FOUR CENTURIES OF GEOMAGNETIC DATA FROM HISTORICAL RECORDS

Art R. T. Jonkers,1 Andrew Jackson, and Anne Murray School of Earth Sciences, University of Leeds, Leeds, UK

Received 14 March 2001; revised 12 February 2002; accepted 19 December 2002; published 21 June 2003.

[1] We present a comprehensive review of historical dented number of early observations of the Earth
s geomagnetic data collected over the last 2 decades by magnetic field, in standardized format, of potential use many workers, culminating in the largest such compila- in many areas of geophysical research. INDEX TERMS: 1714 tion in the world. It spans over 4 centuries, from 1510 to History of Geophysics: Geomagnetism and paleomagnetism; 1560 1930 inclusive, and consists of 151,560 declinations, Geomagnetism and Paleomagnetism: Time variations-secular and long 19,525 inclinations, and 16,219 intensities. The greater term; 1532 Geomagnetism and Paleomagnetism: Reference fields (re- part of the new data set comprises unpublished obser- gional, global); 1507 Geomagnetism and Paleomagnetism: Core pro- vations recorded by mariners engaged in merchant and cesses (8115); KEYWORDS: geomagnetic data, secular variation, mag- naval shipping across the world
s oceans. Earlier pub- netic field, maritime history, navigation, historical geomagnetic lished compilations and printed accounts of voyages observations have also been included, in particular when their original Citation: Jonkers, A. R. T., A. Jackson, and A. Murray, Four centuries sources were inaccessible, missing, or no longer extant. of geomagnetic data from historical records, Rev. Geophys., 41(2), The current effort has brought together an unprece- 1006, doi:10.1029/2002RG000115, 2003.

1. INTRODUCTION [4] Furthermore, the local magnetic direction in the horizontal plane, sometimes called the magnetic merid- [2] The Earth
s magnetic field is commonly assumed ian, is denoted as H, defined by adding X and Y together. to originate in the planet
s fluid outer core. It has been The horizontal angle difference between geographic consciously observed at the surface for the past thousand north (X) and magnetic north (H) is usually written D, years or so, and geographically diverse observations are for the magnetic declination. A compass will try to align available for the last 500 years. It is likely that much can itself with H; navigators therefore have to apply a vari- be learned from an analysis of the field morphology and able correction to their compass courses to regain true evolution deduced from direct measurements. This was courses. Magnetic declination on board ship was often already realized early in the sixteenth century, when the referred to as magnetic variation; the needle was said to first global geomagnetic field models were postulated on be northwesting or northeasting a certain number of de- the basis of a handful of declination measurements, grees from true north. Last, the vertical angle between H compiled in Portuguese sailing directions. More sophis- and F is called the magnetic inclination or dip. It was ticated efforts by sundry individuals ensured in the fol- measured with an inclinometer (dip meter), which con- lowing centuries, largely based on original nautical data sisted of a magnetized needle able to rotate in the vertical [Jonkers, 2000]. Although many of these maritime plane. In the past, declination and inclination were the first sources have since been lost, a sufficiently large number measurements to be made, defining the direction of the have been preserved to warrant extracting a substantial local geomagnetic vector. Intensity (the magnitude of this sample for geomagnetic modeling purposes. vector) was first observed in a relative sense (from the [3] Geomagnetic data come in many shapes and sizes. 1790s) by comparing the “swing time” of a needle at the Nowadays, it is customary to define the field at any desired location with that obtained at a reference site. particular place and time in terms of the three compo- Absolute intensities were only recorded from the 1830s nents (X, Y, and Z) that together yield F, the full geo- onward. At present, directional data have been almost magnetic field vector (see Figure 1). completely replaced by three-component data, recorded at magnetic observatories, on surveys, and by satellites. [5] In 1839, Carl Friedrich Gauss published his classic method for the separation of magnetic fields into those 1Also at Department of Earth Sciences, University of Liv- due to internal and external sources by spherical har- erpool, Liverpool, UK. monic analysis [Gauss, 1839]. This technique was soon

Copyright 2003 by the American Geophysical Union. Reviews of Geophysics, 41, 2 / 1006 2003

8755-1209/03/2002RG000115$15.00 doi:10.1029/2002RG000115 ● 2-1 ● 2-2 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

netic data (notably from ship
s logbooks) through a thorough analysis. Third, we desired to develop a core field model that was suitable for studies of the flow at the top of the core. [7] During the 1980s and 1990s several workers en- gaged in a geomagnetic data collection effort unprece- dented in numerical, temporal, and geographical scope. The result includes the available major compilations published in the past and the largest ever sample of original manuscript sources (mostly English, French, and Dutch ships
logbooks). In particular, global mea- surements of declination (D) before 1850 have received detailed attention, as well as the gathering of early observations of inclination (I) and intensity (H and F, the magnitude of the horizontal and the full, three- dimensional magnetic vector, respectively). In addition, the several different types of relative intensity scales Figure 1. The three-dimensional geomagnetic vector F con- applied in the late eighteenth and early nineteenth cen- sisting of X, Y, and Z. Moreover, X and Y combine to form H, tury have been inventoried (see section 2.3). The com- the horizontal magnetic vector. The horizontal angle of H with plete data set (consisting of nine subsets described in X is defined as D, the magnetic declination, while the vertical section 5) comprises Ͼ187,000 geomagnetic observa- angle between H and F is called I, the magnetic inclination. tions, derived from Ͼ2000 ocean voyages. Each of these has been assigned a unique identification code (hereaf- ter referred to as ID code) and has been fully referenced applied to actual measurements of the Earth
s field, and in the table of sources (see Table 1). A histogram of data it was shown that most of the field was of internal origin. frequencies per decade over the complete interval 1510– In the ensuring years the field was reconstructed at 1930 is given in Figures 2a–2d. different epochs, the interest being primarily in the [8] Table 2 gives an indication of the data frequency evolution of global averages such as the dipole moment. spread for each measurement type per century. The Many such models have been collected by Barraclough unequal division over time is deliberate and will be [1978], who gives an extensive review. Only in the last maintained throughout this paper in Tables 3–7. The few decades has geomagnetic secular variation (the first era (up to 1589) contains the few geomagnetic change in the magnetic field of internal origin over readings taken before the advent of English and Dutch historical time) been modeled at its origin at the core- large-scale exploration of the East Indies. Moreover, the mantle boundary (CMB). Initial attempts mainly relied year 1590 constitutes the lower temporal limit of our upon published compilations of historical measure- time-dependent model gufm1 (1590–1990), and it there- ments, such as the work of Hansteen [1819], Sabine fore seemed useful to mark this boundary in Tables 2 [1872, 1875, 1877], and van Bemmelen [1899], which is and 7 as well. The next period spans the years 1590– treated in section 5. The earliest calculated main field 1699, a slightly extended seventeenth century, covering models at the CMB were epoch models, which reduced the early era of the East India Companies. The eigh- the data to a central year [Bloxham and Gubbins, 1985; teenth century (1700–1799) represents the time of these Shure et al., 1985; Bloxham, 1986; Bloxham et al., 1989]. companies
greatest expansion, as well as vast increases A more sophisticated approach was soon thereafter ap- in Atlantic and Arctic oceanic traffic. The inclusion of plied to determine time-dependent models [Bloxham the first 3 decades of the twentieth century in the last and Jackson, 1989]. However, existing compilations have period (1800–1930), which contains extensive surveys as barely scratched the surface of the number of surviving well as maritime naval and merchant traffic, is due to the original sources, and printed versions were sometimes fact that practically all data points from that last 30 years incomplete, and values were truncated, rounded, or re- stem from the Veinberg subset, which itself covers the duced to grid. Furthermore, maritime historical data late nineteenth century as well. It therefore seemed suffer from peculiar errors (e.g., the ship
s navigational more appropriate to treat this time range as a single error), which have not been adequately accounted for in entity rather than to create an arbitrary division. previous work. [9] After archival processing and initial pruning of [6] Our goals that spurred the present data collection typographical error the large majority of original voy- exercise were threefold. First, we wanted to exploit the ages has been plotted and their estimated longitudes vast amounts of geomagnetic data that remain available brought into coincidence with the identified geographi- in archival and other repositories, going as far back in cal landmarks sighted, using a dedicated piece of soft- time as sources would permit. Second, our aim was to ware we call the voyage editor. The database therefore properly quantify the errors present in historical mag- contains two fields for longitude, reflecting the original 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-3

Table 1. Sources of Dataa [The full Table 1 is available in the HTML version of the article.]

ID code Year(s) Ships and Surveys Authors DIIntensity Locations

2 1598–1600 Gelderland J. Heemskerck 3 0 0 ARA 1.04.01/51 4 1598–1600 Utrecht R. Cornelisz 12 0 0 ARA 1.04.01/62 5 1598–1600 Vriesland and Zeelandia J. C. May 35 0 0 ARA 1.04.01/53 7 1598–1599 Amsterdam and Vriesland J. Pietersz 110 0 0 ARA 1.04.01/46 8 1598–1599 Hollandia S. L. Mau 5 0 0 ARA 1.04.01/44 9 1598–1599 Mauritius C. J. Ceullen 7 0 0 ARA 1.04.01/43 10 1598–1599 Vriesland H. D. Jolinck 66 0 0 ARA 1.04.01/60 12 1599 Hollandia 13 0 0 ARA 1.04.01/45 13 1599–1600 Amsterdam H. D. Jolinck 18 0 0 ARA 1.04.01/60 18 1605 1 0 0 ARA 1.11.01.01/1137 20 1608–1609 Griffioen 3 0 0 ARA 1.11.01.01/1137 21 1611–1612 Vos and Craen J. Cornelissen 4 0 0 ARA 1.01.05/9282 26 1620 Zierikzee 19 0 0 ARA 1.10.30/309 28 1623–1624 Wapen van Delft 32 0 0 ARA 1.05.01/9290 32 1626 Hollandia W. Jansz 4 0 0 ARA 1.04.02/5050 33 1627 Wapen van Hoorn P. Roosenburg 23 0 0 ARA 1.04.02/14345 40 1636 Nieuw Amsterdam H. Brouwer 56 0 0 UBL B.P.L./127 e 41 1636 Swol 15 0 0 ARA 1.10.30/97 42 1637–1638 Wesel 9 0 0 ARA 1.11.06/3704 43 1637–1638 Wezel W. Jansz 9 0 0 ARA 1.04.02/5052 44 1640 Pauw 1 0 0 ARA 1.10.30/131 48 1643 Amsterdam 1 0 0 ARA 1.05.01.01/44 49 1643 Orangeboom C. Pietersz 11 0 0 ARA 1.05.01.01/44 53 1644–1645 Zeemeeuw 35 0 0 ARA 1.10.30/280 a 68 1655–1656 Verenighde Provintien 47 0 0 ARA 1.04.02/11408 71 1662–1663 Maarseveen M. Gerritsz Boos 78 0 0 ARA 1.11.01.01/112 72 1663 Alphen P. v Hoorn 11 0 0 ARA 1.10.45/3 77 1665–1666 Alphen M. Gerritsz Boos 1 0 0 ARA 1.11.01.01/112 78 1666–1667 Nieuwenhovenb M. Gerritsz Boos 6 0 0 ARA 1.11.01.01/112 79 1667–1668 Dijmermeer M. Gerritsz Boos 24 0 0 ARA 1.11.01.01/112 81 1669 Nusenborcht M. Gerritsz Boos 54 0 0 ARA 1.11.01.01/112 82 1670 Hollantsen Thuijn M. vd Brouck 3 0 0 ARA 1.04.02/11409 83 1670–1671 Sparendam R. Padtbrugge 20 0 0 ARA 1.04.02/5053 85 1674 Berkhout M. Kint 5 0 0 ARA 1.10.72.01/239 88 1674 Zeven Provintien J. Andringa 3 0 0 ARA 1.10.72.01/243 90 1677 Africa A. Jongekoe 33 0 0 ARA 1.04.02/5055 93 1677–1678 Asia J. Hoek 60 0 0 ARA 1.04.02/5056 94 1677–1678 Nieu Middelburgh A. Jongekoe 62 0 0 ARA 1.04.02/5055 95 1677–1678 Vrijheijt P. J. Kuijff 56 0 0 ARA 1.04.02/5054 96 1677–1678 Vrije Zee P. Sleeswijk 34 0 0 ARA 1.04.02/5057 97 1680 Hellevoetsluijs C. Bichon 34 0 0 MMPH Hs 101 1681–1682 Japan C. Bichon 30 0 0 MMPH Hs 103 1683 Negombo C. Bichon 10 0 0 MMPH Hs 108 1684 Stavenisse 41 0 0 ARA 1.11.01.01/351 109 1684–1685 Castricum C. Bichon 29 0 0 MMPH Hs 111 1685 Boswijk 39 0 0 ARA 1.11.01.01/351 113 1688 Noordholland? 15 0 0 ARA 1.11.01.01/547 114 1690 Eenhoorn A. Bichon 39 0 0 MMPH Hs 115 1690 Montfoort 11 0 0 ARA 1.11.01.01/351 116 1690 Pampus 30 0 0 ARA 1.11.01.01/552 118 1692 Sirjanslant 43 0 0 ARA 1.11.01.01/351 120 1692–1693 Walenburgh A. Bichon 8 0 0 MMPH Hs 121 1693 Baijeren 33 0 0 ARA 1.04.02/5058 122 1693 Cartago J. Speelman 1 0 0 ARA 1.04.02/5058 123 1693 Driebergen C. Coman 14 0 0 ARA 1.04.02/5058 124 1693 Eenhoorn 18 0 0 ARA 1.04.02/5058 125 1693 Faam 39 0 0 ARA 1.04.02/5058 126 1693 Juffrouw Agata 27 0 0 ARA 1.04.02/5058 127 1693 Koning William 15 0 0 ARA 1.04.02/5058 128 1693 Mijdrecht 25 0 0 ARA 1.04.02/5058 129 1693 Nigtevegt 8 0 0 ARA 1.04.02/5058 130 1693 Ridderschapc 34 0 0 ARA 1.04.02/5058 131 1693 Schoondijk L. P. Clein 39 0 0 ARA 1.04.02/5058 2-4 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Figure 2. Temporal distribution of the database records (a) 1510–1589, n ϭ 162; (b) 1590–1699, n ϭ 13,673; (c) 1700–1799, n ϭ 85,070; and (d) 1800–1930, n ϭ 78,162. Note the difference in vertical scaling; a single record may contain a landsighting and/or one or multiple types of observation.

coordinate used in the dead reckoning, and the modern and islands. These were drawn in exquisite detail using correction (see section 5.1). the coastline database from the “Generic Mapping [10] The voyage editor is an interactive graphical user Tools” plotting package [Wessel and Smith, 1991]. The interface designed by N. Barber of Leeds University. Its voyage editor combines many features in addition to the main window offers a view of the world in cylindrical- course plot proper, such as recorded comments from the equidistant projection in four magnifications (see Figure log and the number of ship days since departure 3). This allows the user to zoom in on specific regions printed next to each plotted point; color-coded dis- and examine voyage data in relation to nearby shores tance traveled per day between points; original me- 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-5

Table 2. Temporal Distribution of Geomagnetic Measurements in the Databasea

Period Records D I H F Total

1510–1589 162 160 2 0 0 162 1590–1699 13,673 12,001 53 0 0 12,054 1700–1799 85,070 68,076 1,747 0 36 69,859 1800–1930 78,162 71,323 17,723 11,404 4,779 105,229 Total 177,067 151,560 19,525 11,404 4,815 187,304 aA single record may contain a land sighting and/or up to three types of measurements (D, I, and intensity (H or F)).

ridian identified in letter code, geographical refer- [13] Jackson et al. [2000] focused on modeling issues ence, and on the chart; coordinates of data points, and introduced a new approach to the quantification of both in degrees and minutes (relative to the original the instrumental and positional error. Here we concen- meridian) and in decimal degrees (relative to Green- trate on the founding measurements and their sources, wich); selection of any range of points to become completing our account of our own researches into mar- temporarily invisible; and tracking of all modifications itime sources, as well as attempting to properly acknowl- in a separate log file. Its main function was, however, edge the substantial efforts made by our predecessors in to correct longitudinal error accumulated between the recent and distant past. We hope that this review will successive landfalls. The identification of geographical not only represent a lasting record of the data processing locations mentioned in the log was invaluable in en- itself but that it will equally serve as a guide to historical abling a reconstruction of the actual course plots; the geomagnetic data, both to prevent any duplication of voyage editor provided instant visual feedback on land processed material in subsequent studies and to inspire sightings and allowed positional adjustments of three novel inquiries along these lines in future times. kinds: (1) single-point editing, (2) translation of voyage [14] The paper has two complementary constituents, a legs, and (3) stretching of voyage legs. description (sections 2–6), and a table of the sources and [11] Single-point edits merely involved correcting in- their references (Table 1). Section 2 discusses the three dividual coordinates. Translation and stretching opera- types of measurements most commonly found in histor- tions acted on multiple points (voyage legs); a start and ical source material. A short overview of previously end point were chosen to define the leg boundaries, and published data and compilations in various forms in either end could be lifted and transported elsewhere. section 3 is then followed by a more in-depth account of This displacement could then be effected uniformly maritime manuscript sources in section 4, discussing for each point, shifting the whole leg to the right or carrier formats, repositories, and coordinate systems left while maintaining latitude. Alternatively, when used. The main description of the nine compilations that stretching, one end of a leg remained fixed, while the make up the present data set can be found in section 5, other points were longitudinally extended to meet the treated in order of occurrence in the database. Some translated other end. For example, when a Hudson
s possible directions of future research are briefly dis- Bay Company ship traveling westward from the cussed in the concluding section 6. These descriptive Orkney Islands finally spotted the tip of Greenland, sections are complemented by Table 1, containing infor- the winds and currents carrying the ship along may mation pertaining to the individual voyages in the data- have been underestimated or overestimated to such base per ID code and the processed original or pub- an extent that the reckoning put the vessel either far lished source(s) that contain the geomagnetic ahead or far before the longitude of Cape Farewell. observations. The last land previously sighted (the Orkneys) could then be fixed as a starting point, and the entry con- taining the comment “seen Cape Farewell” moved to 2. MEASUREMENTS the appropriate location. The stretch was distributed equally over time, and all longitudes were recalculated [15] Historical observations of the Earth
s magnetic to proportionally reflect the change through linear field are heavily biased toward the directional part of the transformation. For a more detailed description of the local magnetic vector; the earliest recorded magnetic voyage editor and the positional corrections made intensities useful for the present modeling studies date therewith, consult Jackson et al. [2000]. from the 1790s (see section 2.3). It was only after [12] In combination with modern observatory, survey, Gauss
s development of a method for measuring abso- and satellite data the historical database has spawned lute intensities in 1832 that the horizontal intensity (H) the first time-dependent model (called gufm1) covering and the total intensity (F) were observed. For a detailed 4 centuries (1590–1990), likewise described by Jackson et description of his instruments and method, see Malin al. [2000]. The change in the declination at the Earth
s [1982]. In more recent times, X, Y, and Z, the eastward, surface as described by this model is shown in Figure 4. northward, and downward components of the magnetic 2-6 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Figure 3. An East India voyage plotted in the voyage editor, reproduced from Jackson et al. [2000].

field, respectively, are reported. A detailed treatment of offered the advantage of stronger and longer magneti- instruments and exact definitions of the measurements zation. obtained therewith has been given by Langel [1987]. We [17] A clear distinction has to be made between a limit ourselves here to a brief overview of developments steering compass and a bearing compass. Two steering in the maritime realm. A more extensive treatment is compasses stood in separate compartments of the bin- given by Jonkers [2000]. nacle in front of the helm for the helmsman to keep a set course. Their cards usually only carried a graduation in 2.1. Declination whole, half, and quarter points in the familiar star- [16] Awareness of magnetic declination in Europe shaped pattern. A bearing compass, on the other hand dates back to the first half of the fifteenth century, when (also known as an observation compass or variation the earliest scattered land observations were made. Mar- compass), had a card divided in degrees and additionally iners in the early sixteenth century could determine the featured a mounted, sometimes rotatable, sighting ap- quantity with the aid of a gnomonic compass; it em- paratus on top of the box in order to “shoot” celestial ployed the shadow cast by a central upright style at equal bodies, landmarks, or the ship
s wake. Most importantly, times before and after noon to find the local meridian, to observation compasses served to keep track of magnetic which magnetic north was compared. In the second half declination as it changed from place to place during an of the sixteenth century it was replaced by the more ocean voyage. Measurements and past experience then familiar round or square box made of wood, copper, or provided an estimate of how much the steering compass brass. This compass “bowl” was suspended in gimbals had to be compensated for to keep a true course. and had a watertight glass lid. The needle was initially [18] Bearing compasses had no fixed place on deck; made of a lozenge-shaped piece of iron wire. By the navigators could momentarily set them up on a stool or second quarter of the eighteenth century it had been tripod as circumstances required. In its early form, largely replaced by a single, straight piece of steel, which known as an amplitude compass, the sighting mechanism 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-7 consisted of two small visors of equal length on opposite that the proponents of the dry type finally had to ac- sides of the box, which only allowed sighting in the knowledge defeat. horizontal plane, directly over the card. The 1750s, how- [22] Up to the middle of the nineteenth century the ever, saw the ascendancy in English, Danish, and Dutch majority of ships were primarily made of wood. The navigational practice of the azimuth compass (see Figure structural introduction of iron and steel in ship construc- 5). tion exacerbated the problem of magnetic deviation, the [19] This innovation had taller visors of unequal difference between measured and true declination. Nee- height, which allowed taking bearings of objects at a far dle deflection due to nearby iron depends on the vessel
s greater altitude above the horizon than previous models. bearing and (magnetic) latitude, the location of the Remarkably, French eighteenth-century improvements wharf where it was built, and the distribution of soft- and to the instrument did not cater to this new method of hard-iron components in its structure. Note that after observation. This failure was due both to institutional the turn of the twentieth century most marine magnetic lethargy on the part of the French Navy, which pro- data were collected by nonmagnetic ships such as the cessed most of the submitted new inventions, and to the Carnegie and the Zarya or were taken from magnetom- maritime practitioners
conservative stance in adopting eters towed behind the ship and away from its magnetic innovations. Existing imperfections in English com- field. passes eventually led to Ralph Walker
s improved de- [23] In 1801, Captain Matthew Flinders noted on his sign, which bore resemblance to a sundial and allowed voyage to Australia that the compass direction was not magnetic declination to be instantly read. It became the always in the direction of the known declination at that Royal Navy
s standard issue by the late 1790s, replacing place and that the discrepancy depended on the direc- tion of a ship
s bow. He found the effect to be opposite an earlier model by Gowin Knight. in southern latitudes to that in northern latitudes and [20] The main advantage of putting the card inside a attributed it to the inductive effects of the vertical iron bowl completely filled with liquid is that needle move- girders in the primarily wooden ships. The deviations ments are damped by the fluid. In addition, needle and were small, of the order of 2Њ to 3Њ, and his method of card can be made to almost float, dispensing with the compensation, that of placing a vertical soft-iron bar age-old problem of pivot-cap friction. The associated close to the compass position, a so-called “Flinders bar,” technical difficulties were, however, far from trivial; in was widely adopted. With the increased use of iron in the addition to the obvious ones such as freezing and leak- mid nineteenth century for hull cladding it was found age, the medium had to remain clear for years and not that the deviation could become as large as 50Њ. interact chemically with the inside of the bowl, while [24] A frequent procedure was to measure the devia- pressure and volume changes dependent on temperature tions for the 16 main compass points and then plot these had somehow to be compensated for. graphically. Since deviation depends on the local values [21] Traces of liquid compass development before the of H and I, it was necessary to “swing the ship” fre- fourth quarter of the eighteenth century are scarce. quently, especially in the event of a large change in After Royal Society experiments with lodestones acting latitude. Correction for deviation for all ship
s headings under water in the 1660s, astronomer Edmond Halley was performed at the English Hydrographic Office by seems there to have demonstrated a prototype liquid interpolation of the curves of deviation taken at adjacent compass in 1690, but nothing more came of it. The first spinning sites. In 1839 the Astronomer Royal, G. B. comprehensive description appeared in 1779, written by Airy, performed a series of experiments and introduced Dutch physician Johannes Ingen-Housz: a large, sensi- a method of correcting the deviation using compensating tive needle inside a steel tube with a float, housed in a magnets and bars. Two schools of thought developed on waterproof casing filled with linseed oil. In 1795, Danish the subject of deviations: physical corrections, favored by instrument maker C. C. Lous published a treatise [Lous, the merchant navy and advocated by Airy, and tabular 1795] in Hamburg, in which he discussed his efforts to corrections, promoted by Archibald Smith and favored give the concept a practical form. In 1813, English watch in the Royal Navy. In practice, when the deviations were maker Francis Crow was able to patent a liquid compass, large, the Royal Navy used a combination of physical and in 1830 the Danish compass production firm of and tabular corrections. Nevertheless, a nonmagnetic Weilbach did the same. The problem of coping with alternative to the magnetic needle was still avidly sought. variable fluid volume due to changes in temperature, It was found in the gyrocompass, operational by 1911, however, remained unsolved until 1856 when W. R. and eventually, the gyrocompass relegated the magnetic Hammersley invented a system of expansion chambers. pathfinder to the function of auxiliary safety measure on Although fluid compasses began to be adopted by large vessels [Jonkers, 1998]. Navies in quantity in the second half of the nineteenth century, they still suffered serious competition from dry- 2.2. Inclination card instruments like William Thomson
s (invented [25] Instead of trying to balance the needle in the 1878), which carried eight needles suspended in parallel. horizontal plane to measure declination, one could al- It was not until the early years of the twentieth century ternatively pivot the needle on a horizontal axis to mea- 2-8 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

sure inclination, the vertical angle between H and F. 2.3. Intensity Thus, in order to work properly, an inclinometer (or dip [28] For hundreds of years the horizontal and vertical circle) had to be aligned with the local declination first. angles were the only available measurements of the local The earliest proper, quantified measurement of inclina- field. Its magnitude remained outside the realm of prac- tion was made by Robert Norman around 1580. A very tical experience, apart from the previously discussed limited number of observations soon gave rise to the (section 2.2) observation that compasses were generally (erroneous) notion that magnetic inclination could serve unreliable in high latitudes. The weaker the intensity is to determine latitude. It was in part through the pursuit in the plane of measurement, the less accurate the of this dream that inclinometers for shipboard use were reading will be, and the longer it will take a needle developed in late sixteenth- and early seventeenth-cen- displaced from its preferred orientation to return to it. It tury England, among others, by Norman, William Gil- was this relationship between field strength and oscilla- bert, and William Barlowe. A broad hoop of brass was tion rate after a given displacement that was exploited by divided into four quarters, each graduated in degrees. German explorer Baron Alexander von Humboldt Two thin plates were then fixed horizontally, and be- (1769–1859), one of the pioneer magnetic surveyors. On tween them held the axis on which a magnetized needle his extensive travels in South America, Europe, and Asia of ϳ5 inches was to turn in the vertical plane. Both sides he compiled observations of relative geomagnetic inten- then covered with circular panes of glass, it could be sity by comparing timed oscillations of a standard dip hand held by a suspension ring and aligned with the needle with the value obtained in 1799 on the magnetic magnetic meridian. Local inclination could then easily equator (212 oscillations in 10 min) at Micuipampa be read. Other designs had the instrument gimbaled in a (Cajamarca region), Peru. The Humboldt unit of rela- box. However, the measurement of inclination never tive magnetic intensity would eventually be replaced on the European mainland by the French standard, also attained a place in standard navigational practice. On known as a French unit. It was based on intensity as land it was replaced by the dip inductor from 1914 measured in Paris in the early nineteenth century, the [Malin, 1987]. horizontal oscillation time there being defined as 1.0000 [26] Two reasons come to mind why the vertical mag- unit and full magnetic intensity being 1.3482 on the same netic vector was so little regarded by mariners. First, scale [von Humboldt, 1814–1829]. The Parisian measure- once the idea of magnetic latitude was abandoned (itself ment of H corresponds to a value of 34,941 nT. Other only a refuge for times when clouds obscured celestial local units in Europe, such as the reference values ob- information), hardly any practical benefit was thought to tained by researchers at Christiania (Oslo) and St. Pe- accrue from inclinometer readings. Second, the mea- tersburg, were eventually standardized to the French surement itself was more difficult than that of declina- system. tion, the instrument having to remain both level with the [29] An even older scale of relative intensity was ap- horizon and precisely aligned with H. On a constantly plied by the French Admiral Jean De Rossel in 1791– moving ship this necessitated a regular compass nearby, 1794, while partaking in B. d
Entrecasteaux
s voyage in which, in turn, allowed deviation to come into play. An search of his colleague J.-F. La Pe´rouse [Walker, 1866]. additional problem reared its head in high latitudes, Rather than using Paris as the needle swing time refer- where inclination was commonly regarded by navigators ence, De Rossel relied on dip needle behavior as ob- as a nuisance, pulling the compass card downward at one served in the port of Brest. His six measurements have end, thereby increasing friction and aversely affecting been stored under ID code 3500. Yet another measure readings. To restore equilibrium, a piece of wax could be of relative intensity was employed by G. A. Erman, on attached to the card, as suggested by mathematician and his worldwide explorations in the 1820s and 1830s [Er- lecturer in navigation Edward Wright. A more drastic man, 1835, 1841]. His measurement of relative total solution was advocated by W. Barlowe in 1597: simply intensity at Portsmouth in 1836 (a value of 1.33019 filing away the end that appeared heaviest. Although this Erman units) could be linked to a similar observation was a purely local solution, it was still being endorsed in recorded by astronomer Edward Sabine for that year, some eighteenth-century nautical manuals despite im- expressed in so-called British units (10.23 BU). This was provements such as adjustable weights to counterbal- the first British measure of absolute intensity, itself ance the needle
s downward tendency. equal to 4610.8 nT [Barraclough, 1978]; the Erman unit [27] These various obstacles notwithstanding, scat- was therefore equal to ϳ35,460 nT. tered sets of measurements have been compiled at sea [30] Sabine has left a rich legacy of compiled obser- by some explorers, natural philosophers, and the odd vations, discussed in section 5.2.5. As far as his intensi- naval survey, for example, by N. Feuille´e, K. G. Ekeberg, ties are concerned, his data set can be partitioned in two: and J. Cook (now all in the Bloxham compilation, see the original measurements by source [Sabine, 1840, section 5.2.3). Past and present archival research has 1843, 1846, 1849] and three compilations by latitude unearthed Ͼ19,000 inclination observations prior to [Sabine, 1872, 1875, 1877]. Like the Portsmouth value 1931 of vital interest for geomagnetic modeling of sec- mentioned above, the latter were uniformly expressed in ular variation. British units. The earlier material, however, still relied 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-9 on a relative scale, the London unit. Sabine first com- These codes have been listed in order of frequency of pared horizontal needle swing times in London and occurrence in Table 5. Paris in 1822 and again in 1826–1827. The second time, [33] The problem of how to measure absolute (hori- Sabine found the Parisian horizontal intensity to be zontal) intensity was finally solved in 1832 by Gauss. His ϳ1.07137 times the value observed at London [Sabine, solution involved a combination of a magnet
s oscillation 1828]. For the year 1836 the total intensity at the English period and its deflection of a second suspended magnet, capital, expressed in this system, was 1.372 units. Ac- both in the horizontal plane [Malin, 1982]. As far as the cording to Sabine [1872] this value was at that time database is concerned, absolute field intensities re- equivalent to 10.27 BU or ϳ47,352 nT; thus 1.000 Lon- corded in nanoteslas mostly express the horizontal di- don unit would equal ϳ34,514 nT. However, in the first rection too; the Russian compilation by Veinberg [1929– of his three data compilations [Sabine, 1872] the astron- 1933] alone is responsible for 7874 instances of AH N omer stressed the fact that since the BU was an absolute (see section 5.2.7). The remainder, together with some scale and the London measure was a relative standard, early instances of AT N, are found in several ID codes in given that the field strength at London was changing the Jackson compilation (see section 5.2.6). with time so would this equivalence. Sabine assumed the [34] An improved version of Gauss
s apparatus (the following progression to be valid: for the period 1830– Kew unifilar magnetometer) had become the standard at 1837, 10.27 BU; for 1838–1846, 10.28 BU; for 1847– magnetic observatories around the world by 1860. Now- 1854, 10.29 BU; for 1855–1862, 10.30 BU; and for 1863– adays, these instruments have been largely relegated to 1869, 10.31 BU. museums, being replaced by electromagnetic devices. In [31] The above developments have been summarized the 1920s, F. E. Smith developed the coil magnetometer, in Table 3, which lists chronologically each unit
s mea- designed to balance the horizontal field component by sured component, the reference location used to derive an artificial field produced by a given current passing the unit value, its period of application (within the through a coil of known dimensions. More recently, the database), and last, the absolute value in nanoteslas (if proton-precession magnetometer and the flux gate mag- known). It needs to be stressed that this list may not be netometer attained prominence. The former is devised exhaustive; individual researchers in the late eighteenth to observe local intensity with an accuracy of up to 0.1 and early nineteenth century may very well have em- nT; the latter directly measures the three directional ployed other relative scales, based on some locally de- components of the magnetic vector. Both use an elec- rived reference value. Nevertheless, the five units listed tromagnetic field to bring the sensor(s) into a known above could already cause substantial confusion while state; the influence of the Earth
s magnetic field can processing these early intensities and thus had to be then be assessed in the way it changes that state. Marine clearly identified in the database. A three-letter code intensity data have been collected by towing magnetom- was therefore defined that could adequately represent eters at a considerable distance behind a vessel, obviat- the possible combinations. The first letter distinguishes ing the need to compensate for deviation effects. More between relative and absolute measurements; the sec- recently, marine surveys have also been carried out by ond distinguishes between H and F. The third identifies specially adapted aircraft from several nations. the actual unit used. Thus, to give some examples, the [35] Continuous three-component data are nowadays code RH F would represent relative horizontal intensity digitally recorded at magnetic observatories around the in French units, whereas AT B would denote absolute world and regularly transmitted to data centers for fur- total intensity expressed in British units. Table 4 lists the ther analysis. These observatories were first set up in the elements that combine to form the presently employed early nineteenth century and have proliferated in the codes. twentieth century. Unfortunately, they are all based on [32] The database contains a grand total of 16,219 land, and relatively few stations operate in the Southern intensities, stored in relatively few places. De Rossel
s Hemisphere. This underlines the importance of oceanic measure (RT R) was already discussed above. The data, potentially covering over two thirds of the planet
s French unit, however, was used both for the horizontal surface and spanning a far longer period than observa- and the full magnetic vector; RH F is given by Vaillant tory measurements. It is mostly this rich source that has [1840–1852]. RT F is found in ID codes 4107 [Becquerel, been exploited in the various compilations that together 1840], 4482 [Vaillant, 1840–1854], 4484 [von Humboldt, make up the historical geomagnetic database. Before 1814–1829], and 4488 [Aubert du Petit-Thouars, 1840– discussing this maritime material in detail per subset (in 1843]. A number of observations have been recorded by section 5), we will briefly dwell on earlier publications Sabine [1840, 1843, 1846, 1849] using London units that contain historical field measurements. (RT L), now stored in ID codes 4275, 4277, 4285, 4483, 4485, and 4487. His British Unit (AT B), however, oc- curs in the various data sets gleaned from his three 3. PUBLISHED DATA SOURCES compilations, filed under several ID codes from 4190 to 4490 (see section 5.2.5). Erman
s [1841] alternative is [36] Although the majority of historical geomagnetic only found in his own material under ID code 4277. data only exist in manuscript form (often kept in archival 2-10 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Figure 4. Magnetic declination at the Earth
s surface in the years (a) 1600, (b) 1700, (c) 1800, and (d) 1900, based on the time-dependent model gufm1. Contour interval is 5Њ. Isogonics for values Ͼ 60Њ declination have been omitted for clarity. Solid lines indicate northeasting; dashed lines indicate northwesting; projection is cylindrical equidistant.

repositories), a substantial number of documents con- Ludgate in London. Purchas [1625, 1905–1907] pub- taining these readings has been printed (frequently ac- lished several collections of travel literature on his own cessible in national libraries). These consist of the com- account (four volumes in 1625 and the modern edition bined efforts of many workers: astronomers, academics, spaning 20 volumes), which contained East India Com- and members of learned societies among them. It is pany records and many manuscripts that are no longer important to distinguish between (1) published measure- extant. Around the turn of the eighteenth century the ments (near contemporaneous, by the observer or fellow booksellers Awnsham Churchill and John Churchill de- scientists, in me´moires, scientific periodicals, observa- cided to follow in their footsteps by compiling their tory annals, etc.) and (2) published compilations (made Collection of Voyages and Travels (1704), based on orig- long after the fact by scholars, based on either original inal English manuscripts, translated documents, and sources or earlier publications). other navigation-related materials. Subsequent enlarged editions of this publication eventually reached eight 3.1. Published Measurements hefty tomes. [37] We will briefly discuss a few of the most notewor- [39] Unfortunately, the magnetic data (forming only a thy types of printed material that may hold relevant small part of the logbook) have not always been fully information. To start with, the closest a reader of the represented in such printed accounts. This is, for in- printed word can get to the literal contents of naviga- stance, true of G. B. Ramusio
s [1563–1606] compilation tional logbooks is studying contemporary or modern Navigazioni et Viaggi (1563–1603) and several published transcriptions of famous voyages of exploration and con- accounts of George Anson
s circumnavigation of the quest. Tales of high-sea adventure and exploits against globe (1740–1744) [Samaurez, 1973]. From the nine- competitors in the former colonial arenas have always teenth century some manuscripts have fortunately ap- enjoyed a large audience at home, and quite a few peared in facsimile edition, while many others have original logbooks have made it to the printer. At the found a home in series such as the Hakluyt Society time itself the most common vehicles for such texts were Publications in English and the Werken der Linschoten compilations of travel accounts, such as those of Hakluyt Vereniging in Dutch. These issues do generally retain [1589], Purchas [1625], and Churchill and Churchill the complete text, including all remarks on the geomag- [1704]. netic field. [38] Geographer Richard Hakluyt (1523–1616) is [40] A second category of sources is formed by the probably the most famous of all collectors and editors of periodicals of learned societies. These can feature voy- travel narratives. His Principall Navigations, Voiages and age data extracted from logbooks but also letters and Discoveries of the English Nation [Hakluyt, 1589, 1598– reports from foreign correspondents, travelers
ac- 1600] went through several editions in his lifetime and counts, compass experiments by instrument makers, and was based on his interviews with travelers, archival re- measurement series to investigate diurnal variation and search of merchant companies, first-hand records of aurorae. Regular scientific periodicals appeared from explorations, and a huge compilation of existing travel- the 1660s; among the most notable are the Philosophical ogues. His work was later continued by his former assis- Transactions of the Royal Society of London (hereafter tant, Samuel Purchas (1577–1626), rector of St. Martin referred to as Philosophical Transactions) and the Me´m- 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-11 oires de l
Acade´mie des Sciences de Paris. Periodicals more directly associated with maritime matters include the Nautical Almanac in England and the Connaissance des Temps and Annales Hydrographiques in France. All of these publications tended to enjoy a wide distribution, and major libraries will usually own at least one com- plete series of the periodical. Sadly, the data sometimes suffer from transcription errors and editorial pruning, which is, additionally, not always explicitly stated. [41] A third category of sources comprises time series for specific locations, normally major cities where inves- tigators have set up permanent instruments, for instance, at national astronomical observatories. Past observers include G. Graham (1722–1723, London clockmaker and the discoverer of diurnal variation) and G. Gilpin in England, academics A. Celsius and O. Hio¨rter (1740– 1747) in Sweden (who studied the correlation of needle disturbance with the occurrence of aurorae), J. Mac- Donald on Sumatra (eighteenth century), and in France scholars and astronomers P. Petit (1630–1667), J. Picard (1660–1682), M. De The´venot (1663–1681), P. De la Hire (1682–1698), the Cassini family (1685–1691, 1699– 1718, 1773–1793), the Maraldi family (1699–1743, 1755– 1771), J.-B. Du Hamel (1754–1769), and P.-C. Le Mon- nier (1771–1791). However, despite their achievements a Figure 5. An eighteenth-century azimuth compass by J. mere handful of cities can boast a series of more or less Fowler. Courtesy of the National Maritime Museum, London. regular observations spanning over a century prior to See color version of this figure at the back of this issue. 1800. A review of recent efforts to make these data series available to a modern audience is given by Alex- andrescu et al. [1996], who also list all early geomagnetic mentioned in nineteenth-century compilations. Suspi- observations made in Paris (1541–1883, based in part on cions of incomplete or edited processing could therefore earlier work by Raulin [1867] and Rayet [1876]). Other not be checked or amended in these cases. In other capitals with a sustained tradition of geomagnetic obser- instances the original documents have been destroyed or vations include London [Malin and Bullard, 1981; Bar- otherwise lost, likewise preventing any attempt to repro- raclough et al., 2000], Rome [Cafarella et al., 1992], and cess the original data carrier. However, this is largely Edinburgh [Barraclough, 1995]. compensated for by the vast number of manuscripts that have been properly referenced; most compilations have 3.2. Published Compilations proved to be valuable aids in tracking down various [42] Whereas an original time series published con- obscure documents that would otherwise have been temporaneously by the observer(s) would be classified as missed. A review of many past compilations of geomag- published measurements, the modern papers mentioned netic data is given by Barraclough [1982]. in section 3.1 (often combining several historical series) [43] Not all compilations are suitable for geomagnetic fall into the second category of published compilations. modeling purposes. The minimum requirements are an Rather than attempting to safeguard contemporary find- identifiable location (usually a coordinate pair of lati- ings from future oblivion, compilers of geomagnetic data tude and longitude, but a clear geographical reference in later centuries usually had different, more geophysical works almost as well), an unambiguous measurement aims in mind. Instead of producing data sets concen- (i.e., observed rather than estimated), and a clear date. trated in temporal or geographical scope, their investi- It is precisely in this last respect that the oldest compi- gations tend(ed) to be of a much broader focus, com- lations are wanting. Polymaths such as Stevin [1599], bining different types of data sets to present a global or Kircher [1654], and Wright [1657] published carefully regional survey for a substantial period of time. This researched declination tables in the sixteenth and sev- brings the problem of completeness to the fore, since enteenth centuries, which would have been of para- many data sources may have been overlooked by the mount importance for field reconstructions if only the compiler or only represented either in part or otherwise authors had been aware of the existence of secular modified. A second problem worth noting is the practice variation. The year of publication obviously renders a of source referencing often not satisfying modern re- terminus ante quem, but that does not alleviate the quirements. In a number of instances it has proved remaining uncertainty before that time. In the case of impossible to uniquely identify certain original sources Stevin it is known that he did not gather the quoted 2-12 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

readings himself but obtained them from geographer Table 3. Early Units of Magnetic Intensity Petrus Plancius, himself an avid collector of magnetic declination data. Kircher listed no fewer than 56 con- Unit Vector Reference Site Period Applied Value, nT tributors to his own compilation, for which the entries De Rossel F Brest 1791–1794 … lacked both longitude and date. Only a few quoted French H Paris 1798–1839 34,941 letters from fellow Jesuits abroad, relating a (clearly London H London 1826–1845 34,514 datable) voyage, provide instances of useful information Erman H Portsmouth 1838–1846 35,460 [Kircher, 1654, pp. 315–316]. British F … 1818–1870 4,610.8 [44] This situation changed for the better after astron- omer Henry Gellibrand had published his findings in 1635 of time-dependent field change at London (over the period 1580–1634). The earliest useful compilations [46] Finally, one of the largest single compilations, thereafter stem mainly from English attempts to map featuring Ͼ28,000 data points of all types, is the Cata- declination on charts (intended for navigational benefit). logue of Magnetic Determinations in USSR and in Adja- The oldest surviving plot of magnetic measurements cent Countries From 1556 to 1926 in three volumes. It gleaned from multiple ship
s logbooks was presented by was compiled and published by Russian physicist B.P. Henry Phillippes to the Royal Society in 1666 (RS CP 9, Veinberg in 1929–1933 and contained original data from volume 2, number 2, folio 195) [Jonkers, 2000]. About 2 Russia and neighboring states, mostly obtained in the decades later, Edmond Halley [1683] published a small first decades of the twentieth century (treated in section table of some 50 observations in the Philosophical Trans- 5.2.7). Since then, magnetic observatories and surveys by actions (probably based on a compilation effort by an- ship, airplane, and satellite have greatly facilitated data other Royal Society Fellow Peter Perkins), followed by gathering, in comparison with the painstaking task of his isogonic charts in 1701–1702, based upon ϳ150 ob- taking personal measurements of magnetic declination, servations. His successors Mountaine and Dodson [1757], inclination, and intensity. who published updated isogonic charts for 1744 and 1756, claimed to have based their grid-averaged tables of declination at different points on the Earth on Ͼ50,000 4. UNPUBLISHED DATA SOURCES original observations of the field. We note in passing that this data set has long since been lost, and all our efforts 4.1. Logbooks to locate its present whereabouts have proved futile. [47] The navigational logbook contains the written [45] The main era of printed compilations of geomag- record of systematic observation and calculation of di- netic data was the nineteenth century, featuring the rection, distance, and position, as well as magnetic dec- work of Hansteen, A.C. Becquerel, Sabine, and van lination, wind, weather, currents, noteworthy landmarks Bemmelen. In 1819 the Norwegian astronomer and and events, drafts of harbors (with anchoring depths), physicist Christopher Hansteen published Untersuchun- the appearance of coastlines, and other useful particu- gen u¨ber den Magnetismus der Erde, which listed data lars. The accumulated experience could be used as a from land surveys and 73 nautical voyages from 1589 to reference for successive voyages by the author himself or 1816. His collection includes many of the great scientific other navigators, by hydrographers to amend charts and expeditions during the latter half of the eighteenth cen- sailing directions, and by modern researchers to recon- tury, including Cook
s three voyages, contributing over struct ship
s tracks of past voyages. Begun as a personal 6500 declination and 1200 inclination observations. Re- diary of maritime matters, it evolved into a highly for- grettably, reformatting, errors, and omissions limit its malized data carrier with preprinted sheets to be handed modern usage to an index of sources. Another valuable in for official inspection within a few days after reaching addition was made by Becquerel
s [1840] Traite´ Expe´ri- the home port. For a discussion of the various maritime mental de l
E´lectricite´ et du Magne´tisme, which contains organizations that supplied the logbooks used in this the only comprehensive collection of relative intensities. research, please consult Jackson et al. [2000]. Regrettably, Becquerel did not reference his sources [48] The way a logbook stored information varied other than by name of the observer, which is why his data widely, depending on period, country of origin, maritime set appears in the database as is (under ID code 4107). organization, and individual author. Longhand entries Several Philosophical Transactions papers by astronomer started to give way to (vertical) semicolumnized and Edward Sabine span the period 1818–1870 with excep- full-columnized layout in the 1620s on Dutch and En- tionally good coverage; his contributions will be treated glish East Indiamen; in the 1680s the Royal Navy and at length in section 5.2.5. Finally, in the 1890s, Dutch the French (both Navy and Compagnie des Indes) fol- physicist Willem van Bemmelen processed 165 nautical lowed this practice. Another, horizontal, block-shaped sources prior to 1741. His data set has become parti- format with fixed designated areas for each type of tioned in time and can now be found in three different observation, fitting two entries to a page, first appeared subsets of the historical database (see sections 5.2.2, in English East India Company (EIC) logs from the 5.2.3, and 5.2.8). 1660s (e.g., Figure 6). 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-13

Table 4. Letter Codes Identifying Intensity Measurement sanitized version for presentation purposes, perhaps Types even compiled after the fact, and generally holding fewer remarks and data. French naval archives frequently kept Code Position Type a separate extract (the de´pouillement du journal) in R 1 relative addition to the logbook for hydrographical purposes. A A 1 absolute starker style is also recognizable in EIC sources during the second half of the eighteenth century, when less H 2 horizontal field T 2 total field (F) space was allowed for general remarks, and nearly all flavor of life onboard outside the navigational realm was R 3 De Rossel units lost. This development is echoed by Danish material. F 3 French units L 3 London units [51] What the daily entries should contain can be E 3 Erman units learned both from navigation manuals, which often sup- B 3 British units (BU) plied a few pages of a sample voyage, and from instruc- N 3 nanoteslas (nT) tions by authorities. Of special interest from the current viewpoint is the requirement to record magnetic decli- nation. Some of the examples given in textbooks lacked a column for variation of the compass, as it was generally [49] It was copied by the Danes some 60 years later called by mariners, while certain directives expressly but was not adopted by the Dutch and the French. The stipulated that all opportunities to observe should be larger maritime organizations eventually achieved in- seized and noted. Early logs often feature the geomag- creased standardization by supplying preprinted forms. netic field in the remarks, while later ones had a separate A Dutch East India Company (VOC) manuscript in the space or column reserved. Ideally, a clear distinction was National Library of Scotland seems to be the oldest made between observed magnetic declination and the surviving Dutch example (1640–1643), whereas the EIC value allowed for in steering, dead reckoning, and shoot- had adopted the practice by 1705 and the French and the ing landmarks. Other occasional remarks concern the Danes adopted it in the 1710s and the 1730s, respec- type of observation made, as regards time of day, am- tively. The English “block” design became the official plitude or azimuth (i.e., whether the Sun was sighted on EIC standard in 1716; the present research uncovered the horizon or above it), and, in rare instances, with what preprinted examples from the 1730s onward, both in make of compass the measurement was performed. The England and Denmark. This was also the format VOC, for example, employed a system of (Ͼ30) abbre- adopted by the Hudson
s Bay Company from the 1750s viation combinations, whereas certain Danish and onward. The EIC introduced new guidelines on logbook French logs had hourly tables per nautical day to desig- layout at several instances, for example, in 1758 (printed nate the time of sighting and applied simple abbrevia- from 1761) and 1791 (with an added entry for chronom- tions in other instances. For an analysis of measurement eter longitude). By this time, even the merchant navies accuracy, illustrating the high precision with which the were exchanging the written diary for neat columns. The declination could be determined (a standard deviation VOC house style was occasionally imported by individ- of 0.46Њ), consult Jackson et al. [2000]. uals on Danish East Indiamen of the Asiatisk Kompagni (around 1740) as well as on Dutch slavers (1760s) and 4.2. Repositories naval vessels (1780s). [52] Most of the research to unearth primary sources [50] In English naval practice both vertical and hori- took place in repositories that were either directly related zontal order existed alongside each other. They seem to to seafaring (i.e., maritime museums) or held sizeable have served different purposes; the blocks version with holdings of particular shipping companies (such as national half a page of space per day contained the rough draft, archives). These will briefly pass review per country. while the columns often appear to have served as a [53] In England the main maritime holdings of the nation are kept in three places (all in London): the Table 5. Frequencies of Intensity Measurement Types in National Maritime Museum (NMM), Greenwich; the the Database Public Record Office (PRO), Kew; and the British Li- brary (BL). The National Maritime Museum owns the Code Frequency ID Codes largest collection of early modern Navy logs (section /LOG and various family papers), consisting of ϳ5000 RT R 6 3500 captain
s logs and 5000 lieutenant
s logs. Unfortunately, RT E 17 4277 RH F 24 4481 from the viewpoint of this particular research most nav- RT L 104 4275, 4277, 4285, 4483, 4485, 4487 igational information is usually found in the master
s RT F 537 4107, 4482, 4484, 4488 logs, which are kept at the PRO (section/ADM). This site AT N 1,157 4509, 4511, 4519, 4528, 4534, 4538 is also the resting place of large parts of the Hudson
s Bay AT B 2,994 4190–4490 Company archives (on microfilm, section/BH 1), Sabine
s AH N 11,380 4449, 4490–4516, 4526–4551, 4553–4555 papers (/BJ) and part of the collection of the Royal Green- 2-14 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Figure 6. Excerpt from the log of the King George on 2 July 1719. Note the azimuth observation of magnetic declination around five o
clock in the afternoon and the recording of meridian distance rather than longitude. By permission of The British Library (source: IOBL L/MAR/B 402 B).

wich Observatory (/RGO). The other half of the RGO which contains charts, documents, and the Admiralty Li- papers is kept at Cambridge University Library (CUL) but brary. Many documents there were at the time of investi- unfortunately yielded very few data points. gation still in the process of being cataloged. An unex- [54] For the English East India Company the India pected find (made through the database of the National Office of the British Library (IOBL) holds some 3000 Registry of Archives in London) was the logbook by Alex- remaining logbooks (section L/MAR/), as well as a ander Keppel (who sailed with Anson on his circumnavi- wealth of other company material. We have used only a gation in 1740–1744), in the “Albemarle papers,” kept at tiny fraction of the available material (325 out of the the Suffolk Record Office (Ipswich). estimated 1500 pre-nineteenth-century logs that are held [55] Moving on to France, the state
s centralized bu- there; however, we have processed all of the seventeenth- reaucracy has ensured that almost all relevant sources century logs). These logs were formerly kept in the India can now be found in Paris. A number of navigation Office, a separate building, but have now joined the other manuals are housed in the Bibliothe`que Nationale collections at the new site of St. Pancras. This main loca- (BNP) (see in particular the sections “French Manu- tion also houses the BL
s Manuscripts Library, which con- scripts” (/FR) and “New French Acquisitions” (/NAF)). tains a limited number of other logbooks. Less success has A far more important repository for these studies has been encountered at the Hydrographic Office (Taunton), been the Archives Nationales (ANP). Most surviving 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-15 material of the Compagnie des Indes is kept here, as well 4.3. Longitude and Meridian Distance as vast collections of naval documents, including many [60] The proper geographical plotting of data points is hundreds of logbooks of both organizations. Note, how- of vital interest to geomagnetic modeling, as it directly ever, that the logbooks have not been stored separately affects an important part of the error associated with but are ordered by destination only (in maritime section/ each reading. An effort has therefore been made to MAR/4JJ). Information on compasses and their appli- reconstruct how the daily procedure of dead reckoning cation can be found in sections/MAR/B4, /MAR/G, and was actually performed aboard ship. At the beginning of Colonial Archives/COL/C2. each new ship
s day (at noon) the Day
s Work was [56] In the Netherlands the huge archive of the VOC performed, preferably with the aid of observed latitude is of similar size as that of its English counterpart at the (using backstaff and later octant or sextant); for discus- India Office, kept in the Algemeen Rijksarchief (ARA) sion of latitudinal accuracy, consult Jackson et al. [2000]. (General State Archives) at The Hague (section The point of departure could be a landmark, the last /1.04.02). However, despite centralizing efforts, sources previous point where latitude was observed, or simply from this organization remain somewhat scattered over (and most often) the best dead-reckoned estimate from the country; several municipal archives (of those cities the previous day. The parallel halfway in between these that had a company chamber) also keep records, as do two latitudes was called the middle latitude (Mid. lat), some provincial archives. The southern province of Zee- used to estimate the length of a longitudinal degree (in land furthermore maintains the paper legacy of the nautical miles) over the interval (see below). The dis- Middelburgsche Commercie Compagnie (MCC), which tance (in miles) that a ship had departed from any resides at the Rijksarchief Zeeland (RAZ), (Provincial meridian was called departure, which had to be con- State Archives) in Middelburg. verted into difference of longitude (diff. long) (in de- grees and minutes). Only on the equator does 1 min of [57] Dutch naval sources can best be researched at the longitude equal 1 (English) nautical mile; at higher Rijksarchief in The Hague (Admiralties, sections/1.01.46 latitudes the length of longitudinal degrees expressed in and /1.01.47). Note, however, that like the VOC docu- miles decreases with the cosine of the latitude (to zero at ments, naval logs may also reside among the family the poles). The so-called meridian distance (the ship
s papers of their authors (section/1.10). One of the finest physical distance from a meridian at the current lati- collections of materials related to navigation in the wid- tude) can be expressed in spherical degrees (or great est sense is the one of the National Shipping Museum
s circle degrees), that is, using degrees of maximum length Library (Nationaal Scheepvaart Museum) in Amster- (for a given radius of the sphere) as found on the equator. dam (NSM) (catalogue I); other maritime museums in [61] Longitude has to be measured from a fixed ref- the Netherlands tend to have at least a few logbooks. erence. The Greenwich meridian was only adopted as an [58] Danish relevant material is mainly confined to international longitude standard in 1884, and some na- holdings at the Rigsarkivet (State Archives) in Copen- tional conventions remained in use later than that date. hagen. The Danish East and West Indian Companies The coordinates contained in the French journal An- have left sizeable holdings there (section /VA XIV), as nales Hydrographiques, for instance, continued to mea- have the Greenland Companies (section /140) and the sure longitude from Paris until at least 1895. However, Danish Navy (sections/85.b.1–2). A large collection of going further back in time, conventions regarding the Danish whaling logs was at the time of investigation not choice of prime meridian from which to reckon from yet accessible to the public. were far more flexible and have to be accounted for in [59] Finally, an extensive search was undertaken in processing maritime data from that period (see section Spain to unearth useful records in general and measure- 5.1). Three groups of islands in the North Atlantic car- ments made during Pacific crossings in particular. After ried most meridians referred to in Dutch and French an initial foray into the Archivo General de Indias logbooks: the Canaries, the Azores, and the Cape Verde (Seville), attention was focused on the Museo Naval Islands. They similarly crop up frequently in English and (Madrid), where most of the material on D.A. Malaspi- Danish logs. Once ocean travel gained momentum, how- na
s expedition (1789–1993) was found. Less success was ever, and Mercator charts started to cover whole oceans, encountered searching the Biblioteca Nac¸ional (Ma- the European mariners
need for a standard became drid), the Museo Nac¸ional de Ciencias Naturales (Ma- more pronounced. However, different countries found drid), the Real Academia de la Historia (Madrid), the different answers to this problem. Archivo General de Marina (Ciudad Real), and the [62] The Dutch case is the most straightforward. When Archivo de Simancas (Simancas, Valladolid). In general, the separate companies that were to form the VOC ven- the few logbooks found often failed to distinguish clearly tured eastward in the 1590s, the twin islands of Corvo and between observations of and steering compensations for Flores (Azores) were the norm. The earliest reference to local magnetic declination. Our hope of finding the Tenerife (Canaries) in the database dates back to 1614. By manuscript legacy of the so-called Manilla Galleons the 1640s the Azorean pair had disappeared in favor of this proved unsubstantiated; perhaps some records still sur- newcomer. Moreover, an island in the Cape Verdes, St. vive in the former Spanish colonies (see section 6). Jago, vied for the navigator
s attention from the 1660s until 2-16 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

about 1710. Afterward, Tenerife
s Pico de Teide
s suprem- acy remained unchallenged throughout the eighteenth cen- tury, and in 1787 it was formally adopted by the Amster- dam Admiralty as the prime meridian of choice. [63] Charting French developments is more difficult, since a discrepancy exists between official policy and actual practice as attested in the logbooks. Already in 1634, King Louis XIII determined that Ferro (Canaries) was to become the prime meridian for the French, but their navigators seem to have preferred Tenerife and St. Jago (so they could rely on Dutch charts). Differences were only reconciled with the ascendancy of the Paris meridian in the 1750s, which completely wiped out the competition within the space of a single decade. [64] The only constant in Danish practice regarding meridian usage is change. Corvo, Tenerife, Paris, Green- wich, all choices seem based purely on the fancy of the navigator in charge. Often too, the English practice is followed, to which attention will presently be turned. [65] Until the advent of the Greenwich meridian in the 1750s (chosen because the Royal Observatory was built there), English oceanic navigation adhered to a Figure 7. A dead reckoning course triangle on a plane chart. rather idiosyncratic practice. It fundamentally differed ␣ ␪ Here is course, r is distance, 1 and “lat.left” is latitude of ␪ ␪ in two aspects from the Dutch and French manner: (1) departure, 2 and “lat.in” is latitude of arrival, d is change in ␾ ␾ Each major land sighting provided a new, local meridian latitude, 1 and “lon.left” is longitude of departure, 2 and from which to reckon. (2) Position east or west from a “long.in” is longitude of arrival, d␾ is change in longitude; the ␪ ␪ reference meridian was often measured in meridian dis- middle latitude is the midpoint between 1 and 2; the “quot- tance (expressed in spherical degrees or miles) rather ed” terms are found in English logbooks than in longitudinal degrees. [66] Both aspects have peculiar problems associated ␪ with them, which have severely slowed the processing of rhumb, r is the length of the rhumb, is the latitude, and ␾ English logbooks. Since documents stating longitudinal is the longitude, and assuming a spherical Earth, the distance in miles have been studiously avoided in the displacement along each axis can be expressed as sample, there remained a large number of sources either d␪ ϭ cos ␣dr (2) using longitudinal or spherical degrees, which are indis- tinguishable in format. Both use degrees and minutes, d␾ ϭ sec ␪͑sin ␣dr͒ ϭ sec ␪͑tan ␣d␪͒. (3) and both are of equal length on the equator. The higher the latitude is, however, the larger the difference is, since For the east-west difference the conversion from merid- a spherical degree is of fixed length (ϳ111 km), while the ional to longitudinal degrees then requires resolving the converging meridians reduce the distance between them following integral:

(and thus the length in miles or kilometers of a longitu- ␪2 dinal degree) at higher latitudes. In other words, merid- ⌬␾ ϭ tan ␣ ͵ sec ␪d␪ ional (i.e., spherical) degrees (m) relate to longitudinal ␪1 degrees (␾) as a function of latitude (␪): ␪ ␲ ␪ 2 ϭ tan ␣ͭlnͫtanͩ ϩ ͪͬͮ . (4) 1 4 2 ␪ ␾ ϭ m. (1) 1 cos ␪ The seventeenth-century English mariner was, unfortu- [67] This leads to the confusing situation that a ship nately, unaware of techniques involving integrals, so heading south from the equator will see its distance east instead he took the middle latitude in his Day
s Work as or west from some prime meridian elsewhere (expressed the best estimate of the length of a longitudinal degree in spherical degrees) diminish gradually, whereas its over the daily interval. The database, however, only longitudinal coordinate would naturally remain con- contains points for dates when compass data or land stant. In general, the size of the vertical component of sightings were reported, necessitating the adoption of an the course vector determines how much the ratio be- additional assumption of constant headings between tween the two systems changes over the distance trav- successive stored coordinates. A visual check on the eled. Taking a course-distance triangle (an example is resulting course plot was also required, because most shown in Figure 7), where ␣ is the direction of the sources gave no information whatsoever regarding the 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-17

Table 6. Database Subset Statisticsa

ID Code(s) Data Set Records D I Intensity Total

0001–2272 Jonkers 56,396 40,881 35 0 40,916 3001–3500 Murray 25,246 23,792 98 6 23,896 4002–4107 Bloxham 9,880 7,828 1,568 484 9,880 4108–4187 Paris/Madrid 2,645 2,616 28 0 2,644 4188–4488 Sabine 9,000 5,834 4,923 3,197 13,954 4489–4552 Jackson 11,758 9,984 4,722 4,658 19,364 4553–4555 Veinberg 13,835 12,355 8,114 7,874 28,343 4556–4662 Hutcheson 2,971 2,934 37 0 2,971 4663–4664 Cartes and Plans 45,336 45,336 0 0 45,336 Total 177,067 151,560 19,525 16,219 187,304 aA single record may contain a land sighting and/or one or multiple geomagnetic observations (D, I, or intensity (H or F)).

applied system, while others changed from one type to the 1760s should be aware of the potential pitfalls of the other on subsequent legs of a voyage, depending on mistaking spherical for longitudinal degrees. the chart in use for that particular part of the journey. [68] English naval sources are lacking for the first half of the seventeenth century and are sketchy at best for 5. DATA DESCRIPTION the second. The cornucopia of the East India Company
s archives, on the other hand, yields excellent temporal [69] The geomagnetic database in its present form is resolution, with only a small gap around the 1650s. available to interested parties through the World Data Although the first longitudinal degrees already appear in Centre for Geomagnetism at the British Geological Sur- an EIC logbook from 1621, the abundance of meridional vey, Edinburgh, Scotland. In addition to the data file degrees in the seventeenth century and the EIC
s per- itself (a fixed-format ASCII table) a separate listing of sistent commitment to them during the first half of the prime meridians (ditto) and a data extraction utility (in eighteenth century is quite remarkable. In the processed FORTRAN) are supplied, which allow the prospective Navy logs, only two instances (in 1673 and 1689) attest to user to define and write out a subset using one or several meridional reckoning being favored, while longitude had selection criteria, such as time, geographical area, and been uniformly adopted there by the 1690s. Fortunately, required type(s) of measurements. The output either the standard prime meridian of Greenwich (like Paris in follows the database format (see below) or consists of France) made rapid and determined headway in English user-defined selection of variables, such as ID code, shipping from the 1750s; the last occurrence of meridian year, latitude, longitude (relative to Greenwich or the distance in the survey is dated 1759. Nevertheless, any original meridian), and type(s) of observation, all either future researcher processing navigational data prior to in degrees and minutes or decimal degrees. More infor-

Figure 8. Geographical data distribution of declination observations made before 1590. Here n ϭ 160; some points may overlap; projection is cylindrical equidistant. 2-18 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Figure 9. Geographical data distribution of declination observations made in 1590–1699. Here n ϭ 12,001; some points may overlap; projection is cylindrical equidistant.

mation regarding data and program can be found in the important statistics. Geographical distribution plots per accompanying description file of the World Data Centre epoch have also been provided for declination (see Fig- database. ures 8–11), inclination (see Figures 12–14), and intensity [70] The main compilation resulted from a number of (see Figures 15–16), data collection efforts since the early 1980s, primarily [71] Some of these subsets consist purely of primary conducted at the universities of Cambridge, Leeds, and sources (historical manuscripts), such as those of Amsterdam. These several phases are represented by Jonkers and Cartes and Plans. Others are solely based separate subsets in the database and sections 5.2.1–5.2.9. upon printed sources (data and compilations published Each individual source has been assigned a unique iden- in the past) such as those of J. Bloxham, E. Sabine, B. P. tification code (ID code) in order to link all data points Veinberg, and K. A. Hutcheson. A third category con- to their respective source. Neither the database itself tains information of both kinds; examples include the (not included here) nor Table 1 of references are chro- sets of A. Murray, A. Jackson, and Paris/Madrid. All nologically ordered, but they merely reflect the various subsets will be treated in detail in sections 5.2.1–5.2.9. compilations made; each of these spans a certain range Their order in the database bears no particular signifi- of ID codes. Table 6 gives a brief overview of their most cance; it merely reflects the sequence of adding the

Figure 10. Geographical data distribution of declination observations made in 1700–1799. Here n ϭ 68,076; some points may overlap; projection is cylindrical equidistant. 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-19

Figure 11. Geographical data distribution of declination observations made in 1800–1930. Here n ϭ 71,323; some points may overlap; projection is cylindrical equidistant. several subsets to the main compilation. Listed in chro- because of multiple types of observations being taken at nological order of the modern processing of sources (i.e., the same time and coordinates. the time when recast in machine-readable form for geo- magnetic analysis), the ranking from eldest to youngest 5.1. Data Formats would have been J. Bloxham (ϳ1982–1985), E. Sabine [72] The database itself contains the following fields (ϳ1982–1985), B. P. Veinberg (ϳ1986–1987), A. Jack- (character length in parentheses): ID code (4), year (4), son (ϳ1989), K. A. Hutcheson (ϳ1989–1990), A. month (2), day (2), degrees of latitude (3), minutes of Jonkers (ϳ1993–1998), A. Murray (ϳ1995–1997), latitude (3), seconds of latitude (3), original longitude
s Cartes and Plans (ϳ1998), and Paris/Madrid (1998– degrees (4), minutes (3), and seconds (3), corrected 1999). The distribution in time of the measurements longitude
s degrees (4), minutes (3), and seconds (3), themselves is given in Table 7, which highlights the prime meridian (2), voyage edited? (1), magnetic decli- temporal range differences between the compilations. nation in degrees (4) and minutes (3), number of Each subset
s geographical distribution is additionally declination measurements on which this value is based displayed in a separate plot; the total number of plotted (2), magnetic inclination in degrees (4) and minutes (3), points can be lower than the last column in Table 7 number of inclination measurements on which this

Figure 12. Geographical data distribution of inclination observations made in 1590–1699. Here n ϭ 53; some points may overlap; projection is cylindrical equidistant. 2-20 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Figure 13. Geographical data distribution of inclination observations made in 1700–1799. Here n ϭ 1747; some points may overlap; projection is cylindrical equidistant.

value is based (2), intensity (10, of which 3 are deci- used in the reckoning. The format of this table of me- mals), number of intensity measurements on which ridians consists of the reference code (2), the geograph- this value is based, (2) type of intensity measurement ical description (40), and its latitude (6) and longitude (3), and remarks (50). All fields except the last always (7), both in degrees and minutes. Each record is termi- contain an entry; some of these can be “999” if data is nated by a “newline.” missing. The total length of a single record is 128 [74] The ID code of each database record refers back characters (including one “newline”); a total of to Table 1, which describes the data
s origin, that is, the ϳ177,000 records then results in a file size of ϳ21 and processed data carrier(s). These can include an original a half megabytes. manuscript, a nineteenth-century data compilation, or a [73] Dates follow the calendar style in use by the scientific paper, for example. Table 1 consists of the author of the consulted document. For the conversion following fields (character length in parentheses): ID from Julian to Gregorian calendar, consult, for instance, code (4), year of departure (and return) (10), ship(s) Stephenson [1997]. The variable designating “prime me- (25), author(s) of record (27), number of declinations ridian” contains a two-letter code which refers to a (5), number of inclinations (5), number of intensities (5), separate list of 286 identified geographical landmarks and source(s) (50). Additional information for the char-

Figure 14. Geographical data distribution of inclination observations made in 1800–1930. Here n ϭ 17,723; some points may overlap; projection is cylindrical equidistant. 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-21

Figure 15. Geographical data distribution of intensity observations made in 1700–1799. Here n ϭ 36; some points may overlap; projection is cylindrical equidistant. acter fields Ships and Surveys, Authors, and Locations 1998. The first part encompassed solely Dutch source may be found in the Table 1 footnotes. References to material, while the second part extended the research to archival manuscripts in the last field rely extensively on France, Denmark, and England (in that order). This a system of abbreviations given in the Table 1 footnotes. sequence is still reflected in the assignment of ID codes Published and manuscript sources are given following to the various sources. In a temporal sense this project the author-date convention and can be found in the last followed maritime developments from the start of Eu- column of Table 1. If an entry contains multiple refer- ropean East India ventures in the late 1590s until the ences, these are listed in order of their occurrence in the turn of the nineteenth century. Its main aim was to study database. compass use and compensation onboard ship for local magnetic declination, between nations, shipping routes, 5.2. Description of Compilations and (types of) maritime organizations. Thus the sam- pling has been quite broad; not only East India Compa- 5.2.1. Jonkers nies and Navies were included, but also any other type of [75] The Jonkers subset, shown in Figure 17, was shipping that would allow a reasonable sample size to be compiled by one of us (A.R.T.J.) over the period 1993– taken was included. In addition to slavers and whalers,

Figure 16. Geographical data distribution of intensity observations made in 1800–1930. Here n ϭ 16,183; some points may overlap; projection is cylindrical equidistant. 2-22 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Table 7. Temporal Distribution of All Defined Database Ͼ2000 original ship
s logbooks processed for geomag- a Subsets per Time Period netic research (of which some 1300 yielded useful data). In addition, these voyages have been scrutinized in the Data Set 1510–1589 1590–1699 1700–1799 1800–1930 voyage editor, and all longitudes between landfalls have Jonkers 0 6,574 34,338 4 been adjusted to reflect identified geographical land- Murray 0 1,729 19,706 2,461 marks where possible. Extensive procedures have also Bloxham 149 476 8,788 467 been applied to filter out typographical errors and other Paris/Madrid 0 135 2,509 0 outliers. For the numerous archival references, please Sabine 0 0 37 13,917 consult Table 1 (footnotes). Jackson 0 0 2 19,362 Veinberg 13 78 462 27,790 Hutcheson 0 2,964 7 0 Cartes & Plans 0 98 4,010 41,228 5.2.2. Murray Total 162 12,054 69,859 105,229 [77] The Murray subset was compiled by one of us aAll types of observation are combined. (A.M.) over the period 1995–1997 and is shown in Fig- ure 18. Much of the work has been done in conjunction with the previous subset in order to avoid overlap. The large merchant companies such as the Hudson
s Bay geographical focus was, however, more limited than the Company have consequently provided valuable data. Jonkers data set. The main aims here were twofold: (1) Printed, published ship
s logbooks have been completely to strengthen the historical database with new, original excluded; the focus throughout has been on manuscript manuscript sources, mostly from the English East India sources only. This approach implies that “famous” voy- Company and the Royal Navy, kept at the BL and the ages of discovery are almost totally absent from this PRO, and (2) to fill in many of the gaps left by previous subset, as these have often been (re)published else- (published) compilation efforts. where. The only claim to fame of this material can be [78] The time window for this source-processing exer- made by the very early voyages to the East Indies by the cise was slightly later than the Jonkers subset, spanning Dutch and the English. Sparse recorded land sightings in the period 1615 to 1856 inclusive. The brunt of the Dutch East India logbooks have been supplemented material stems from the eighteenth century, with minor with information on departures and arrivals compiled by contributions from before and after. Most observations Bruijn et al. [1987], the standard reference work for all are of magnetic declination, but nearly 100 values of voyages made under that company
s flag to and from the inclination were also secured. Of particular note is a late East Indies. addition (filed under ID code 3500) containing the rel- [76] Numbering Ͼ40,000 observations, this subset is ative intensities and inclinations recorded by De Rossel the second largest in the database and is the strongest of in 1791–1794 on d
Entrecasteaux
s voyage [Walker, all for the seventeenth century and the eighteenth cen- 1866]. These constitute some of the earliest known rel- tury. This is, however, offset to some extent by the fact ative intensities ever observed (see section 2.3). Once that it almost exclusively contains readings of declina- again, the ID codes have been assigned in order of tion. Nevertheless, this compilation alone represents source processing in the various repositories.

Figure 17. Geographical data distribution of the Jonkers subset. Here n ϭ 40,913; some points may overlap; projection is cylindrical equidistant. 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-23

Figure 18. Geographical data distribution of the Murray subset. Here n ϭ 23,779; some points may overlap; projection is cylindrical equidistant.

[79] The second spearhead of this project, the filling (in collaboration with D. Gubbins and D. Barraclough) in of gaps in the data record left by predecessors, has over the period 1982–1985 has a number of specific enriched the database with many famous voyages. strong points worth noting. For one, it has an extensive Among them are the exploits of T. Roe (1614–1615), G. temporal range, from the sixteenth (largest contribu- Anson (1740–1744), H. de Bougainville (1766), W. Bligh tion) to the nineteenth century. Furthermore, the set (1788), A. Malaspina (1789–1793), G. Vancouver contains both many inclinations and intensities (in addi- (1791), B. d
Entrecasteaux (1792), and M. Flinders tion to numerous declination measurements). The com- (1801). Most of these data have been extracted from piler has earlier published a cursory description of published sources, either modern or contemporary, in- sources [Bloxham, 1986], in which he also described how cluding both originals and facsimiles of single voyages he plotted all voyages and adjusted the longitudes of [e.g., Pote´rat, 1815; Malaspina, 1885; Saumarez, 1973], most to bring them into coincidence with reported land and compilations made long after the fact [e.g., Han- sightings. This was done without the aid of the voyage steen, 1819; van Bemmelen, 1896; Erman, 1835, 1841]. In editor. The geographical distribution of the subset is particular, Malaspina
s Pacific explorations deserve shown in Figure 19. more attention than hitherto given by scholars. Nearly [82] The data then provided the input for an effort to contemporaneous with Cook
s voyages, this Spanish ex- model the geomagnetic field at the CMB for two time pedition recorded Ͼ350 declinations on its 4-year voy- windows: 1695–1735 (centered on a 1715.0 epoch age. Additional data from this journey, including 28 model) and 1760–1789 (centered on a 1777.5 epoch previously unknown measurements of inclination, were model). Consequently, much of the data cover only these later found in Spanish archives and will be treated in two periods. Furthermore, famous voyages abound in section 5.2.4. this subset; they include the logbooks of Dampier [80] As the Murray subset was processed in tandem (1699–1701), Feuille´e (1708), Roggeveen (1721), Wallis with that of Jonkers, the same postprocessing was per- (1766–1768), Carteret (1766–1769), Le Gentil (1766– formed on both: All voyages have been scrutinized in the 1771), Ekeberg (1766–1777, 1770–1771, 1774–1775), voyage editor, and all longitudes between landfalls have Wales (1772–1775), Phipps (1773), Cook (1776–1780), been adjusted to reflect identified geographical land- and La Pe´rouse (1784–1788). Some observations on marks where possible. Extensive procedures have also land have also been included. In general, the data have been applied to filter out typographical errors and other been ordered alphabetically by author. outliers. Although numbering fewer voyages than the [83] An aspect of this compilation worth noting is the previous subset, the data yield per logbook has proved fact that is entirely based on printed material. The list of much higher; the Murray subset constitutes the fourth- sources was initially constructed from two published largest contribution to the overall total. compilations [Hansteen, 1819; van Bemmelen, 1899], af- ter which published logbooks were located for some 5.2.3. Bloxham voyages, and their (presumably more accurate) data [81] Although of more modest size than the previous replaced the values recorded in the two compilations. two compilations, the data set compiled by J. Bloxham The sixteenth-century data comprise a separate set (ID 2-24 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Figure 19. Geographical data distribution of the Bloxham subset. Here n ϭ 9366; some points may overlap; projection is cylindrical equidistant.

code 4106), itself derived from nine earlier compilations was processed by K. Hutcheson (see section 5.2.8 for a (including the van Bemmelen [1899] compilation). Many more extensive treatment of van Bemmelen
s work). intensities (1791–1829) have been gleaned from Becque- rel [1840]. 5.2.4. Paris/Madrid [84] Note also that because Bloxham modeled specific [85] The smallest subset of all consists mainly of dec- time windows, he has processed only those sources of lination values from the seventeenth century, with some Hansteen
s [1819] and van Bemmelen
s [1899] work that earlier data and some inclinations. It is shown in Figure satisfied his temporal criteria. Some of Hansteen
s other 20. The main source is a single document in the Parisian voyages were therefore processed about a decade later Archives Nationales, archival reference ANP MAR 3JJ/ and can be found in Murray
s subset (see section 5.2.2). 35, entitled Recueil de Toutes les Observations de Varia- As far as van Bemmelen is concerned, the material from tion, a geomagnetic compilation from the second half of the sixteenth century resides in ID code 4106; some the eighteenth century. It was made for hydrographical seventeenth and early eighteenth century material has purposes by a M. Mandille, at the De´poˆt des Cartes et received separate ID codes in Bloxham
s subset; the rest Plans de la Marine of the French Navy (Paris) based in

Figure 20. Geographical data distribution of the Paris/Madrid subset. Here n ϭ 2373; some points may overlap; projection is cylindrical equidistant. 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-25

Figure 21. Geographical data distribution of the Sabine subset. Here n ϭ 8939; some points may overlap; projection is cylindrical equidistant. part on earlier work by his predecessor Guillaume 5.2.5. Sabine Delisle. The information in this document spans the [88] Major General Sir Edward Sabine (1788–1883) period 1682–1754; most data refer to the first half of the began collecting geomagnetic observations when he eighteenth century. Sources include the work of served as astronomer on two voyages in search of the Feuille´e, Jesuit reports, extracts from the Me´moires de Northwest Passage under J. C. Ross (1818) and W. A. l
Acade´mie Royale des Sciences (Paris) and the Philo- Parry (1819–1820). For the next half century he collated sophical Transactions of the Royal Society (London), and data on declination (almost 6000 points), inclination numerous original logbooks, kept at the De´poˆt des (almost 5000 points), and intensity (Ͼ3000 points) from Cartes et Plans de la Marine. Some of these have been hundreds of different sources, including the voyages of processed by the famous hydrographer d
Apre`s de Man- the Krotkoi (1828–1842), the Beagle (1834–1836), the nevillette. Bonite (1836–1837), and the Ve´nus (1836–1839) and the [86] The data in this document fall into two broad cat- United States Coastal Survey (1843–1863) (see Figure egories: those recorded on voyages and those collected on 21). For a historical analysis of the quest for magnetic land. In the former case each datum is presented together data in the early nineteenth century, and in particular with the appropriate measurements of latitude, longitude, Sabine
s role in this, see Cawood [1977, 1979]. Sabine, and the date of observation; longitudinal corrections in the who was president of the Royal Society, also published voyage editor have been applied in most cases. In the latter three compilations in the Philosophical Transactions, category the date and place of observation alone are based on his earlier work published separately; these sufficient to fix the datum in time and space. were part of his numerous Contributions to Terrestrial [87] A number of important additions have been made Magnetism. Sabine [1840, 1843, 1846, 1849, 1868, 1872, to this subset later, which came to serve as a resting place 1875, 1877] has provided the mainstay of the observa- for those individual processed voyages and small data col- tions contained in this subset. These were originally lections that resisted categorization under any of the other processed by J. Bloxham, and we have augmented them main headings. The most significant of these results stems with additional material relating to the covered voyages from the work done by collaborator J. Luque in Spain. The (see below). three main additions from his hand are the voyages of De [89] The Sabine subset covers the period 1773–1893 los Rios (1611, from a manuscript in the Biblioteca Nac¸io- and is based in part on published accounts of expedi- nal, Madrid), De Boenechea (1775, from a Spanish printed tions, both British and foreign, and on manuscripts to source), and Malaspina (1789–1793, from a document in which Sabine had access. Fortunately, Sabine has listed the Museo Naval, Madrid). The Malaspina records (ID his sources, albeit that some will be difficult to identify. code 4148) contain 28 Pacific inclinations, which comple- To aid future research, Sabine
s nomenclature and al- ment the declination data earlier compiled from this voy- phabetical order by author has been maintained. A se- age by A.M. (in ID code 3335). Finally, the published rious shortcoming of the compilation is the fact that logbook of La Barbinais Le Gentil (1727), describing a Sabine omitted many data when forming his compilation voyage in 1717, has also found a home in this subset. from the original sources. Since the world compilations Owing to the various additions at different times the he published in the 1870s are only for the zones 40ЊN– voyages are in no particular order. 85ЊN, 0Њ–40ЊN, and 0Њ–40ЊS, data from greater than 40ЊS 2-26 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Figure 22. Geographical data distribution of the Jackson subset. Here n ϭ 11,756; some points may overlap; projection is cylindrical equidistant.

have not been included (notwithstanding the data orig- and before this date most of the original results were still inating in the magnetic survey of the South Pole regions kept in the geomagnetism data archive of the British undertaken between 1840 and 1845, reported by Sabine Geological Survey, Edinburgh. Manuscript sources kept [1868]). We therefore used Sabine
s data set as the basis at this institute from the ships Alert, Dart, Flying Fish, for creating a new data set in which individual voyages Liverpool, Myrmidon, and Thalia have also found their are represented, and we used the original sources (both way into this subset. The most important oceanic mea- published and in manuscript) to reinstate missing data surements of all three elements come, however, from from the voyages. Among the more significant additions two ships that were contemporaneously sent out on are PRO ADM 53/142 (logbook Blossom), PRO ADM scientific expeditions, HMS Challenger [Nares, 1882] and 55/73 (logbook Herald), and the publications by von SMS Gazelle [Hydrographisches Bureau, 1888–1890], which Humboldt (1814–1829), Erman [1835, 1841], Fitzroy both measured D, I, and F on their circumnavigations. [1839], Aubert du Petit-Thouars (1840–1843), and Vail- [92] In a geographical sense the Jackson subset is lant (1840–1852). particularly strong in high latitudes (see Figure 22). The [90] Another important voyage was the one by Ross late part of the nineteenth century saw much activity in and Crozier, in the ships Erebus and Terror (1839–1843). the Arctic because of the search for the Northwest Sabine [1843, 1868, 1872, 1875, 1877] did extensive work on this material; it is mentioned in his contributions III, Passage from the Atlantic to the Pacific and the quest for V, VI, X, and XI, and the compilation by latitude is the North Pole. Arctic exploration records that were mentioned in contributions XIII–XV. The original data relied upon include the German expeditions under K. from the Erebus nowadays reside in PRO ADM 55/48- Koldewey and F. Hegemann of 1868 and 1869 in the 53. Regrettably, following Sabine
s several publications, Germania and Hansa; the Austro-Hungarian expedition the data have become spread over five separate ID codes of 1872 under C. Weyprecht and J. von Payer in the in the database. These are 4247 (Crozier, 1840–1843), Tegetthoff; the Swedish Arctic expedition of 1872–1873; 4274 (Ross, 1841), 4275 (Erebus, 1839–1841), 4429 the British Arctic expedition of 1875 led by G. S. Nares (Ross, 1827–1849), and 4456 (Terror, 1840–1841). in the Alert and Discovery; the Norwegian North Atlantic expedition (1876–1878) of C. Wille and H. Mohn; the 5.2.6. Jackson expedition of the Vega under A. E. Nordenskio¨ld in [91] The Jackson subset was compiled by one of us 1878–1880; the Greenland expeditions of 1891–1892 and (A.J.); it is based to a large extent on published sources 1891–1893; the voyage of the Manche in 1892; and F. covering the interval 1860–1900 [Jackson, 1989]. This Nansen
s Norwegian North Pole expedition from 1893 period saw the increasing importance of charts of the to 1896 in the Fram. American expeditions include the geomagnetic elements, especially declination, for navi- U.S. Arctic expedition of 1871–1873 led by B. Hall, the gation. The British Admiralty published charts of D from Lady Franklin Bay expedition of 1881–1884 under A. W. 1858 onward, but unfortunately, most of the data on Greely, and the subsequent Greely Relief expedition. In which the charts were based remain unpublished. How- contrast, there was practically no exploration in the ever, post-1890 data were published [Admiralty, 1901], Antarctic during this period. 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-27

Figure 23. Geographical data distribution of the Veinberg subset. Here n ϭ 13,832; some points may overlap; projection is cylindrical equidistant.

[93] Regional compilations of land-based data have [95] This extensive data set has been put into ma- furthermore been used in four areas: the United States, chine-readable form by V. N. Vadkovsky of World Data Canada, North Africa, and western Russia. The latter Centre B2 (Moscow) and was communicated by him to was based on von Tillo
s [1881, 1885] catalogues, which workers in England in 1987. It has since been incorpo- contain measurements between latitudes 40Њ–70ЊN and rated in its entirety into the historical geomagnetic da- longitudes 16Њ–70ЊEin“European Russia”; these have tabase, per volume, under ID codes 4553–4555. The been omitted to prevent overlap with the catalogues of geographical distribution of this subset is shown in Fig- Sabine. Additional land surveys were taken from original ure 23. publications and covered the Philippines, Japan, Indo- nesia, east Brazil, and South Africa. In all, these multi- farious contributions have resulted in the largest subset 5.2.8. Hutcheson for the nineteenth century, comprising Ͼ 19,000 obser- [96] The Hutcheson subset, shown in Figure 24, was vations. provided by K. A. Hutcheson in the late 1980s and has been described by Hutcheson [1990] and Hutcheson and Gubbins [1990]. It contains many early voyages, notably 5.2.7. Veinberg concerning the English search for a Northwest Passage [94] B. P. Veinberg graduated from the Institute of under Wilkens, Lancaster, Wymouth, B. Hall, H. Hud- Physics at St. Petersburg University in the late nine- son, and W. Baffin. Other famous explorations include teenth century. Afterward, he embarked upon a large those of L. V. de Torres (1606) and the Dutch Pacific project to collect geomagnetic data from Russia and voyage of W. C. Schouten and I. Le Maire (1615–1616). beyond. Over 30 years later, his Catalogue of Magnetic This compilation represents one of the first attempts to Determinations in USSR and in Adjacent Countries from gather historical geomagnetic data for core field model- 1556 to 1926 was published by the Central Geophysical ing. As a result, both its temporal scope (1600–1700) Observatory at St. Petersburg (then Leningrad) in 1929– and its size (Ͻ 3000 points) are limited. This data set 1933. The three volumes contain original measurements constitutes another example of a compilation purely of all components, loosely ordered by geography, mostly made on the basis of published sources. In fact, all but spanning the period from the late nineteenth century up 425 observations are found in a single work by W. van to 1930. The catalogue is, in fact, the largest single Bemmelen. As the efforts of this nineteenth-century source representing the twentieth century in the histor- Dutch meteorologist appear in other compilations as ical database, with Ͼ 18,000 readings after 1899. It also well (see sections 5.2.2 and 5.2.3), we will devote some contains the largest number of absolute intensity obser- specific attention to his data set here. vations, all made in the horizontal plane. Veinberg [97] van Bemmelen
s Ph.D. thesis (mathematics and [1929–1933, Table 1] furthermore used special codes to physics, Leyden University, 1893) contained a recon- indicate the geographical zone (Z), the reference num- struction of the Earth
s surface magnetic field for the ber (R) of his sources, and each data point
s weight (W) period 1540–1690. It was born out of desire to amend (as attributed by Veinberg). earlier work by Hansteen [1819], in particular to fill the 2-28 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Figure 24. Geographical data distribution of the Hutcheson subset. Here n ϭ 2971; some points may overlap; projection is cylindrical equidistant.

remaining data gap for the seventeenth century. The same truncation of single minutes was applied to the Dutchman reconstructed four isogonic maps (for 1610, noon latitude. Longitude values were rounded to whole 1640, 1665, and 1680), based on ϳ150 land observations degrees. Only near known landfalls were the actual of magnetic declination and ϳ1000 made at sea, gath- coordinates copied to the minute. In some cases, no ered from 38 ship
s logbooks. Six years later, as director coordinates were recorded in the log, only course and of the Royal Magnetical and Meteorological Observa- distance run. In such instances, van Bemmelen had to tory at Batavia, he greatly expanded on this early work reconstruct the ship
s daily position through manual with a similar publication [van Bemmelen, 1899]. This calculation, based in part upon the Wind and Current time, he had compiled a total of 5276 magnetic measure- Map of the Hydrographic Office (London, 1872). Of ments (388 on land and 4888 at sea), gathered from 165 course, the accuracy of these corrections will, on the logbooks kept in Dutch, English, and French archives. whole, be less than that achieved by processing logbook These observations resulted in five isogonic maps, cov- data with the voyage editor. ering the area between 60Њ north and south latitude (for [100] Because the processing of van Bemmelen
s data the years 1500, 1550, 1600, 1650, and 1700). set has taken place in stages (because of the temporal [98] Looking in more detail at the Dutchman
s source restrictions applied to various compilation efforts, see references, it becomes clear that van Bemmelen only above), his voyages have been incorporated into the processed primary sources in little more than half the database in three separate subsets: total number of cases (89 out of 165); these documents 1. van Bemmelen
s numbers 2–15 (period were primarily perused in the Dutch National Archives 1529–1600) have ended up together in a single ID code (ARA) (The Hague), with smaller contributions from (4106) as part of Bloxham
s subset (see 5.2.3). No fur- abroad (mainly the British Library and logs kept at the ther longitudinal corrections have been applied to these De´poˆt de la Marine in Paris). As far as the remainder is data. (van Bemmelen
s number 1 contained Columbus
s concerned, apart from nine unreferenced voyages, the 1492–1493 discovery voyage, which was not based on data were extracted from printed accounts of voyages primary sources and has therefore been excluded.) and travels, such as early publications from the Hakluyt 2. van Bemmelen
s numbers 16–117 (period 1601– Society and the famous compilation by Purchas. Regard- 1695) have found a home in the Hutcheson subset, ing the latter, it is unfortunate that many different edi- under ID codes 4556–4657. Three separate voyages tions of this work (entitled Purchas His Pilgrimes) exist (added by van Bemmelen [1899, appendix] have later and that van Bemmelen failed to identify which partic- been included under ID code 4662. Again, no further ular one he relied upon. To complicate matters, Purchas longitudinal corrections have been applied. himself caused potential confusion by naming two of his 3. van Bemmelen
s numbers 118–149 (period 1696– other works Purchas His Pilgrimage (on world history) 1722) are part of Bloxham
s subset as separate ID codes and Purchas His Pilgrim (on religion). (see Table 1), either using van Bemmelen
s compilation [99] Moving on to the recorded data, magnetic mea- or an earlier (published) source (see section 5.2.3). As surements (of declination only) were averaged per day the voyage editor had not yet been developed at the time and deemed to be accurate up to tens of minutes. The that these data were processed, Bloxham applied his 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-29

Figure 25. Geographical data distribution of the Cartes and Plans subsets. ID code is 4664 (Bibliothe`que Nationale); n ϭ 35,760; some points may overlap; projection is cylindrical equidistant.

own (full or partial) corrections to the ship
s tracks, [104] Both documents originated at the De´poˆt des using less sophisticated software. Please consult Blox- Cartes et Plans de la Marine of the French Navy (Paris) ham [1986] for further details on these corrections. as part of an attempt to construct isogonic charts for the [101] van Bemmelen
s numbers 150–162 (period early nineteenth century, based on readings taken on 1721–1743) had been copied from Hansteen [1819] and French Naval and Hydrographic Service ships. Sadly, the have therefore been excluded. charts themselves have not been found, but the base [102] Finally, regarding the remainder of the grids containing the (listed or plotted) original measure- Hutcheson subset, the most interesting entry is ID code ments did survive. The largest of these now resides at the 4658, which itself represents a separate compilation of Bibliothe`que Nationale in Paris under manuscript cata- seventeenth-century data (including 37 values of I) from logue number /NAF 9460. Its title is Recueil des Varia- a variety of published sources (ranging from Halley tions de la Boussole, and it is believed to have been [1683] to Malin and Bullard [1981]), collated by D. Bar- compiled not long after the year 1870. This manuscript raclough in the 1980s and communicated to the Univer- contains measurements predominantly made during the sity of Leeds in 1987. (A full list of Barraclough
s sources eighteenth and nineteenth centuries. A list of the ships is given in Table 1 footnotes.) These data have been involved is given at the end of the document, but it is not supplemented with a few more points from contempo- clear whether or not this list is comprehensive. In addi- rary compilations of travel accounts such as those of tion, the manuscript includes a few seventeenth-century Purchas [1625]. The Indian voyage of Sir Thomas Roe as observations, made in European cities and ports. The recorded by Churchill and Churchill [1704] was originally data in this document are arranged in a geographical processed in part by Hutcheson too (six points). How- grid, divided by lines of latitude and longitude with a ever, the full record (27 points) has been reprocessed by spacing of 1Њ. Values of magnetic declination, given in one of us (A.M.) and now forms part of the Murray whole degrees or degrees and minutes, together with the compilation instead under ID code 3153. year in which they were observed, were recorded in the appropriate grid cell. A total of 35,760 observations from 5.2.9. Cartes and Plans this document has been incorporated into the database; [103] The last subset to be described here is the largest their geographical distribution is shown in Figure 25. of all, comprising Ͼ45,000 data points, extracted from [105] The second compilation is slightly more concen- two French manuscripts (located by A.R.T.J., largely trated in time, containing readings for the interval 1719– processed by A.M.). Its character combines aspects of 1857. It concerns Observations des Variations de l
Aiguille several previously discussed compilations. It consists Aimante´e, possibly compiled shortly after 1850, and now solely of unpublished declination measurements, is geo- kept in the French Archives Nationales (Paris) under graphically ordered, and covers large parts of the globe. catalogue references ANP MAP 6JJ/80 a–b and 6JJ/81 Furthermore, the data center around a specific time a–c. These manuscripts originated in the Service Hy- window within the nineteenth century (1820–1850 in this drographique and comprise a series of large plans which, case, although the complete range spans the years 1594 like the Bibliothe`que Nationale document, present dec- to 1899 inclusive). lination values and the year of observation within a 2-30 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Figure 26. Geographical data distribution of the Cartes and Plans subsets. ID code is 4663 (Archives Nationales); n ϭ 9576; some points may overlap; projection is cylindrical equidistant.

graticule of latitude and longitude. In this case, however, necessary to complete this latest collection effort, any the grid is divided into squares representing 10 min of future endeavor to process more English or French latitude and longitude each rather than whole degrees. logbooks from this period will have to balance new Again, the declination values are given in degrees and investments against potentially diminished returns. minutes and are accompanied by the year of observa- Nearly all Dutch material before 1800 has now been tion. Although a listing of the particular ships supplying processed, while Iberian sources have proved disap- the data was not given, it is clear from a color-coded pointing so far. Moreover, about half the remaining legend included with the plans that some were employed French manuscripts and about a quarter of the surviving by the Service Hydrographique (marked in red and eighteenth-century English logbooks have been seen. identified as such in the compilation). There are 9576 However, much of the data are repetitive, traversing useful observations in this document, shown in Figure (Atlantic and Indian) oceanic areas already more than 26. adequately covered by the present collection. [106] Since the data were not recorded in their origi- [108] In our opinion, subsequent investigations should nal manuscript form but had been transcribed onto a focus primarily on unearthing more material prior to geographical grid, there is a quantization error in the 1700, in particular land-based surveys, measurement se- observation positions that must be taken into account. A ries, and scattered observations from poorly represented second problem concerns duplicate points, 1709 of regions (basically all continents except Europe). As far which were identified and eliminated (before the above as nautical sources are concerned, the whole of the totals were computed). Another consequence of the Pacific is still severely underrepresented in the sample, carrier format is the fact that the original logbook con- and data from any period would enhance the resolution text of individual voyages has been lost. Longitudinal of core field models for this hemisphere. Two other corrections of ship
s tracks, based on identified geo- areas poorly covered so far are the polar latitudes (the graphical landmarks, was therefore not possible. higher the latitude, the better); these too would benefit from any additional information. [109] Of course, this raises the question of where this 6. FUTURE WORK AND CONCLUSIONS material might be found. As section 4.2 made clear, our investigations have mainly taken place in England, [107] The preceding description and Table 1 represent France, and the Netherlands, with smaller archival ven- the most comprehensive attempt to date to catalogue tures into Denmark and Spain. Within Europe, archives and reference geomagnetic observations from historical in Portugal, Sweden, and Russia are perhaps the most records. Although several thousand eighteenth- and promising new territories to explore (even though much nineteenth-century logbooks have not yet been pro- Portuguese material is known to be no longer extant). cessed in this manner, we believe that the large majority We also remind the reader that several hundred Danish of the available primary sources in northwest Europe whaling logs (once properly inventoried) could enhance before 1700 have now been entered into the database. eighteenth-century Arctic coverage from Davis Strait to Given the substantial expenditure in time and resources Spitsbergen. Furthermore, new material may await pro- 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-31 cessing once the cataloguing efforts at the Hydrographic la Noticias Hidrograficas no. 12, Santiago, 1890), 12, 247, Office at Taunton have been completed. Outside of 1890. Europe, repositories in former colonies of European Annales Hydrographiques, Resultats des observations de de´- clinaison magne´tique faites en 1893 et en 1894 sur la coˆte de maritime powers may contain rich holdings as well. We Norve`ge, entre Christiania et Skolver (Efterretninger for mention again the lost logbooks of the Spanish Manilla Sjøfarende, no. 3/139, Christiania (Oslo), March 1895), 17, Galleon, which completed annual Pacific crossings be- 95, 1895. tween South America and the Philippines for many “Arethuse”, Resultats des me´sures magne´tiques effectue´es, decades. Remnants of West Indian merchant company Ann. Hydrogr., 10, 600–625, 1888. Asher, G. M. (Transl./Ed.), Henry Hudson the Navigator: The records may furthermore still exist in various places Original Documents in Which His Career Is Recorded,HS along South American coasts. ser. I, vol. 27, Hakluyt Soc., London, 1860. [110] These suggestions, in combination with the pre- Aubert du Petit-Thouars, A., Voyage autour du monde sur la ceding description of compilations, underline the fact fre´gate la Ve´nus, pendant les anne´es 1836–1839, relation, 4 that the quest to find historical geomagnetic data has vols., Gide, Paris, 1840–1843. Aubry, Note sur les observations magne´tiques faites` a bord de been, and will continue to be, an ongoing concern for la Minerve (1886–1888) par M. Aubry, Lieutenant de Vais- geomagnetists, maritime historians, and historians of seau, Ann. Hydrogr., 10, 290–303, 1888. science alike. Consecutive compilations have greatly Barr, W., and G. Williams (Eds.), Voyages to Hudson Bay in benefited from the work of predecessors, and we firmly Search of a Northwest Passage 1741–1747, H. S. Ser. II, vol. hope that our contribution will itself both stimulate and 1, no. 177, The Voyage of Christopher Middleton 1741–1742, Hakluyt Soc., London, 1994. advance future geomagnetic research by others. The Barraclough, D. R., Spherical harmonic models of the geomag- potential rewards of enhancing coverage and under- netic field, Inst. Geol. Sci. Geomagn. Bull., 8, Her Majesty
s standing of the Earth
s magnetic field in former times Stationery Off., London, 1978. should prove a sufficient and lasting incentive to that Barraclough, D. R., Historical observations of the geomagnetic effect. field, Philos. Trans. R. Soc. London, Ser. A, 306,71–78, 1982. Barraclough, D. R., Halley
s Atlantic magnetic surveys, in Historical Events and People in Geosciences, edited by W. Schro¨der, Peter Lang, New York, 1985. [111] ACKNOWLEDGMENTS. We have benefited from Barraclough, D. R., Observations of the Earth
s magnetic field the help of a large number of colleagues in the course of this in Edinburgh from 1670 to the present day, Trans. R. Soc. work. Jeremy Bloxham, David Barraclough, and Ken Edinburgh Earth Sci., 85, 239–252, 1995. Hutcheson have shared their data and expertise with us, for Barraclough, D. R., J. G. Carrigan, S. R. C. Malin, Observed which we are very grateful. We are furthermore indebted to geomagnetic field intensity in London since 1820, Geophys. V. N. Vadkovsky of World Data Centre B2 (Moscow) for J. Int., 141,83–99, 2000. communicating Veinberg
s data set. Mioara Mandea, Jose´ Bauer, L. A., The earliest values of the magnetic declination, J. Geophys. Res., 13,97–104, 1908. Luque, and Stuart Humber have all helped with data collec- Beattie, J. C., Report of a Magnetic Survey of South Africa, Univ. tion, verification, and homogenization. We thank Nick Barber Press, Cambridge, U.K., 1909. for his development of the voyage editor software. We would Becquerel, A. C., Traite´ experimental de l
electricite´ et du mag- also like to mention Karel Davids, Erik Gøbel, David Gubbins, ne´tisme, Fermin Didot Fre`res, Paris, 1840. Willem Hackmann, Alan Lemmers, Jeff Love, Willem F. J. Bernardie`res, M., Re´sume´ des de´terminations magne´tiques ef- Mo¨rzer Bruyns, and Glyndwr Williams for useful discussions. fectue´es en 1882–1883 par la mission charge´edel
observation Finally, we extend our gratitude to all members of staff at du passage de Venus au Chili, et de la mesure de diffe´rences de
museums, libraries, archives, and other repositories every- longitude sur la coˆte occidentale de l Ame´rique du sud, Extr. where for providing information on and access to the historical Annal. Hydrogr., 1er semestr. Imprimerie Natl., Paris, 1884. material. This work was supported by NERC grants GR9/ Bessels, E., Scientific Results of the U.S. Arctic Expedition, Steamer Polaris, vol. 1, Physical Observations, Dep. of State 01848 and GR3/10581 and the Royal Society; ARTJ was ad- and Public Inst., Navy Dep., Washington, D. C., 1876. ditionally supported by the Foundation for Historical Sciences Bloxham, J., Models of the magnetic field at the core-mantle with financial aid from the Netherlands Organisation for Sci- boundary for 1715, 1777, and 1842, J. Geophys. Res., 91, entific Research (NWO). 13,954–13,966, 1986. [112] Louise Kellogg was the Editor responsible for this Bloxham, J., and D. Gubbins, The secular variation of Earth
s paper. She thanks an anonymous cross-disciplinary reviewer magnetic field, Nature, 317, 777–781, 1985. and an anonymous technical reviewer. Bloxham, J., and A. Jackson, Simultaneous stochastic inversion for geomagnetic main field and secular variation II, 1820– 1980, J. Geophys. Res., 94, 15,753–15,769, 1989. Bloxham, J., D. Gubbins, and A. Jackson, Geomagnetic secular REFERENCES variation, Philos. Trans. R. Soc. London, Ser. A, 329, 415– Admiralty, List of magnetic observations made by the officers 502, 1989. of HM ships and Indian Marine Survey during the years Borchgrevink, C. E., Magnetic and Meteorological Observations 1890–1900, Hydrographic Dep., Admiralty, London, 1901. Made by the
Southern Cross
Antarctic Expedition, 1898– Alexandrescu, M., V. Courtillot, J.-L. Le Moue¨l, Geomagnetic 1900, Harrison and Sons, London, 1902. field direction in Paris since the mid-sixteenth century, Phys. Brewster, D., A Treatise on Magnetism, Adam and Black, Ed- Earth Planet. Inter., 98, 321–360, 1996. inburgh, 1837. Annales Hydrographiques,De´terminations magne´tiques effec- Bruijn, J. R., F. S. Gaastra, and I. Scho¨ffer, Dutch-Asiatic tue´es en 1889 dans diverses localite´s du Chili par les As- Shipping, vols. 165–167, Rijks Geschiedkd. Publ., The tronomes de l
observatoire National de Santiago (Extrait de Hague, 1987. 2-32 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Cafarella, L., A. De Santis, and A. Meloni, Secular variation in Delano, A., A Narrative of Voyages and Travels, in the Northern Italy from historical geomagnetic field measurements, Phys. and Southern Hemispheres: Comprising Three Voyages Earth Planet. Inter., 73, 206–221, 1992. Round the World; Together With a Voyage of Survey and Carlheim-Gyllensko¨ld, V., Sur la forme analytique de Discovery, in the Pacific Ocean and Oriental islands, Boston, l
attraction magne´tique de la Terre exprime´e en fonction Mass., 1817. du temps, Astron. Iakttag. Unders. Anstalda Stockholms Delisle, G., Histoire de l
Acade´mie des Sciences (…), Paris, Obs., 5(3), 29, 1896. 1721. Cassini, J. D., De la de´clinaision et des variations de l
aguille de Moidrey, J., Note sur quelques ancie`nnes de´clinaisons, J. aimante´e, J. Phys.,52–55, 1791. Geophys. Res., 9,18–24, 1904. Cawood, J., Terrestrial magnetism and the development of Douglas, R., The variation of the compass; containing (a) 1719 international collaboration in the early nineteenth century, observations to, in, and from, the East Indies, Guinea, West Ann. Sci., 34, 551–587, 1977. Indies, and Mediterranean, with the latitudes and longi- Cawood, J., The Magnetic Crusade: Science and politics in tudes at the time of observation (…), Philos. Trans. R. Soc. early Victorian Britain, Isis, 70(254), 493–518, 1979. London, 66,18–72, 1776. Chief Astronomer, Report of the Chief Astronomer for the Drygalski, E. V., Gro¨nland-Expedition der Gesellschaft fu¨r Erd- year ending March 31 1909, Dep. Inter. Sessional Pap. 25a, kunde zu Berlin (1891–93), vol. 2, W. H. Kuhl, Berlin, 1897. Dep. of the Inter. Ottawa, Ont., Canada, 1910. Duperrey, L. I., Voyage autour du monde, exe´cute´ par ordre du Christy, M. (Ed.), The Voyages of Captain Luke Foxe of Hull, Roi, sur la corvette de Sa Majeste´ La Coquille, pendant les and Captain Thomas James of Bristol, in Search of a North- anne´es 1822, 1823, 1824 et 1825, sous le ministe`re et con- West Passage, in 1631–32; With Narratives of the Earlier forme´ment aux instructions de S. E. M. le marquis de Cler- North-West Voyages of Frobisher, Davis, Weymouth, Hall, mont-Tonnerre (…), 4 vols., Navigation et hydrographie, at- Knight, Hudson, Button, Gibbons, Bylot, Baffin, Hawkridge, las, Minist. de la Mar. et des Colonies, Paris, 1826–1827. and Others, HS Ser., vol. I, nos. 88–89, Hakluyt Soc., Lon- Erman, G. A., Reise um die Erde durch Nord-Asien und die don, 1894. beiden Oceane in den Jahren 1828, 1829 und 1830 ausgefu¨hrt Churchill, A., and J. Churchill (Eds.), A Collection of Voyages von A. Erman, Abteilung I: Historischer Bericht, 3 vols., G. and Travels, Some Now First Printed From Original Manu- Reimer, Berlin, 1835. scripts. Others Translated Out of Foreign Languages, and Erman, G. A., Reise um die Erde durch Nord-Asien und die Now First Publish
d in English. To Which Are Added Some beiden Oceane in den Jahren 1828, 1829 und 1830 ausgefu¨hrt Few That Have Formerly Appear
d in English, but Do Now von A. Erman, Abteilung 2: Physikalische Beobachtungen,G. (…) Deserve To Be Reprinted. With a General Preface Giving Reimer, Berlin, 1841. an Account of the Progress of Navigation, 6 vols., Churchill Fanning, E., Voyages and Discoveries in the South Seas 1792– and Churchill, London, 1704. 1832, vol. 5, Mar. Res. Soc., Salem, Mass., 1924. Cirera, P. R., El Magnetismo Terrestre en Filipinas, Obs. Me- Favereau, De´terminations magne´tiques faites dans l
Oce´an teorol. de Manila, Manila, 1893. Indien en 1884–1885, par M. Favereau, Lieutenant de Clarke, J. S., The Progress of Maritime Discovery, From the Vaisseau, Ann. Hydrogr., 8, 445–449, 1886. Earliest Period to the Close of the 18th Century, Forming an Feuille´e, L., Journal des observations physiques, mathe´matiques Extensive System of Hydrography, vol. 1, A. Strahan, London, et botaniques faites par ordre du Roi sur les coˆtes Orientales de 1803. l
Amerique Meridionale et aux Indes Occidentales, 2 vols., Cook, J., J. King, and W. Bayly, The original Astronomical Pierre Giffart, Paris, 1714. (Reprinted by Jean Mariette, Observations Made in the Course of a Voyage to the Northern Paris, 1735.) Pacific Ocean, Commiss. of the Longitude, London, 1782. Fitzroy, R., Narrative of the Surveying Voyages of HMS Adven- Coreal, F., Voyages de Franc¸ois Coreal (…) avec une relation de ture and Beagle, Between the Years 1826 and 1836, Describing la Guiane de Walter Raleigh et le voyage de Narborough a`la Their Examination of the Southern Shores of South America, Mer du Sud par le de´troit de Magellan (…), vol. 3, Amster- and the Beagle
s Circumnavigation of the Globe, 3 vols., dam, 1722. Henry Colburn, London, 1839. Cornwall, Observations of the variation on board the Royal Freeden, W. V., Die wissenschaftlichen Ergebnisse der ersten African Pacquet in 1721, Philos. Trans. R. Soc. London, 32, Deutschen Nordfahrt 1868, Geogr. Erforsch. Polar-Reg. 27 55–56, 1722. Justus Perthes, Gotha, Germany, 1869. Courmes, L., Observations magne´tiques faites pendant la cam- Fritsche, H., U¨ber die Bestimmung der geographischen La¨nge pagne du croiseur Le Dubordieu (1889–1891), par M. L. und Breite und der drei Elemente des Erdmagnetismus durch Courmes, Lieutenant de Vaisseau, Ann. Hydrogr., 14, 302– Beobachtung zu Lande sowie erdmagnetische und geogra- 349, 1892. phische Messung an mehr als tausend verschiedenen Orten in Creak, E. W., On the results of the magnetical observations Asien und Europa, ausgefu¨hrt in den Jahren 1867–1891, St. made by the officers of the Arctic Expedition 1875–76, Proc. Petersburg, Russia, 1893. R. Soc. London, 29,29–42, 1879. Gauss, C. F., Allgemeine Theorie des Erdmagnetismus, in Cunningham, J., Observations of the weather, made in a voy- Resultate aus den Beobachtungen des magnetischen Verein im age to China, Philos. Trans. R. Soc. London, 24, 1639–1674, Jahre 1838, pp. 1–52, Leipzig, Go¨ttingen Magn. Ver., Ger- 1704. many, 1839. Dalrymple, A., A Collection of Voyages, Chiefly in the South Geelmuyden, H., Astronomical Observations (section D), in Atlantick Ocean, 5 parts, London, 1775. The Norwegian North Polar Expedition 1893–1896: Scientific Dalrymple, A., Journal of a voyage to the East Indies, in the Results (1900–06), edited by F. Nansen, vol. 2, no. 6, Fridt- ship Greville, Captain Burnet Abercrombie, in the year jof Nansen Fund for the Adv. of Sci., New York, 1901. 1775, Philos. Trans. R. Soc. London, 68, 389–418, 1778. Greely, A. W., International Polar Expedition: Report on the Dampier, W., A Voyage to New Holland, James and John Proceedings of the U.S. Expedition to Lady Franklin Bay, Knapton, London, 1729. (Reprinted in Dampier
s Voyages, Grinnel Land, vol. 2, Gov. Print. Off., Washington, D. C., edited by J. Masefield, Grant Richards, London, 1906.) 1888. Davidson, G., Pacific Coast: Coast Pilot of California, Oregon, Hakluyt, R., Principall Navigations, Voiages and Discoveries of and Washington Territory, Coast and Geod. Surv., U.S. Dep. the English Nation …, G. Bishop and R. Newberie, deputies of State and Public Inst., Washington, D. C., 1869. to C. Barker, printer to the Queen, London, 1589. 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-33

Halley, E., A theory of the variation of the magnetical com- with a summary of previous magnetic work in northern pass, Philos. Trans. R. Soc. London, 13, 208–221, 1683. Africa, Surv. Dep. Pap. 6, Minist. of Finan., Egypt, Cairo, Halley, E., An account of the cause of the change of the 1907. variation of the magnetic needle, with an hypothesis of the Kircher, A., Magnes sive de Arte Magnetica (Opus Tripartitum) structure of the internal parts of the Earth, Philos. Trans. R. 3rd ed, Mascardi, Rome, 1654. Soc. London, 16, 563–578, 1692. Knott, C. G., and A. Tanakadate. A magnetic survey of all of Halley, E., Observations of latitude and variation, taken on Japan by order of the President of the Imperial University, board the Hartford, in her passage from Java Head to St. J. Coll. Sci. Imp. Univ., Tokyo, II(III), 1888. Helena, Philos. Trans. R. Soc. London, 37, 331–338, 1732. Koldewey, K., Die Zweite Deutsche Nordpolarfahrt 1869–70, Hansteen, C., Untersuchungen u¨ber den Magnetismus der Erde, vol. 2, F. A. Brockhaus, Leipzig, Germany, 1874. Lehmann and Gro¨ndahl, Christiania, Norway, 1819. La Barbinais Le Gentil, Nouveau voyage autour du monde; par Harris, J., Navigantium atque Itinerantium Bibliotheca: or, a Monsieur Le Gentil, Avec une description de l
Empire de la Compleat Collection of Voyages and Travels, Consisting of Chine beaucoup plus ample & plus circonstancie´e que celles Above Four Hundred of the Most Authentick Writers, 2 vols., qui ont parues jusqu
a` present, ou` il est traite´ des moeurs, T. Bennet, J. Nicholson, and D. Midwinter, London, 1705. religion, politique,´ education & commerce des peuples de cet Harris, J., An account of some magnetical observations made Empire, Flahault Paris, 1727. in the months of May, June and July 1731 in the Atlantick Lancelin, G., Observations de de´clinaison magne´tique dans la or Western Ocean, Philos. Trans. R. Soc. London, 38,75–79, Me´diterrane´e orientale (Feb–May 1894) par M. G. Lance- 1733. lin, Enseigne de Vaisseau` a bord du Croiseur le Cosmao, Hazard, D. L., United States magnetic tables and magnetic Ann. Hydrogr., 16, 141–143, 1894. charts for 1915, U.S. Coast Geod. Surv. Spec. Publ. 44, U.S. Langel, R. A., The main field, in Geomagnetism, vol. 1, edited Coast and Geod. Surv., Washington, D. C., 1917. by J. A. Jacobs, pp. 249–512, Academic, San Diego, Calif., Heeres, J. E., 1899. Het aandeel der Nederlanders in de 1987. ontdekking van Australie¨ 1606–1765, K. Ned. Aardrijksk. Le Cannellier, Observations magne´tiques faites dans diffe´rents Genoots., Leyden, Netherlands, 1917. mouillages de la Me´diterrane´e Orientale, par M. Le Can- Herbert, W. H., and J. H. McKnight, Magnetic results in nellier, Lieutenant de Vaisseau` a bord du Cuirasse western Canada, Bull. 52, Topogr. Surv. of Canada, Ottawa, d
escadre Amiral-Baudin, Ann. Hydrogr., 12, 240–245, 1890. Ont., 1924. Leconte de Roujou, De´terminations magne´tiques en extreˆme- Houette and Morache, Mission magne´tique en Islande et en Orient par M. Leconte de Roujou, Lieutenant de Vaisseau Scandinavie, de l
Aviso- transport La Manche, par M. Hou- `a bord de la Triomphante, Ann. Hydrogr., 14, 112–155, 1892. ette, Capitaine de Fre´gate, Commandant, et M. Morache, Lephay, J., Extrait des observations magne´tiques faites dans le Lieutenant de Vaisseau` a bord de ce baˆtiment, Ann. Hy- Levant en 1885–1886 par M. J. Lephay, Lieutenant de drogr., 18,49–68, 1896. Vaisseau` a bord de La Venus, Ann. Hydrogr., 11, 49, 1889. Howay, F. W. (Ed.), Voyages of the Columbia to the Northwest Leroux, E. (Ed.), Voyage de La Manche a`l
ˆIle Jan Mayen et au Coast, 1787–1790 and 1790–1793, vol. 79, Mass. Hist. Soc., Spitzberg, E. Leroux, Paris, 1894. Boston, 1941. Loschober, F., De´terminations magne´tiques faites en 1889– Hoxton, W., The variation of the magnetic needle, as observed 1890, et ramene´s au commencement de 1890, sur les coˆtes in three voyages from London to Maryland, Philos. Trans. de la Mer Adriatique, par le capitaine de corvette Franz R. Soc. London, 41, 171–175, 1742. Loschober, Directeur de l
observatoire Impe´rial et Royal Hunnewell, J., Journal of the Voyage of the Missionary Packet, de Pola, (Hydrographische Nachricht, no. 28, Pola, 1891.), Boston to Honolulu, 1826 (…) With Maps and Plates, and a Ann. Hydrogr., 13, 266–267, 1891. Memoir, Charlestown, Mass., 1880. Lous, C. C., Versuche und Vorschla¨ge betreffend die Theorie der Hutcheson, K. A., Geomagnetic field modelling, Ph.D. thesis, Navigation, Kiel, Germany, 1795. Univ. of Cambridge, Cambridge, U.K., 1990. Lundquist, K. Z., Magnetic observations in Svålbard 1596– Hutcheson, K. A., and D. Gubbins, Earth
s magnetic field in 1953, Skr. 110, Norsk Polarinst., Det K. Dep. for Ind. og the seventeenth century, J. Geophys. Res., 95, Håndverk, Oslo, 1957. 10,769–10,781, 1990. Malaspina, A., Viaje pol´ıtico-cient´fiı co alrededor del mundo por Hydrographisches Bureau, Die Forschungsreise SMS “Gazelle” las corbetas Descubierta y Atrevida al mando de los capitanes in den Jahren 1874 bis 1876 unter Kommando des Kapita¨n zur de navo D. Alejandro Malaspina y Don Jose´ de Bustamante y See Freiherrn von Schleinitz, 5 vols., part II, Physik und Guerra desde 1789 a´ 1794, 2nd ed., Impr. de la viuda el hijos Chemie, Hydrogr. Amt des Reichs-Mar. Amts, Ger. Dep. of de Abienzo, Madrid, 1885. State and Public Inst., Admiralty, Berlin, 1888–1890. Malin, S. R. C., Sesquicentenary of Gauss
s first measurement Jackson, A., The Earth
s magnetic field at the core-mantle of the absolute value of magnetic intensity, Philos. Trans. R. boundary, Ph.D. thesis, Univ. of Cambridge, Cambridge, U. Soc. London, Ser. A, 306,5–8, 1982. K., 1989. Malin, S. R. C., Historical introduction to geomagnetism, in Jackson, A., A. R. T. Jonkers, and M. Walker, Four centuries Geomagnetism, vol. 1, edited by J. A. Jacobs, pp. 1–49, of geomagnetic secular variation from historical records, Academic, San Diego, Calif., 1987. Philos. Trans. R. Soc. London, Philos. Trans. Math. Phys. Malin, S. R. C., and E. C. Bullard, The direction of the Earth
s Eng. Sci., 358, 957–990, 2000. magnetic field at London, 1570–1975, Philos. Trans. R. Soc. Jonkers, A. R. T., The fading tradition: The declining use of London, Ser. A, 299, 357–423, 1981. magnetic declination in Dutch pelagic navigation, in Frutta Marion du Fresne, M.-J., Extracts From Journals Relating to the di Mare: Evolution and Revolution in the Maritime World in Visit to New Zealand in May–July 1772 of the French Ships the 19th and 20th Centuries, edited by P. C. van Royen, L. R. Mascarin and Marquis de Castries Under the Command of Fisher, and D. M. Williams, Batavian Lion Int., Amster- M.-J. Marion du Fresne, transcription and translation from dam, 1998. French by I. Ollivier, Alexander Turnbull Libr. Endow. Jonkers, A. R. T., North by northwest: Seafaring, science, and Trust with Indosuez N. Z. Ltd., Wellington, N. Z., 1985. the Earth
s magnetic field, Ph.D. thesis, Free Univ., Am- Markham, C. R. (Ed.), The Voyages of Sir James Lancaster, Kt., sterdam, 2000. to the East Indies, With Abstracts of Journals of Voyages to the Keeling, B. F. E., Magnetic Observations in Egypt 1895–1905, East Indies During the Seventeenth Century, Preserved in the 2-34 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

India Office. And the Voyage of Captain John Knight (1606), the years 1819–20, in His Majesty
s Ships Hecla and Griper, to Seek the North-West Passage, HS Ser. I, vol. 56, Hakluyt Under the Orders of William Edward Parry (…) With an Soc., London, 1877. Appendix, Containing the Scientific and Other Observations, Markham, C. R. (Ed.), The Voyages of William Baffin, 1612– Lords Comm. of the Admiralty, London, 1821. 1622, From Narratives and Journals by John Gatonbe, Robert Pickersgill, R., Track of His Majesty
s armed brig Lion from Fotherby, and Others, With Baffin
s Letters, Journals, and England to Davis
s Streights and Labrador, with observa- Other Observations, and Various Treatises on the Probability tions for determining the longitude by Sun and Moon and of a North-West Passage, HS Ser. I, vol. 63, Hakluyt Soc., error of common reckoning; also the variation of the com- London, 1881. pass and dip of the needle as observed during the said Maurelle, F. A., Voyage of the Sonora in the Second Bucareli voyage in 1776, Philos. Trans. R. Soc. London, 68, 1057– Expedition (…) The Journal (…) Translated by the Hon. 1063, 1778. Daines Barrington (…) Reprinted Line for Line (…) From Pond, C. F., Oce´an Pacifique Nord, Californie, Observations Barrington
s Miscellanies, 1781. With concise Notes (…)by de de´clinaison magne´tique, d
inclinaison et d
intensite´, Thomas C. Russell, etc., T. C. Russell, San Francisco, Calif., faites par le Lieutenant Charles F. Pond, du steamer hy- 1920. drographe des E´tats-Unis, Commander F. A. Cooke, en Middleton, C., Observations on the weather, in a voyage to 1887–1888, Ann. Hydrogr., 10, 626, 1888. Hudson
s Bay, Philos. Trans. R. Soc. London, 37,76–78, Pond, C. F., De´terminations magne´tiques faites sur les coˆtes 1731. ouest de la basse Californie par le Lieutenant Charles F. Middleton, C., Observations of the variations of the needle and Pond, de la marine des E´tats-Unis,` a bord du navire hy- weather made in a voyage to Hudson
s Bay, Philos. Trans. drographe Ranger, (Extrait de la Notice to Mariners no. 47 R. Soc. London, 38, 127–133, 1733. de Washington 1889.), Ann. Hydrogr., 11, 233, 1889. Middleton, C., Observations made of the latitude, variation of Pond, C. F., Oce´an Pacifique Nord. Basse Californie (coˆte the magnetic needle, and weather, Philos. Trans. R. Soc. ouest). De´terminations magne´tiques faites dans la baie London, 39, 270–279, 1736. Ballenas et sur l
ˆIle Ascuncion, par le Lieutenant Charles F. Mion, M., De´terminations magne´tiques`laco a ˆte ouest de Pond,` a bord du navire hydrographe Ranger, (Notice to Madagascar, et aux Comores, par M. Mion, Ingenieur hy- Mariners no. 50 Washington 1890.), Ann. Hydrogr., 13, 86, drographe de la Marine, 1888–1890, Ann. Hydrogr., 15, 1890. 205–207, 1893. Pote´rat, Journal d
un Voyage au Cap de Horn, au Chili, au Mizon, L., Observations magne´tiques recueillies`laco a ˆte oc- Pe´rou, aux Iles Philippines, et a`lacoˆte de la Nouvelle-Espagne cidentale d
Afrique par M. L. Mizon, Lieutenant de Vais- (…) pendant les anne´es 1795–1800, F. Didot, Paris, 1815. seau, Ann. Hydrogr., 11,36–47, 1889. Preston, E. D., Oce´an Atlantique: De´terminations magne´t- Morrell, B., A Narrative of Four Voyages to the South Sea, North iques faites par M. E. D. Preston, Assistant du Coast et and South Pacific Ocean, Chinese Sea, Ethiopic and Southern Geodetic Survey, pendant un voyage du navire Pensacola Atlantic Ocean, Indian and Antarctic Ocean, From (…) 1822 dans l
oce´an Atlantique, 1890–1891, Ann. Hydrogr., 13, 268, to 1831, New York, 1832. 1891. Mottez, L., De´terminations magne´tiques faites sur la coˆte Purchas, S., Purchas his Pilgrimes: In Five Bookes. The First, ouest d
Ame´rique pendant une campagne dans l
oce´an Contayning the Voyages (…) Made by Ancient Kings, (…) Pacifique` a bord du croiseur Le Dubordieu (1892–1893) par and Others, to and Thorow the Remoter Parts of the Knowne M. Luis Mottez, Lieutenant de Vaisseau, Ann. Hydrogr., 15, World, etc., 4 parts, H. Fetherstone, London, 1625. 408–431, 1893. Purchas, S., Hackluytus Posthumus (or) Purchas His Pilgrimes: Mountaine, W., A letter from Mr. William Mountayne, F. R. S. Containing a History of World in Sea Voyages and Lande to the Rt. Hon. James Earl of Morton, PRS, Philos. Trans. Travells by Englishmen and Others, 20 vols., J. Maclehose R. Soc. London, 56, 216–223, 1776. and Sons, Glasgow, U. K., 1905–1907. Mountaine, W., and J. Dodson, A letter to the Right Honour- Ramusio, G. B., Navigazioni et Viaggi, 1563–1606; (Facsimile able the Earl of Macclesfield, President of the Council and ed., Theatrum Orbis Terrarum, Amsterdam, 1970). Fellow of the Royal Society, concerning the variation of the Raulin, V., Quelques vues generales sur les variations se´cu- magnetic needle; with a set of tables annexed, which exhibit laires du magnetisme terrestre, Actes Soc. Linn. Bordeaux, the results of upwards of fifty thousand observations, in six 26,1–92, 1867. periodic reviews, from the year 1700 to the year 1756, both Rayet, M. G., Recherches sur les observations magne´tiques inclusive; and are adapted to every 5 degrees of latitude and faites`l a
observatoire de Paris de 1667` a 1872, Mem. Ann. longitude in the more frequented oceans, Philos. Trans. R. Obs. Paris, 13, 1876. Soc. London, 50, 329–350, 1757. Rogers, W., The variation of the magnetical compass, observed Mulert, F. E., Scheepsjournaal gehouden op het schip Tienhoven by Captain Rogers, Philos. Trans. R. Soc. London, 31, 173– tijdens de ontdekkingsreis van Mr. Jacob Roggeveen (1721– 176, 1721. 1722), K. Zeeuws Genoot. van Wet., Middelburg, Nether- Ryder, C., Meddedelser om Grønland, Den Ostgrønlandske Ex- lands, 1911. pedition 1891–1892, Meteorol. Inst., Dep. of State and Pub- Nares, G. S., Report on the Scientific Results of the Voyage of lic Inst., Copenhagen, 1895. HMS Challenger during the years 1873–1876, under the com- Sabine, E., Experiments to ascertain the ratio of the magnetic mand of Captain George S. Nares RN FRS and Captain forces acting on a needle suspended horizontally, in Paris Frank Tourle Thomson RN, vol. 2, Narrative, The Stationary and in London, Philos. Trans. R. Soc. London, Part I, 118, Off., London, 1882. 1–14, 1828. Norwood, R., The Seaman
s Practice, Contayning a Fundamen- Sabine, E., Contributions to terrestrial magnetism, I, Philos. tall Probleme in Navigation Experimentally Verified; Namely Trans. R. Soc. London, 130, 129–155, 1840. Touching the Compasse of the Earth and Sea, and the Quan- Sabine, E., Contributions to terrestrial magnetism, V, Philos. tity of a Degree in our English Measures, Also an Exact Trans. R. Soc. London, 133, 145–231, 1843. Method (…) of Keeping a Reckoning at Sea (…), London, Sabine, E., Contributions to terrestrial magnetism, VIII, Phi- 1659. los. Trans. R. Soc. London, 136, 337–432, 1846. Parry, W. A., Journal of a Voyage for the Discovery of a North- Sabine, E., Contributions to terrestrial magnetism, IX, Philos. West Passage From the Atlantic to the Pacific; Performed in Trans. R. Soc. London, 139, 173–234, 1849. 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA ● 2-35

Sabine, E., Contributions to terrestrial magnetism, XI, Philos. Earthquake Investigation Committee, J. Coll. Sci., Imp. Trans. R. Soc. London, 158, 371–416, 1868. Univ. Tokyo, 14, 1904. Sabine, E., Contributions to terrestrial magnetism, XIII, Phi- The´venot, J., Relation d
un voyage fait au Levant, dans laquelle los. Trans. R. Soc. London, 162, 353–433, 1872. il est (…) traite´ des estats sujets au Grand Seigneur, etc. (Suite Sabine, E., Contributions to terrestrial magnetism, XIV, Phi- du Voyage (…) dans laquelle (…) il est traite´ de la Persie, etc.) los. Trans. R. Soc. London, 165, 161–203, 1875. Voyages de M. de The´venot contenant la relation de Sabine, E., Contributions to terrestrial magnetism, XV, Philos. l
Indostan, des nouveaux Mongols, etc., L. Billaine, Paris, Trans. R. Soc. London, 167, 461–508, 1877. 1664–1684. Sanderson, W., Observations on the variation of the needle Truxtun, T., Remarks, Instructions, and Examples Relating to the made in the Baltic, Anno 1720, Philos. Trans. R. Soc. Lon- Latitude and Longitude; Also, the Variation of the Compass don, 31, 120, 1720. (…) To Which Is Annexed, a General Chart of the Globe (…) Saumarez, P., Log of the Centurion: Based on the Original With a Short (…) Account of Variable Winds, etc., T. Dob- Papers of Captain Philip Saumarez on Board HMS Centu- son, Philadelphia, 1794. rion, Lord Anson
s Flagship During his Circumnavigation U.S. Hydrographic Office, Contributions to terrestrial magne- 1740–44, edited L. Heaps, Hart-Davis, MacGibbon, Lon- tism, the variation of the compass, from observations made don, 1973. in the United States Naval Service during the period from Schomburgk, R. H. (Ed.), The Discovery of the Large, Rich, and 1891 to 1894, Publ. 109, Govt. Print. Off., Washington, D. Beautiful Empire of Guiana, With a Relation of the Great and C., 1894. Golden City of Manoa (…). Performed in the Year 1595, by Vaillant, A. N., Voyage autour du monde exe´cute´ pendant les Sir W. Ralegh, Reprinted From the Edition of 1956: With anne´es 1836 et 1837 sur la corvette La Bonite (commande´ e Some Unpublished Documents Relative to That Country. Ed- par M. Vaillant), 14 vols., A. Bertrand, Paris, 1840–1852. ited With Copious Notes and a Biographical Memoir, HS Ser. van Bemmelen, W., De isogonen in de XVIde en XVIIde Eeuw, I, vol. 3, Hakluyt Soc., London, 1850. J. van Druten, Utrecht, Netherlands, 1893. Schott, C. A., The magnetic observations made on Bering
s van Bemmelen, W., The Observations Made With the Compass first voyage to the coasts of Kamchatka and eastern Asia in on Tasman
s Voyage and the Isogonic Lines Across the Indian the years 1725 to 1730, Bull. 20, U.S. Coast Geod. Surv., and Pacific Oceans in 1640, F. Muller, Amsterdam, 1896. Washington, D. C., 1890. van Bemmelen, W., Die Abweichung der Magnetnadel: Schott, C. A., Secular variation of the Earth
s magnetic force in Beobachtungen, Sa¨cular-Variation, Wert- und Isogonen- the United States and in some adjacent foreign countries, systeme bis zur Mitte des XVIIIten Jahrhunderts, Obs. R. U.S. Coast Geod. Surv., part II, app. 1, 167–320, Washing- Magn. Meteorol. Obs., Batavia, 21, 1899. ton, 1896. van den Broecke, P., Korte Historiael ende Journaelsche Aan- Schwerer, A., E´tude sur le magne´tisme terrestre a Terre- teykeninghe van al
t gheen merckwaerdich voorgevallen is in Neuve, par M. A. Schwerer, Lieutenant de Vaisseau, Ann. de langdurighe Reysen, soo nae Cabo Verdi, Angola, etc., als Hydrogr., 14,88–111, 1892. inzonderheydt naar Oost-Indie¨n, H. Passchiers van Wes- Schwerer, A., Observations` a la mer de M. Schwerer, Lieuten- busch, Haarlem, 1634. ant de Vaisseau` a bord du Dubordieu, Ann. Hydrogr., 18, van Musschenbroeck, P., Physicae experimentales et geometricae 177–187, 1896. de magnete, tuborum capillarium vitreorumque speculorum Shadwell, C., A contribution to terrestrial magnetism; being attractione, magnitudine terrae, cohaerenta corporum firmo- the record of observations of the magnetic inclination or rum: Dissertationes ut et ephemerides meteorologie ultrajecti- dip, made during the voyage of HMS Iron Duke to China nae, Samuel Luchtmans, Leyden, Netherlands, 1729. and Japan etc., 1871–1875, Philos. Trans. R. Soc. London, van Rijckevorsel, E., Verslag aan Zijne Excellentie den Minister 167, 138–143, 1877. van Kolonie¨n over eene Magnetische Opneming van den In- Shure, L., R. L. Parker, and R. A. Langel, A preliminary dischen Archipel, in de jaren 1874–1877 gedaan, Johannes harmonic spline model from Magsat data, J. Geophys. Res., Muller, Amsterdam, 1880. 90, 11,505–11,512, 1985. van Rijckevorsel, E., and C. E. Engelenburg, Magnetic Survey Stanley, H. E. J. (Ed.), The Philippine Islands, Moluccas, Siam, of the Eastern Part of Brazil, Johannes Muller, Amsterdam, Cambodia, Japan, and China, at the Close of the Sixteenth 1890. Century. by Antonio De Morga. Translated From the Spanish, Veinberg, B. P., Catalogue of Magnetic Determinations in USSR With Notes and a Preface, and a Letter From Luis Vaez de and in Adjacent Countries From 1556 to 1926, 3 vols., Cent. Torres Describing his Voyage through the Torres Straits. HS Geophys. Obs., St. Petersburg, Russia, 1929–1933. Ser. I, vol. 39, Hakluyt Soc., London, 1868. Vidal Gormaz, F. (Annot.) Documentos para la Historia de la Steen, A. S., Terrestrial Magnetism, The Norwegian North Polar Na´utica en Chile, vol. 2, Del Anuario-Hidrogra´fico, Viaje de Expedition 1893–1896: Scientific Results (1900–06), edited Enrique Brouwer, Viaje de Domingo de Boenechea. Viaje de by F. Nansen, vol. 2, no. 7, Fridtjof Nansen Fund for the Jose´ And´ıa i Varela, Santiago de Chile, 1892. Adv. of Sci., New York, 1901. von Humboldt, F. H. A., Personal Narrative of Travels to the Stelling, E., Magnetische Beobachtungen auf einer Reise nach Equinoctial Regions of the New Continent During the Years Urga im Sommer 1893, Bemerkungen u¨ber die A¨ nderungen 1799–1804, by A. de Humboldt and A. Bonpland, translated der erdmagnetischen Elemente in Ost-Sibirien, Me´m. Acad. from French by H. M. Williams, 7 vols., Addison-Wesley- St. Petersbourg, VIIIe Se´r. Classe physico-mathematique. vol. Longman, Reading, Mass., 1814–1829. 2, St. Petersburg, Russia, 1895. von Kotzebue, O., A Voyage of Discovery, into the South Sea and Stephenson, J. R., Historical Eclipses and Earth
s Rotation, Beering
s Straits, for the Purpose of Exploring a North-East Cambridge Univ. Press, New York, 1997. Passage (…)in(…) 1815–1818, 3 vols., Longman, Hurst, Stevin, S., De Havenvinding, Plantijn, Leyden, Netherlands, Rees, Orme, and Brown, London, 1821. 1599. von Tillo, A., U¨ ber die geographische Verteilung und sa¨cula¨re Taillemite, E. (Ed.), Bougainville et ses compagnons autour du A¨ nderung der Deklination und Inklination im Europa¨is- monde, 1766–1769, journaux de navigation, Imprimerie chen Russland, Rep. Meteorol., 8(2), 1–82, 1881. Natl., Paris, 1977. von Tillo, A., U¨ ber die geographische Verteilung und sa¨cula¨re Tanakadate, A., A magnetic survey of Japan reduced to the A¨ nderung der erdmagnetischen Kraft im Europa¨ischen epoch 1895 and the sea level carried out by order of the Russland, Rep. Meteorol., 9(5), 1–78, 1885. 2-36 ● Jonkers et al.: GEOMAGNETIC DATA 41, 2 / REVIEWS OF GEOPHYSICS

Wales, W., Astronomical Observations Made in the Voyages Wijkander, A., Observations magne´tiques faites pendant Which Were Undertaken by Order of His Present Majesty for l
expe´dition Arctique Sue´doise en 1872–1873, Kongl. Svens. Making Discoveries in the Southern Hemisphere and Succes- Vetenskaps Akad. Handl., 13(15), 1876. sively Performed by Commodore Byron, Captain Wallis, Cap- Wijkander, A., Observations magne´tiques, faites pendant tain Carteret and Captain Cook in the Dolphin, Tamer, Swal- l
expe´dition de la Vega 1878–1880. Vega Expeditions Veten- low and Endeavour, Comm. of the Longitude, London, skapliga Iakttagelser, vol. 2, Stockholm, 1883. 1788. Wille, C., and H. Mohn, Den Norske Nordhavs-Expedition Wales, W., and W. Bayly, The Original Observations Made in the 1876–1878, vol. 5, Magnetiske Observationer, Christiania, Course of a Voyage Towards the South Pole and Around the Norway, 1882. World, Board of Longitude, London, 1777. Wright, E., Certaine Errors of Navigation Detected and Cor- Walker, E., Terrestrial and Cosmical Magnetism: The Adams rected, With Many Additions That Were Not in the Former Prize Essay for 1865, Deighton and Bell, London, 1866. Editions, 3rd ed., Joseph Moxton, London, 1657. Wessel, P., and W. H. F. Smith, Free software helps map and display data, Eos Trans. AGU, 72(41), 441, 445–446, 1991. Weyprecht, C., Die Magnetische Beobachtungen der O¨ ste- reichisch-Ungarischen Arctischen Expedition 1872–1874, A. Jackson, A. R. T. Jonkers, A. Murray, School of Earth Denkschr. Kaiserlichen Akad. Wiss. Math.-Naturwiss. Classe. Sciences, University of Leeds, Leeds LS2 3GP, UK. Wien, 35,69–292, 1878. ([email protected]; [email protected]) 41, 2 / REVIEWS OF GEOPHYSICS Jonkers et al.: GEOMAGNETIC DATA

Figure 5. An eighteenth-century azimuth compass by J. Fowler. Courtesy of the National Maritime Museum, London.

● 2-11 ●