, starts 2.5 m below the top, and the last section ends 2.5 m above the bottom. Because equation image is a diagonal dominated matrix the determination of its inverse is a well-posed problem, which has to be calculated only once for all sections. Whenever for a given section the depth zi+j lies outside of the upper or lower bounds of the section, the magnetizations are assumed to be known and, after multiplication with the appropriate geometric factors, are subtracted from the respective datum yi. For zi+j < ziN/2 · h the already determined magnetizations in the section above are inserted, for zi+j > zi + N/2 · h approximations Mh = 2 · ΔH0 and Mz = −ΔZ0 from the field at depth zi+j. These approximations are used also for the uppermost and lowermost 2.5 m sections, which are excluded from the more elaborate conversion scheme.

[34] For improved determinations an iteration process has been adopted with overlapping sections while going down. After finding the N + 1 magnetizations of a certain section, the calculations are repeated in a new section shifted downward by only 1.7 m, that is the magnetizations in the lower two thirds of the previous section are derived again and those in its central third are stored as final determinations.

4. Correlation With Lithology and Other Data Sets

[35] The subaerial/submarine transition at 1079 mbsl depth is marked by the disappearance of aa, pahoehoe and transitional lava flows and the appearance of massive flows and hyaloclastites. Pahoehoe and aa are Hawaiian terms to describe the common lava types found on Mauna Loa and Kilauea. Aa lava flows show extremely irregular surfaces, usually covered by fragments of broken crust, whereas the surface of pahoehoe is smooth and continuous [Kilburn, 2000]. Hyaloclastite is a deposit consisting of fragments of volcanic glass formed by nonexplosive shattering [Batiza and White, 2000]. In the magnetic log the subaerial/submarine transition is marked by a distinct change from highly fluctuating to quiet sections. In the following five displays, each with lithology columns on the left [Hawaii Scientific Drilling Project 2, 2000] and with rock magnetizations derived from logging on the right, the major causes of these anomalies will be discussed.

[36] The first example (Figure 5) shows a clear correlation between lithology and magnitude of magnetization, that is about 4–6 A/m for massive lava flows and 1–2 A/m for hyaloclastites. Such correlations are typically in the submarine depth interval and are distinguished by the high magnetization contrast at massive/hyaloclastite boundaries. Within the depth interval 1780–1798 mbsl we can compare magnetizations M derived from logging with five NRM (Natural Remanent Magnetization) laboratory determinations at cores [Kontny et al., 2003]. The magnetizations derived from logging match the NRM values satisfactorily.

Figure 5.

Magnetization derived from magnetic logging (GBM, blue curve) shows good correlation with lithology in submarine section. Strong magnetization (about 4–6 A/m) is with massive units (blue, units U0249, U0251 and U0253), low with hyaloclastite units (red, units U0248, U0250, U0252 and U0254). Single NRM (red crosses) and susceptibility (green crosses) determinations are derived from core measurements in the laboratory. The magnitude of magnetization from logging is comparable to NRM from cores. Magnetization (from logging) and susceptibility (from cores) are well correlated.

[37] For the same depth interval Kontny et al. [2003] measured susceptibility κ along the core. A very good correlation of κ with the magnitude M of our logging magnetization was found. Inserting κ ≈ 5 · 10−3SI (from Figure 5) and the total field intensity F0 = 35099 nT (from Table 1) the induced magnetization Mi of the massive lava flow U0251 (1784–1793 mbsl) can be estimated:

display math

The observed mean magnitude for the same flow, as sum of induced and remanent magnetization, is M = 5 A/m, yielding a Koenigsberger ratio of remanent magnetization to the magnetization induced by the Earth's magnetic field of about Q = 35. This dominant remanent magnetization has been found to be characteristic of massive lava flows in general. It allows us to interpret our logging results in terms of magnitude and inclination of the remanent part of magnetization.

[38] Besides magnetization changes at lava flow boundaries, there are many cases with strong changes inside a single lava flow. A first example is the subaerial aa lava flow unit U0107 in Figure 6. Between 614.1 and 614.7 mbsl the magnetization decreases from 13 A/m to 2 A/m and half a meter below the inclination increases from about 30° to 60°. Looking at the core scan image [Hawaii Scientific Drilling Project 2, 2000] a significant discoloration in the same depth interval can be seen. This could stand for an alteration of the aa flow, or there exists a boundary between two different lava flows. Because of the strong change in inclination we consider that a long time gap exists between the two flows.

Figure 6.

Within the aa lava flow U0107 between 614.1 and 614.7 mbsl magnetization decreases from 13 A/m to 2 A/m, and half a meter below 614.7 mbsl inclination increases from 30° to 60°. In consideration of the core scan discoloration this indicates a boundary between two different aa flows at 614.8 mbsl.

[39] Figure 7 shows the strongest anomaly (25 A/m magnetization magnitude) that we have observed. The maximum in magnetization correlates with manifestly discolorations within the aa flow and a potentially open fissure. The inclination, such as in most anomalies within flows, is not affected, a remarkable observation in itself. It argues for a real fissure within a single lava flow, and not for two different flows. We plotted inclination versus intensity, but there was no correlation between inclination and intensity of magnetization.

Figure 7.

The strongest magnetization magnitude (25 A/m) is encountered at 791.7 mbsl within the aa unit U0132. The core scan shows a potentially open lava flow fissure and discolorations at this depth. The inclination is not affected significantly by the fissure.

[40] The next displayed interval 1544–1553 mbsl (Figure 8) contains segments with poor core recovery and hence core scan images are missing. We observe magnetization spikes at 1547.7 mbsl and 1549.2 mbsl and strong inclination changes and do not find the expected massive flow coincidence typical for submarine section. In view of our logging data we relocate tentatively unit U0217 two meters upward to achieve better correlation between lithology and magnetization.

Figure 8.

According to lithology, unit U0217 is a massive lava flow between two hyaloclastite units U0216 and U0218 in the depth range 1548.7–1551.8 mbsl. But in U0217 no correlation to changes in magnitude or inclination of magnetization is observed. Because of poor core recovery within this depth interval core scans are missing. Thus magnetic logging suggests a 2 m upward shift of the massive flow to obtain better correlation between lithology and magnetic logging.

[41] The last detailed logging example (Figure 9) shows three successive hyaloclastite units U0242 to U0244. Remarkable is the strong anomaly within the central unit U0243. An interpretation is difficult because there exist no solid core, but only many core fragments. From the magnitude of magnetization we would consider an intrusion of massive material at 1739 mbsl.

Figure 9.

The central unit U0243 of the hyaloclastite sequence U0242–U0244 is with up to 7 A/m magnetization. This high value is unusual for hyaloclastites. Because of poor core recovery classification of U0243 is difficult. From magnetic logging an intermediate massive flow is suspected. Inclination is only well determined in high magnetized intervals, thus the very low inclinations at 1738.6 mbsl and below 1741.2 mbsl are uncertain. In any case the stable inclinations are remarkable, where magnitudes are high around 1739.3 mbsl.

5. Inclination Studies

5.1. Frequency Distribution of Magnetization and Inclination

[42] In Figure 3 we have seen very variable magnitudes and inclinations of rock magnetization versus depth. To obtain an improved overview about magnetizations of logging interval 594–1806 mbsl, magnitude and inclination of magnetization are presented as normalized frequency distributions in Figure 10. Inclinations derived from weak anomalous horizontal fields with ∣HH0∣ < 500 nT are not considered, because they are not well determined. After removal of these low amplitude anomalies the data set contains 12120 magnitudes and 9858 inclinations.

Figure 10.

Frequency distribution on rock magnetization (top magnitude, bottom inclination) derived from magnetic logging. The distribution consists of mean values averaged within 10 m depth intervals without considering different lava flows. Mean values are 3.5 A/m magnitude and 22.7° inclination. The mean inclination is 12.9° below the geocentric axial dipole inclination of 35.6° (vertical line). There is no evidence for an inverse geomagnetic field polarity.

[43] Magnitudes are classified in 30 groups on a logarithmic scale spanning a 0.1–100 A/m range. At 5–6 A/m we observe the predominant class, 3.5 A/m is the mean value of the distribution. The distribution does not look symmetric, that is classes with low magnitudes are more occupied than classes with high magnitudes.

[44] In contrast the frequency distribution of inclinations is symmetric with regard to the most occupied central class with 20°–25° inclination. The mean value is 22.7° and thus 12.9° less than the GAD inclination, which is 35.6° for the borehole site. The inclinations calculated from magnetic logging are dependent on the chosen reference field. If we select H0 and Z0 from GBM measurements at the Earth's surface (Table 1) we obtain 27.8° as mean inclination of rock magnetization. This value is closer to, but still definitely lower than the GAD inclination. Only few negative inclinations are observed, thus there is no evidence for an inverse geomagnetic field polarity.

[45] The 30° scatter of inclinations, as visible from the half width of the inclination diagram (Figure 10), does not seem to support the often cited hypothesis of low secular variations in the Pacific area during the Brunhes epoch. Thus we can agree with the statement of McElhinny et al. [1996] that “the hypothesis of the Pacific dipole window may confidently be rejected.”

[46] Figure 10 does not consider the effects of different lava flow types. To study an influence of lithology on magnitude and inclination of rock magnetization, we reclassified our data in intervals with aa lava, pahoehoe, transitional, massive and transitional, respectively. To prevent an overbalance of lava flows of extreme thickness, a weighted average is used with weights w = 1/d (d = thickness of lava flow).

[47] Depending on the five different flow types from above the new magnitude and inclination frequency distributions are assembled in Figure 11. The magnetization magnitudes of subaerial flows aa, pahoehoe and transitional and of massive submarine flows are similar with maxima at about 4–8 A/m, whereas there are three hyaloclastite maxima at 4 A/m, 1.5 A/m and 0.8 A/m. If we split up the submarine section into 2 depth intervals, lower magnetization magnitudes of hyaloclastites are mostly observed in the lower (>1670 mbsl) submarine section.

Figure 11.

Frequency distribution of magnetization for subaerial (aa, pahoehoe, transitional: solid lines) and submarine (massive, hyaloclastite: dashed lines) lava flow types. The magnetization magnitudes of all subaerial flows and of massive submarine flows are similar with maxima at about 4–5 A/m, whereas there are two hyaloclastite maxima at 3 A/m and 1.2 A/m. The inclinations of aa, pahoehoe and transitional flows are similar with maxima between 25° and 30°, hyaloclastite around 22°, and massive with two peaks at 22° and at 8°. The inclinations of submarine (older) units are less than inclinations of subaerial (younger) units.

[48] The frequency distribution curves of inclination are spiky, but different distributions are observed for subaerial (solid lines) and submarine lava flows (dashed lines). The subaerial flows (aa, pahoehoe and transitional) are similar with maxima around 30° inclination, whereas the submarine flows (massive and hyaloclastite) show inclination maxima around 20°. There seems to be a tendency that the inclinations of submarine and thus older units are smaller than inclinations of subaerial and thereby younger units.

5.2. Indications for a Decreasing Inclination With Depth

[49] To verify this tendency, the inclinations are averaged now over single lava flow units. Averaging has been performed by the “inclination only method” [McFadden and Reid, 1982] with a computer program named “incfish” [Tauxe, 1998], assuming a Fisher distribution of inclinations. To be free from effects around flow boundaries, 0.3 m depth intervals above and below are omitted. Consequently only flows thicker than 0.6 m are considered.

[50] The mean magnitudes and inclinations for any flow unit are assembled in Table 2; inclinations versus depth are shown in Figure 12. As already stated, the mean inclination is clearly smaller than the 35.6° GAD inclination. Obviously this shift cannot be removed by the choice of a slightly different geomagnetic reference field to define the anomalous field as shown above.

Figure 12.

Inclinations averaged over single lava flow units versus depth (scale left) and age (scale right) from magnetic logging. Ages are from Sharp et al. (manuscript in preparation, 2003). Different flow types are indicated by different colors. Green vertical line indicates the geocentric axial dipole inclination of 35.6°.

Table 2. Magnetization of Lava Flow Units U0104–U0256a
UnitDepth Interval, mbslLithologyMagnitudeInclination
nM, A/mS.D., A/mError, A/mnI, degS.D., deg
  • a

    M, mean magnitude; I, mean inclination (Fisher distribution [Tauxe, 1998]); n, number of values; S.D., 95&amp;percnt; deviation; Error, error of mean value; 999.9 denote indeterminable values. Depth intervals 0.3 m above and below unit boundaries are not used and inclinations I derived from weak anomalous horizontal fields with ∣H − H0∣ < 500 nT are not considered.


[51] Additionally, there is a decrease of the inclination with depth in the submarine section, which probably cannot be explained as secular variation, and also not by the increase of the main field with depth. With respect to the surface the horizontal field H0 rises only by 12 nT and the vertical field Z0 only by 10 nT, respectively, at 1 km depth (see Table 1).

[52] The effect of borehole deviation on inclination has not been considered yet. For simplification we assume an angular deviation φ from the vertical in the HZ plane. To eliminate the deviation's influence on the measured H and Z components we rotate the data in the HZ plane by φ. The intuitive assumption, that this rotation affects the calculated inclination of rock magnetization in the order of φ is wrong, because the crucial ratio to calculate the inclination is not Z/H but ΔZH. Especially for the case of low anomalous horizontal magnetic fields ΔH, a rotation by φ may yield modifications in inclination far in excess of φ.

[53] From our own measurements the slant angle of borehole deviation is known, but not the direction in reference to geographical coordinates. If we rotate the data by the slant angle, and this is the worst case, the subaerial section inclinations are affected by not more then 0.5 degrees because these are sections of the strong anomalous magnetic field. For the submarine section with weak anomalies, inclinations can be modified up to 5 degrees. Because of the tool orientation is unknown we have abstained, however, from any slant-correction of our measured fields. Thus a small portion of submarine inclination trend plotted in Figure 12 may be deviation induced.

[54] We already noted in section 3.2 that the inclination is affected by a nonzero dip of the magnetized layers while cooling down of the lava flows. Estimates about layer dippings are maintained from the slopes of Mauna Kea, investigated by Mark and Moore [1987]. The lower subaerial slopes on the east side of Mauna Kea are 6° to 9°. The submarine slopes are about two times steeper than the subaerial slopes. The dipping direction is radially outward from the summit of the volcano and therewith S75°E at the HSDP-2 drillsite.

[55] However, the lava flow slopes may be altered even after cooling down. First, the Hawaiian Ridge is the site of numerous landslides caused by the steep slopes. When sliding downward the tilt of the layers is often modified additionally. Second, a geotectonic tilting of the HSDP lava flow units, increasing with depth and time, caused by a flexural deformation beneath the Hawaiian volcano chain is noted. According to Moore [1987] and Wessel [1993] the Hawaiian Islands represent some of the largest volcanic loads on the seafloors resulting in regional deformation of the lithosphere beneath them. The tilting would be approximately to the southeast in the range of 5° to 7°.

[56] Several of the effects described here have to act similarly to explain both the downhole inclination trend and the decided inclination discrepancy up to 200 at 1800 mbsl depth. To explain the trend, a dip continuously changing and extending to about 600 m depth would be required, which seems to be unlikely.

[57] Additionally to the depht scaling on the left, an age scaling is inserted on the right of Figure 12. The ages are based on a model depth-age curve from Sharp et al. (manuscript in preparation, 2003). For more details about age estimates we refer the reader to section 5.4.

5.3. HSDP-2 Inclinations From Cores and From Magnetic Logging

[58] The good core recovery and the dense quasi-continuous magnetic logging in HSDP-2 borehole permits a comparison of paleomagnetic laboratory studies on cores with our logging results, concentrating on inclinations. Even though it would have been preferable to restrict our determinations of inclination to the interior of lava flows to avoid edge effects of various origins, layer boundaries are not sufficiently well documented throughout lithospheric logs to make such a restriction viable. We tested both, an inclination average over different lava flows and over equidistant depth sections without considering boundaries between different lava flows, whereby we mostly found no remarkable discrepancy. In cases of thick flows and suspecting that not all flow boundaries have been detected, averaging over equidistance is preferred. Readers interested in mean inclinations averaged over stratigraphic units will find the values in Table 2.

[59] Thus to obtain mean values, the logged depth interval 600–1800 mbsl has been subdivided into equidistant sections of 5 m length each. With 0.1 m sampling rate, each section contains 50 values. Mean values and standard deviations of each section in the depth range 550–1050 mbsl are plotted as red line with error bars in Figure 13.

Figure 13.

Comparison of inclinations derived from cores [Laj and Kissel, 1999] and from logging. Inclinations from magnetic logging (red line with error bars) are values averaged over 5 m depth intervals. Inclinations from cores are determined after Thellier-Thellier experiments (green line) and thermal demagnetization (blue crosses): The “logging inclinations” are in good agreement with the “core inclinations”.

[60] Inclinations derived from HSDP-2 cores up to 1050 mbsl depth are determined by C. Laj and C. Kissel (personal communication, 2001) in their paleomagnetic laboratory. They investigated inclinations both from Thellier-Thellier experiment and from thermal demagnetization. These preliminary inclinations, among them some isolated determinations, are shown in Figure 13 also.

[61] There is no obvious discrepancy between the few inclination values after thermal demagnetization and the Thellier-Thellier experiment, respectively, thus we compare logging results with Thellier-Thellier experiment inclinations only. The consistence of inclinations from logging and from cores is very satisfying, particularly in the depth range 750–870 mbsl. Discrepancies at 880 mbsl and 895 mbsl may reflect the uncertainty connected to single Thellier-Thellier experiment determinations.

[62] For logging purposes two results are important: First the fluctuation ranges of “logging inclinations” and of “core inclinations” are in agreement, and, Second, their absolute values do not differ greatly. The second observation supports our choice of the geomagnetic reference field IGRF95 in deriving magnetizations from logged horizontal and vertical fields. There is no offset between “logging inclinations” and “core inclinations”.

5.4. Age-Inclination-Depth Relations From Inclinations in HSDP-1 and HSDP-2

[63] Age-depth relations of Mauna Kea volcano have been published for HSDP-1 borehole by DePaolo and Stolper [1996]. For the HSDP-2 borehole they will be the subject of an article by Sharp et al. (manuscript in preparation, 2003). The HSDP-1 model of DePaolo and Stolper is based on Mauna Kea's lava accumulation rate, and its subaerial fraction is validated by Ar-Ar and K-Ar ages by Sharp et al. [1996]. For the HSDP-2 borehole, Sharp et al. (manuscript in preparation, 2003) proposed a linear depth-age curve using both HSDP-1 and HSDP-2 Ar-Ar ages and the lava accumulation rate 9.1 ± 3.4 m/ka from 382 mbsl to the bottom of the hole. Ages for samples in this depth section are estimated using Age(ka) = 286 ka + Depth (m) * 0.11 ka/m. Both the model of DePaolo and Stolper [1996] and the model of Sharp et al. (manuscript in preparation, 2003) are displayed within the depth interval 580–1080 mbsl in Figure 16. The model curves are close together at the top of the depth interval and differ by about 25 ka at the bottom.

[64] Magnetic logging provides no opportunity on its own for age determinations of the penetrated rocks. But it affords the comparison of observed patterns, e.g. the depth dependence on inclinations, with well dated inclinations from other boreholes. Thus with the assumptions, First, of a proper correlation between the two different holes' inclination data, and, Second, of a reliable age determination of the reference inclination data, an indirect age estimate of magnetic logging data is possible. We indicate three different inclination data comparisons with such “age determination” facilities in the following.

[65] HSDP-1 is a pilot borehole about 2.1 km north of HSDP-2, penetrating 43 lava flow units of Mauna Loa and then from Mauna Loa/Kea contact at 279.5 m penetrating 184 Mauna Kea flows to 1056 m total depth [Stolper et al., 1996]. Laj and Kissel [1999] published tables of depths, ages and inclinations from paleomagnetic HSDP-1 core measurements. They connect the 126 lava flow units within the depth range of 500–1050 m to the age interval 340–418 ka, yielding a mean time interval of 619 years between successive flows. Considering the age model of Sharp et al. (manuscript in preparation, 2003) within 352–413 ka for 600–1150 mbsl and the 82 lava flow units within, we get 743 years as the mean time interval between successive HSDP-2 flows.

[66] The different intervals between successive HSDP-1 and HSDP-2 flows are indications that different sequences of lava flows have been penetrated. This fact complicates comparisons, but nevertheless a mean time interval of about 1000 years should be sufficient to identify geomagnetic secular variations on a timescale of, say, 10,000 years. In other words, despite of different flows in both boreholes the longer wavelength magnetic pattern can be expected to be similiar.

[67] Inclinations of the two HSDP boreholes, both versus depth and versus age, are compared in Figure 14. The HSDP-2 “logging inclinations” in the left part of Figure 14 are mean values averaged over 10 m depth intervals. The displayed depth section resp. age section, is 500–1150 mbsl and 341–413 ka, whereas ages are calculated by the linear age/depth-relation (Sharp et al., manuscript in preparation, 2003), mentioned above. HSDP-1 “core inclinations” and associated ages [Laj and Kissel, 1999] are plotted on the right of Figure 14. Laj and Kissel's ages are interpolated ages from radiometric datings of HSDP-1 samples.

Figure 14.

Comparison of “logging inclinations” from HSDP-2, on the left, with “core inclinations” from pilot borehole HSDP-1 [Laj and Kissel, 1999] on the right. In the display depths for HSDP-1 data have been shifted 100 m downward to achieve the best possible fit between the data sets. HSDP-2 ages are from Sharp et al. (manuscript in preparation, 2003), and HSDP-1 ages are from Laj and Kissel [1999].

[68] To obtain a reasonable fit between “logging inclinations” and “core inclinations” by visual inspection, the core data set of HSDP-1 has to be shifted downward relative to HSDP-2 by 100 m. Now the two age scalings of Figure 14 match with 402 ka at 1050 mbsl HSDP-2 depth, but the ages diverge by +12 ka at 600 mbsl and −5 ka at 1150 mbsl, respectively. Asigning the HSDP-1 age scaling to HSDP-2 via the correlation of inclinations leads to the “HSDP-1/2 Correlation” curve in Figure 16. In comparison to the model of Sharp et al. (manuscript in preparation, 2003) this straight line yields a 12 ka lower age at 600 mbsl, the same age at 1040 mbsl and a lower lava accumulation rate of 7.3 m/ka. The shift in depth by 100 m would imply a 2.7 degrees upward dip in a northward direction, considering the distance between the two sites. There exist two particular flow boundaries which could verify the postulated shift in depth: first, the contact between Mauna Loa and Mauna Kea flows (ML) and second, the subaerial/submarine transit (SSB). The ML is at 279.5 m depth in HSDP-1 and at 246 mbsl in HSDP-2 [Hawaii Scientific Drilling Project 2, 2000]. This would imply a 0.9° down dip in a northward direction and thereby is in conflict with the proposed shift. The SSB has been penetrated only in HSDP-2 and not in HSDP-1, unfortunately.

[69] A more detailed presentation of inclinations from logging and from cores, including error bars, is given in Figure 15. The cyan curve [Laj and Kissel, 1999] and the dark blue curve [Holt et al., 1996] result from HSDP-1 cores, the green curve from HSDP-2 cores (C. Laj and C. Kissel, personal communication, 2001) and the red curve from our magnetic logging, with “logging inclinations” averaged over 5 m depth intervals, as in Figure 13.

Figure 15.

Inclinations of HSDP-2 magnetizations from logging (red) compared with three different inclinations from cores. “Core inclinations” from HSDP-2 [Laj and Kissel, 1999] are displayed as a green line. Two “core inclinations” from pilot borehole HSDP-1 [Laj and Kissel, 1999; Holt et al., 1996] are displayed as cyan and dark blue lines, respectively. Depths of HSDP-1 data are shifted 100 m downward against displayed depths of HSDP-2 to achieve the best possible fit between the data sets by visual inspection.

[70] The two “core inclinations” from HSDP-1 agree very well, while the good agreement between “core inclinations” and “logging inclinations” from HSDP-2 has been evident already from Figure 13. Now we observe a satisfying or even good (750–840 mbsl) correlation between HSDP-1 and HSDP-2 data. Examples for such correlations are the sinusoidal variations in inclinations between 770 and 830 mbsl, with a peak to peak amplitude of 45°, and the continuous increase of inclination from a low at 990 mbsl to a maximum at 1070 mbsl.

[71] The mean inclinations of the four different data sets displayed in Figure 15, averaged over 600–1150 mbsl, are given in Table 3. The four mean values are around 30° and match well considering the upper and lower bounds, as derived with Tauxe's [1998] incfish program.

Table 3. Mean Inclinations With Upper and Lower Bounds in HSDP-1 and HSDP-2 Boreholes, With Depths of HSDP-1 Data Shifted 100 m Downwarda
 Inclination (Mean)Inclination (Upper Bound)Inclination (Lower Bound)
  • a

    Depth range of the boreholes is 600–1150 mbsl (see Figure 14).

HSDP-1 core [Laj and Kissel, 1999]30.0°32.6°27.3°
HSDP-1 core [Holt et al., 1996]29.0°31.6°26.3°
HSDP-2 core [Laj and Kissel, 1999]28.8°32.9°24.6°
HSDP-2 logging29.7°32.9°26.5°

[72] Considering possible evidence for excursions of the geomagnetic field in magnetic logging is another interesting subject. An overview on known excursions in Brunhes, the recent epoch with normal (positive) geomagnetic polarity, has been published by Langereis et al. [1997]. The Brunhes/Matuyama transition is at 778 ka, thus the whole HSDP-2 is in Brunhes. Dominant excursions in the younger part of Brunhes are the “Calabrian Ridge 1” (315–325 ka) and the “Calabrian Ridge 2/West Eifel” (515–525 ka) excursions.

[73] Considering Figure 14, our magnetic log concerns the interval between the two Calabrian Ridge excursions which themselves are outside of our logging interval, however. Within our interval Langereis et al. list two minor excursions, “Levantine” (360–370 ka) and “Unknown” (400–420 ka). We allocate the inclination minimum at 785 mbsl (372 ka) with “Levantine” and the minimum at 990 mbsl (395 ka) with “Unknown”, even though the two ages differ by +7 ka for “Levantine” and −15 ka for “Unknown” compared to the ages given by Langereis et al. [1997].

[74] Either excursion positions are plotted in Figure 16 as ellipses, whose axis stand for the approximate excursion spread in depth and age (ages from Langereis et al. [1997]). The position of “Levantine” agrees well with the “HSDP-1/2 Correlation” graph, whereas the excursion “Unknown” is located between the models of Sharp et al. (manuscript in preparation, 2003) and DePaolo and Stolper [1996]. With the assumption that the inclination low at 785 mbsl is actually connected to the “Levantine” excursion this agreement supports the depth shift between HSDP-1 and HSDP-2 (Figure 14).

Figure 16.

Age versus depth relations for Mauna Kea. The solid lines are based on models from DePaolo and Stolper [1996] for HSDP-1 and Sharp et al. (manuscript in preparation, 2003) for HSDP-2. The dashed lines are deduced from magnetic logging, but are not independent agings. The dark blue dashed line results from a correlation procedure between HSDP-1 paleomagnetic core data and HSDP-2 magnetic logging data. The light blue dashed line is derived from a comparison of “logging inclinations” at HSDP-2 with “core inclinations” at ODP hole 1062D around an excursion at 410 ka. The two excursions detected while magnetic logging in HSDP-2 are marked by ellipses.

[75] Lund et al. [1998] published an inclination record measured on cores and samples from ODP hole 1062D at Bahama Outer Ridge. They plotted the inclination versus 59–63 meters sediment depth below seaflor, assumed a sedimentation rate of 0.2 m/ka and assigned the observed inclination low, named 11A, to an excursion at 420 ka. The shape of their 20 ka long inclination record is comparable to our “logging inclinations” around the “Unknown” excursion at 990 mbsl. The assumption of a 7.6 m/ka HSDP-2 lava accumulation rate yields the best fit between the two inclinations records. Using this accumulation rate as the slope and the “Unknown” excursion at (990 mbsl, 410 ka) as a fixed data point we obtain the short age-depth curve, marked with “Lund et al. (ODP 1062D) 1998” in Figure 16.

6. Conclusions

[76] 1. Magnetic logging in the HSDP-2, completed in two days, yielded about 12000 horizontal and vertical rock magnetizations and thereby inclinations, continuously with depth between 594 mbsl and 1806 mbsl. There are no data gaps as they may occur in connection with poor core recovery in the case of laboratory determinations. Uphole and downhole magnetic logs show a satisfactory agreement.

[77] 2. Particularly in submarine section we found a possibility to link lithology to magnetic logging results.

[78] 3. The overall magnitude of magnetizations derived from logging are comparable with natural remanent magnetizations (NRM) measured in core samples. A mean magnitude of 4–8 A/m was for aa, pahoehoe, transitional and massive lava flows. Hyaloclastites have a somewhat lower magnetization. As a general rule, remanent magnetization of aa, pahoehoe, transitional and massive lava appears to be at least ten times stronger than the induced one.

[79] 4. The mean inclination of magnetization is 30.1° ± 1.5° for the subaerial lava flows and 15.2° ± 0.7° for the submarine section. Thus, particularly submarine inclinations are significantly shallower than the 35.6° geocentric axial dipole inclination.

[80] 5. Besides strong depth- and thereby age-dependent variations of inclination, a systematic decrease in inclination with depth is observed, especially in the submarine section. The trend's basic cause is not yet clearified. The borehole's deviation from vertical, the dip of the layers, the subsidence of the Hawaiian Islands and thereby a changing slope of the Earth's crust and landslides are possible, but on their own presumably insufficient causes.

[81] 6. Inclinations from HSDP-2 logging fit well to inclinations from HSDP-2 cores. It was found, however, that a downward shift of 100 m of the pilot borehole HSDP-1 core data is necessary to obtain agreement of the two sets of borehole data. After increasing HSDP-1 depths by 100 m, inclinations from HSDP-1 cores correlate equally well with inclinations from magnetic logging in HSDP-2.

[82] 7. Besides the correlation between the two HSDP borehole inclination data, a further correlation between magnetic logging inclinations and an inclination record of about 20 ka length around 400 ka age from the Bahama Outer Ridge borehole is observed [Lund et al., 1998]. With the assumption of proper age determinations in HSDP-1 and Bahama Outer Ridge the correlations yield HSDP-2 lava accumulation rates of 7.3 m/ka and 7.6 m/ka, respectively, in the depth range 600–1070 mbsl. These values are consistent with Sharp et al.'s (manuscript in preparation, 2003) accumulation rate 9.1 ± 3.4 m/ka.

[83] 8. Reversals of the geomagnetic field were not encountered. Langereis et al.'s [1997] excursions “Levantine” at 360–370 ka and “Unknown” at 400–420 ka, respectively, are assigned to inclination minima at 785 mbsl and 990 mbsl, respectively.

[84] 9. Advantages and limitations of magnetic logging are demonstrated in this study. Magnetic logging cannot substitute paleomagnetic core investigations in laboratories, but logging is a good alternative where cores are unavailable. Important requirements are a well chosen magnetic reference field, knowledge about possible deviations of the borehole axis from the vertical and sufficiently strongly magnetized rocks. The high conformity of inclinations from logging and from cores have encouraged us to upgrade our borehole magnetometer with three Fiber Optic Gyros for orientation control. Thus, in the future we will be able to derive also the declination of magnetization. Measurements in HSDP-2 with the upgraded tool are planned after reaming of the 1800–3000 m section.


[85] We thank the three PIs of the HSDP project, Don DePaolo, Ed Stolper, and Don Thomas and furthermore Roy Wilkens, which supported and enabled our logging program in HSDP-2. Only with energetic help of Joerg Schliebe and Wilfried Steinhoff, technicians in the Institute of Geophysics Goettingen, a timely scheduling of the tools was possible. The GeoForschungsZentrum Potsdam staff Karl Bohn, Mike Hoenig, Jochem Kueck, and Miel Kuehr run winch and other facilities when logging and made sure a faultness tool recovery. We thank Carlo Laj and Catherine Kissel (CEA-CNRS, Gif-sur-Yvette) for providing us with unpublished inclination data from HSDP-2 cores. We thank Ulrich Schmucker for very helpful reviews of manuscript drafts. Reviews of three anonymous reviewers helped improve the manuscript. The project has been supported by NSF and the German Science Foundation (DFG, Ste 371/6).