Magnetic Fields in Stars Magnetic Fields in Stars Magnetism—the force that deflects the needle of a compass—and magnetic fields have been found in some hundreds of stars during the past 50 yr. Magnetic fields have been detected in T Tauri stars and other pre-mainsequence stars, several types of main sequence stars, white dwarfs and neutron stars. We now know a number of methods by which such magnetic fields may be detected, we are in the process of learning more about how they are distributed over stellar surfaces, and we understand some of the ways in which these fields reflect—and sometimes influence—the evolution of the stars which possess them. The first stellar magnetic fields were detected in sunspots (see SUNSPOT MAGNETIC FIELDS) in the Sun by George Ellery Hale in 1908. Almost forty years later, in 1947, the first magnetic field in a star other than the Sun was found by Horace W Babcock, who discovered a magnetic field in the star 78 Virginis, a ‘chemically peculiar’ main sequence star (see STELLAR EVOLUTION) about twice as massive as the Sun. Magnetic fields are now known in perhaps 200 other A and B stars of the middle main sequence, all of which are, like 78 Vir, chemically peculiar (which means that they have very unusual surface chemical compositions). In such stars the fields are generally found to be roughly dipolar in form; in other words, they have an overall structure reminiscent of that of a simple bar magnet or of the Earth’s magnetic field, with a north and a south pole, between which the the magnetic force (as represented by magnetic lines of force) points along loops connecting one pole to the other. The typical field strength of such stars— over the whole stellar surface—is of the order of 1000 G (0.1 T), some 3000 times greater than the strength of the Earth’s surface field, and about as strong as the magnetic field of a good permanent horseshoe magnet. The discovery of PULSARS in 1967 by Jocelyn Bell Burnell and Anthony Hewish was soon recognized to be both the discovery of NEUTRON STARS and of magnetic fields in such stars. The pulsed radio radiation emitted by these spinning, magnetized neutron stars is still almost the only means for detecting single neutron stars. The roughly dipolar fields of neutron stars are initially of the order of 1012 –1013 G (108 –109 T) and then seem to decay in strength by about a factor of 100. Most or all neutron stars thus appear to be formed initially with fields about 1010 times stronger than are found in the magnetic middle main sequence stars. Three years later the first magnetic field was detected in a WHITE DWARF by James Kemp, John Swedlund, John Landstreet and Roger Angel. Fields are now known in about 50 other white dwarfs. These fields range from about 105 to 109 G (10–105 T) in strength, roughly a factor of 104 stronger than those of middle main sequence stars. The white dwarf fields also appear to be approximately dipolar in structure. Unlike neutron stars, only a few per cent of all white dwarfs have detectable magnetic fields. The first magnetic fields in stars of the lower main sequence were detected in 1980 by Richard Robinson, Pete E N C Y C LO P E D IA O F A S T R O N O M Y AN D A S T R O P H Y S I C S Worden and Jack Harvey. Fields are now known in more than 50 cool stars, mostly rather young, active stars, or in stars in which a companion enforces rapid rotation (see the SOLAR–STELLAR CONNECTION). Recently, fields have been found in a few PRE-MAIN-SEQUENCE STARS and T TAURI STARS. In cool stars, the fields detected have very different distributions over the stellar surface from those of the stellar types already described. Instead of simple, roughly dipolar structure, these fields seem to occur in forms more like giant sunspots, or in patches on the stellar surface that may resemble solar active regions. The field strengths are typically of the order of 103 G (0.1 T), while the fraction of the stellar surface covered is typically only of order 20–50%. Unlike the fields of middle main sequence stars, white dwarfs and neutron stars, all of which are observed to change in structure only very slowly with time, the distribution of magnetic flux on the surface of a lower main sequence star usually changes substantially in a period of weeks or months. It is generally supposed that, in the cool stars, the fields observed are generated by a dynamo process operating in the convective outer envelope of the star, while the more stable fields of middle main sequence stars, white dwarfs and neutron stars are ‘fossil fields’—large-scale fields produced during an earlier stage of evolution, and subsequently frozen into the highly electrically conductive matter of the star. In this article we will focus on the fields of middle main sequence stars and of white dwarfs, leaving those of solar-type stars and of neutron stars to other articles. Methods of detecting stellar magnetic fields Basic physics Most of the magnetic fields detected in main sequence and white dwarf stars are found by detecting the ZEEMAN EFFECT in the stellar spectrum. This effect splits each energy level of an atom in a magnetic field into several magnetic substates, leading to a number of effects that can in favorable cases be detected. When placed in an external magnetic field, a state i of an atom with energy Ei and total angular momentum quantum number J splits into 2J + 1 magnetic substates equally spaced in energy, with a spacing which varies from one atomic level to another. Transitions between level i and another level f of energy Ef are characterized by the frequency νif = (Ef − Ei )/ h when no magnetic field is present. When a field B is applied, the splitting of the lower and upper energy levels by the magnetic field leads to the splitting of the spectral line associated with this transition into three closely spaced group of lines. These groups arise because most atomic transitions allow the magnetic quantum number M to change by −1, 0, or 1. The M = Mf − Mi = 0 group, called π components, are distributed symmetrically about νif . The two groups of lines with M = ±1, called σ components, are shifted systematically to frequencies above and below νif , with the M = +1 group on one side and the M = −1 group Copyright © Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK 1 Magnetic Fields in Stars on the other. The typical separation between the π and one of the σ groups is λB = ḡeBλ2 /4πmc2 where ḡ is a number of order 1 which varies from one transition to another. In familiar units, the splitting is λB (nm) = 4.67 × 10−3 ḡB(kG)λ(µm)2 . Thus a field of 1 kG (0.1 T) leads to a π –σ separation of the order of 0.001 nm at 500 nm = 0.5 µm. The π and σ groups of lines are polarized. If the magnetic field is transverse to the line of sight, the π components (in emission) are linearly polarized parallel to the applied field and the σ components are linearly polarized in the orthogonal direction. If the field is parallel to the line of sight, the π components are suppressed and the two groups of σ components have opposite circular polarizations. When the magnetic field strength is strong enough (of order 105 G) that the perturbation of the atom by the field is larger than the spin–orbit perturbation, the relatively complex Zeeman effect is superseded by the Paschen–Back effect, which leads essentially to splitting of all lines into simple triplets. A field of order 106 G results in a significant quadratic Zeeman effect, which systematically shifts lines of large upper principal quantum number nf relative to lines of smaller nf . As a field of order 107 G is reached and the magnetic interaction energy becomes comparable with the Coulomb energy of the atomic electrical field, the atomic spectrum of any atom, even H, becomes extremely complicated. At fields above about 106 G, another magnetic effect occurs that is very useful for detection of fields in white dwarfs: polarization of continuum radiation. The broadband light from a star with a field strength of this order is circularly polarized, essentially because the field forces electrons to spiral about field lines in a preferred direction. For still larger fields, broad-band linear polarization can also occur. Field measurement methods The splitting, shifting and polarization of spectral lines by the Zeeman, Paschen–Back and quadratic Zeeman effects, and the occurrence of continuum polarization for sufficiently large fields, have provided astrophysicists with a number of methods of detecting and measuring magnetic fields in stars. The most straightforward of these methods is useful for stars that have a very small projected equatorial rotational velocity veq sin i and hence sharp spectral lines. In this case, one can directly observe the splitting of spectral lines into components if the field is of the order of a few kilogauss or more in main sequence stars, or about 106 G in white dwarfs. The separation of the observed line components provides a direct measurement of the modulus of the magnetic field averaged over the surface E N C Y C LO P E D IA O F A S T R O N O M Y AN D A S T R O P H Y S I C S of the star, a quantity called the mean field modulus, Bs , as shown in figure 1. The polarization introduced into spectral lines by magnetic splitting provides a second powerful method of field detection. This method depends on the fact that, in a magnetic field with a significant component along the line of sight, the σ components on one side of the line center absorb circularly polarized light of one sense of polarization, while the σ components on the other side of line center absorb the opposite circular polarization. If we observe the stellar spectrum through polarizers that pass each of the two senses of circular polarization, the absorption lines in one of the two circularly polarized spectra are not at precisely the same wavelengths as the same lines in the other polarized spectrum, because of the small wavelength difference between the two sets of σ components. The presence of a field can be detected by this shift in the position of spectral lines between spectra observed in right- and left-circularly polarized light, or equivalently in the spectrum formed by the difference of the two spectra divided by their sum, as shown in the lefthand panel in figure 2. The fact that, in the presence of a magnetic field transverse to the line of sight, the absorption by the π components is orthogonally polarized with respect to the polarization of the σ components leads to a similar effect when a spectrum is observed through linear polarizers oriented parallel to and perpendicular to the field direction, as seen in the right-hand panel of figure 2. In general, the observable effect in linear polarization is considerably smaller than in circular polarization. Detection of the polarization effects from a stellar magnetic field is possible only if there are substantial regions on the stellar surface over which the field does not change direction too much. Clearly, a field structure in which many small tubes of magnetic flux directed out of the star are closely mixed in with small tubes of inwarddirected flux will lead to no net polarization, as the effects of adjacent oppositely directed flux tubes will cancel. On the other hand, the detection of circular polarization, which is not readily produced in line profiles by other mechanisms, is a very robust indicator of the presence of a field. Furthermore, very small levels of polarization (0.1% or even 0.01%) can be measured reliably. The result is that circular polarization methods, which measure what is called the ‘mean longitudinal field’ B , provide much more sensitivity to weak but geometrically simple magnetic fields than methods that depend on studying line profiles. In the best cases, B values as small as tens of gauss can presently be detected. Magnetic fields in middle main sequence (‘peculiar A’) stars The oblique rotator Magnetic fields are detected in middle main sequence stars (stars of between about 2 and 10 times the mass of the Sun, which are burning hydrogen in their cores) both by the detection of circular polarization in spectral Copyright © Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK 2 Magnetic Fields in Stars E N C Y C LO P E D IA O F A S T R O N O M Y AN D A S T R O P H Y S I C S Figure 1. A portion of the spectrum of the very slowly rotating magnetic Ap star HD 94660 (the spectrum at the top of the figure), showing splitting of spectral lines by a magnetic field of Bs ≈ 6400 G. Below the stellar spectrum is a schematic diagram of the Zeeman splitting pattern of several strong stellar spectral lines. Each of the vertical lines above and below the line at intensity 0.4 (respectively π and σ components) represents one Zeeman component; the position of each component is calculated for a field of 6400 G, and the height of each line is proportional to the strength of that component. The ion responsible for each strong line is identified at the bottom of the figure. (From G Mathys 1990 Astron. Astrophys. 232 151, reproduced by permission of Astronomy and Astrophysics.) lines and by observing Zeeman splitting of spectral lines in stars with small veq sin i values. All well-confirmed detections of fields in middle main sequence stars are in members of a class of stars known as ‘chemically peculiar A stars’, often called Ap (or Bp) stars (see also PULSATING AND CHEMICALLY PECULIAR UPPER MAIN SEQUENCE STARS). These stars have long been known to have (sometimes very) anomalous atmospheric chemical composition compared with the Sun. Their chemistry is typically characterized by unusually large amounts of chemical elements such as Sr, Cr, Eu and other rare earths, Si and (only in the most massive Bp stars) He. Most of these stars also have unusually small amounts of a few elements as well, often He and O. The elements which are anomalous depend systematically on the mass of the star. It appears that all chemically peculiar stars having the same general chemical anomalies as the known magnetic stars probably have fields. A first question about the magnetic middle main sequence stars (I will call them ‘magnetic Ap stars’) is that of determining the geometry of the magnetic field over the stellar surface. This cannot be determined by directly observing the disks of such stars, of course; they are much too small in angular size for direct imaging. We must use other kinds of information to deduce the magnetic geometry. The magnetic Ap stars are in most cases observed to vary periodically in apparent brightness, in the strengths and profiles of some or most spectral lines, and in the measured components of the magnetic field. All these quantities are generally observed to vary with the same period. The period of variations is typically in the range 1–10 days, although periods as short as 0.5 days and as long as some decades or known. An example is shown in figure 3. An important clue to the origin of the variations is furnished by the observed fact that the value of the projected rotational velocity veq sin i is closely correlated with the period of variation. Large values of veq sin i are only found for stars with short periods and, the longer the period, the smaller the values of veq sin i are. The facts that the observed periods are found with an enormous range of values and that the periods are closely related to the projected rotational velocities clearly indicate that the observed variations are due to the rotation of the magnetic Ap star. Variation in the average magnetic field indicates that the star must have a magnetic field that varies from one place to another at the surface, either in strength or in inclination or both. Thus, when we see mainly magnetic field lines directed towards the observer, we measure a large value of B , but, when the field lines are mainly perpendicular to the line of sight, B is small. The variations in spectral line intensity and shape indicate that the relative abundances of various chemical elements vary from one place on the star to another. When we are looking at a part of the star in which some element (such as He) is relatively abundant, the spectral lines are strong and deep; when we look at a different part of the star which has relatively less He, the spectral lines are weaker. The variations in surface chemistry in turn influence the amount of light emitted at various wavelengths and lead to the variations observed in brightness as the star rotates. We observe that the value of B of a magnetic Ap star generally varies in a fairly sinusoidal fashion. When we Copyright © Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK 3 Magnetic Fields in Stars Figure 2. Effects of a magnetic field on the line profile and polarization of stellar spectral lines. The panels show schematically the effect of a magnetic field with the Zeeman pattern displayed in (a) for a field parallel to the line of sight (left) and perpendicular to it (right). (b) The change in the line from the non-magnetic case (dotted) to the magnetic case (solid) for Zeeman splitting comparable with the Doppler width of the spectral line. (c) The absorption line in left- and right-circular polarization (left) and in linear polarization parallel to and perpendicular to the stellar field (right). (d) The net circular (left) and linear (right) polarization in the line. (From D J Landstreet 1980 Astron. J. 85 611, reproduced by permission of the American Astronomical Society.) try to reproduce this behavior by calculating the variations expected from various simple models of the field geometry, we find that this observation is consistent with the idea that the structure of the magnetic field over the surface of the star is in the general form of a dipole, typically inclined (oblique) to the rotation axis of the star by some fairly large angle. As the star rotates, we thus usually see one pole of the dipolar distribution, and then the other. This model is known as the ‘oblique rotator model’. Astronomers are also actively working to use observed spectrum variations to deduce the distributions of different chemical elements over the surface of the star. Often the models that fit the observations have a distribution of the elements that is roughly axisymmetric around the axis of the magnetic dipole. It is not uncommon to find very large differences in the fractional abundances of some elements over the stellar surface; in some cases E N C Y C LO P E D IA O F A S T R O N O M Y AN D A S T R O P H Y S I C S Figure 3. Variations of the massive magnetic Bp star HD 184927 (whose most striking abundance peculiarity is a considerable excess of atmospheric He) with a period of 9.53 days. The horizontal axis gives time in phase units (fractions of one cycle from the periodically repeating time when the helium spectral line intensities are strongest). The three panels show (from top to bottom) the mean longitudinal field in kilogauss, the strength of one helium line in arbitrary units and the brightness of the star (in magnitudes) seen through a narrow (Strömgren) ultraviolet filter. (From G A Wade et al 1997 Astron. Astrophys. 320 172, reproduced by permission of Astronomy and Astrophysics.) there may be more than 100 times more atoms of some elements per unit volume of gas in one part of the star’s atmosphere than in another part. Origin of the observed magnetic fields and of the chemical anomalies Astronomers generally accept two possible origins for observed stellar fields. One is that a field may be generated by the interplay between convection (boiling motions of the gas) in the outer layers of a star and the overall rotation of the star. These motions may act as a dynamo in the highly conducting outer layers of a star, producing a complex and time-varying field. This mechanism seems to be the cause of the magnetic field observed in the Sun. In spite of much theoretical effort, such dynamo fields are still not well understood. The second possible origin is that the field is the result of the collapse of a huge gas cloud to form a tiny star, trapping in the partly ionized, electrically conducting gas some small fraction of the weak galactic magnetic field. As the field lines of the galactic field are squeezed together, the strength of the entrained field is amplified by a very large factor. Thus, the observed fields of the magnetic Ap stars may be ‘fossil’ magnetic fields. This is not as unreasonable an idea as may at first appear. Owing to the very large bulk and high electrical conductivity of a star, a field formed in this way would take a very long time, of the order of 1010 yr, to decay. This is longer than the main sequence lifetime Copyright © Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK 4 Magnetic Fields in Stars of a middle main sequence star, and so a fossil field could easily persist throughout the life of such a star. The absence of any long-term changes in the observed magnetic field strengths of magnetic Ap stars (such as the solar activity cycle), other than the periodic variations caused by stellar rotation, and the simple overall geometry deduced for these fields, suggests that they are probably fossil fields rather than dynamo-generated ones. This view is supported by the lack of any obvious means of generating a large dynamo field in a star that may be rotating 100 times slower than the Sun and that has almost no convection in its outer layers. The presence of remarkable chemical peculiarities in these magnetic stars also requires an explanation. In fact, this is not quite as anomalous as it seems. The lowermass stars of the main sequence (M ≤ 2M ) show quite homogeneous abundance patterns, with overall content of ‘heavy elements’ (everything heavier than He) that depends only on the age of the star. (Older low-mass stars, formed when the universe was less rich in the elements synthesized in many generations of supernova explosions, have less of all heavy elements than younger stars.) Very massive main sequence stars (M ≥ 10M ) lose mass rapidly, and this ensures that their surfaces reveal their bulk composition which, because of their very short main sequence lifetimes, is essentially that of the contemporary interstellar material from which they form. The intermediate-mass stars, among which the magnetic Ap stars are found, almost all exhibit some degree of chemical individuality. In most A and B stars the variations are no more than some tens of per cent of excess or deficiency compared with other similar stars, but several other families of stars (mostly slowly rotating and apparently non-magnetic stars) are known in which certain deficiencies or enhancements can reach much larger values. ‘Metallic-line A’ (Am) stars often have 10 times less Ca and Sc than other main sequence stars of similar mass, and 10 times more of some rare earths. ‘Mercury–manganese’ (HgMn) stars have enhancements of some elements by factors of from order 102 (V, Mn, Ga) up to 105 or more (Eu, Pt, Hg). The abundance anomalies of the magnetic Ap stars are simply some of the most spectacular types of anomaly in a mass range filled with variety. These chemical abundance anomalies are generally believed to be confined to the atmospheres and outer envelopes of intermediate-mass stars, rather than being representative of the bulk chemical composition of these stars, for several reasons. First, the wide variety of observed compositions, in stars all of which formed relatively recently in galactic history, does not correspond to any similar variety of compositions in the interstellar clouds which form stars, or in other young stars of low or high mass. Furthermore, the extremes of anomaly are so great (factors of 105 or more) that it is not possible to imagine any way in which star formation could have led to gas clouds with such peculiar composition. Instead, we believe that the observed chemical anomalies are E N C Y C LO P E D IA O F A S T R O N O M Y AN D A S T R O P H Y S I C S essentially surface phenomena, due to powerful processes that separate elements, raising some into the atmosphere while others sink out of sight. The main sorting process leading to chemical anomalies is microscopic diffusion of atoms of lowabundance elements, relative to the dominant hydrogen of the stellar gas. Under the influence of gravity, elements with higher atomic mass than hydrogen tend to sink into the interior of the star. In a sufficiently stable atmosphere, this process would eventually lead to an exterior layer made up only of hydrogen, as is actually observed in many white dwarfs. However, there are competing processes. One of the most important is the outward force felt by atoms and ions which can absorb photons of many wavelengths from the outward flow of radiation through the star. This absorption imparts an outward acceleration to such ions and lifts them up to higher levels in the stellar envelope. Thus, the overall effect of diffusion is to allow some elements to sink in the atmosphere under the dominant influence of gravity, while others are lifted towards the surface by radiation. These sorting processes compete with various mixing processes such as convection. Thus, because the outer layers of low-mass main sequence stars are strongly convective, all sorting processes are strongly inhibited, and these Sun-like stars exhibit very similar compositions. In contrast, the main sequence stars of intermediate mass are precisely the stars with sufficiently stable atmospheres to allow diffusion to sort the chemical elements, at least to some extent. Rapid rotation is capable of generating slow mixing currents, and so the more rapidly rotating A and B stars have only modestly sorted surface chemistries. Most of the more peculiar middle main sequence stars are slowly rotating. The magnetic Ap stars have the additional feature that the presence of the magnetic field rather strongly inhibits mixing motions in the outer layers. The chemical peculiarities of the magnetic Ap stars are simply a particularly strongly developed aspect of a characteristic found in all stars in this mass range. Magnetic fields in white dwarfs Observations and modelling Magnetic fields are detected in white dwarfs by the same methods used for magnetic Ap stars, namely by direct observation of magnetic splitting of spectral lines and by observation of circular polarization in line wings. Fields are also detected by means of the continuum polarization produced by fields of more than about 106 G. The deduced fields range in strength from about 105 up to 109 G. At the low end of this range, the spectrum of a white dwarf is hardly perturbed at all by the field. For fields in the range from about 106 to 3 × 107 G, splitting of familiar spectral lines is easily seen. For still larger fields, the wavelengths and shapes of spectral lines are so strongly altered by the field that the spectrum is not recognizably related to that of any non-magnetic white dwarf. The spectrum of one magnetic white dwarf is shown in figure 4. Copyright © Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK 5 Magnetic Fields in Stars E N C Y C LO P E D IA O F A S T R O N O M Y AN D A S T R O P H Y S I C S more than about 3 × 107 G. Modelling of such fields leads to deduced magnetic field geometries that are roughly dipolar, like the fields of magnetic Ap stars. For stronger fields, the main difficulties come from the complex and uncertain behavior of even simple atoms such as H in the presence of fields so strong that they influence the motion of the electron(s) as strongly as the central Coulomb attraction of the atomic nucleus. The observations of spectra of white dwarfs with fields of 108 G or more are in reasonable accord with the results of atomic physics calculations, but no detailed modelling has yet been possible. Fields as small as about 10 kG could be detected in most white dwarfs. Surveys of many white dwarfs to about this level of precision have shown that the great majority of white dwarfs do not possess detectable magnetic fields. About 4% of the total population of white dwarfs have fields, with about equal probabilities per decade of field strength over the range 105 –109 G. Figure 4. The flux and polarization spectrum of the magnetic white dwarf GD 229, which has a field of order 109 G. The lowest curve shows the wavelength variation of the flux, the second lowest of circular polarization, the third lowest the percentage linear polarization and the top the position angle of linear polarization. The polarization and/or line splitting is observed to be variable in about one-quarter of the known magnetic white dwarfs. Observed variations are periodic, with periods in the range from about 1 h to 20 days. These periods are so long compared with any reasonable oscillation period of a white dwarf that they must be rotation periods, and so we are again quickly led to the oblique rotator model for the variations. The observed variations in the magnetic field strength and in spectral line shapes are again interpreted as simply being due to the fact that we see a magnetic field that is inclined to the stellar rotation axis from different directions as the star rotates. The fact that most magnetic white dwarfs do not vary may imply that, in most magnetic white dwarfs, the magnetic field is axisymmetric about the rotation axis or possibly that most magnetic white dwarfs rotate with periods of decades or more. Modelling of observed spectra and their variations is possible if the fields are not too large, say not much Origin of white dwarf fields Trying to understand the origin of the fields observed in white dwarfs presents us with substantial challenges. There are no obvious mechanisms for producing largescale, ordered, static fields in either magnetic Ap stars or white dwarfs after they are formed. We observe that a small fraction of middle main sequence stars, and of white dwarfs, have magnetic fields large enough to detect, in the range 102 –105 G on the main sequence and 105 –109 G in white dwarfs. The observed magnetic Ap fields may be due to magnetic flux retention during star formation, and the fields of white dwarfs could be due to the further retention of that same flux as magnetic Ap stars collapse to become white dwarfs. This hypothesis is consistent to some extent with the relative values of observed field strength, since, if the magnetic flux threading a star’s equator is retained during a collapse, the magnetic field strength will increase as B ∝ /R 2 where R is the stellar radius. Thus the decrease in radius by a factor of 102 as a star becomes a white dwarf could lead to a field strength increase by a factor of 104 , about the difference observed between the ranges of field strength on the main sequence and among white dwarfs. However, this does not explain how that magnetic flux is retained in the evolution stages between the main sequence and white dwarf stages; the intervening giant state is expected to be largely convective, which might be expected to expel much of the magnetic flux in a star. Furthermore, this idea does not explain why the largest (108 –109 G) fields are as common as fields 103 times smaller; on the main sequence the largest fields are a modest tail on a distribution that is very strongly peaked around fields of less than 103 G. Magnetic fields in neutron stars present us with further challenges. It appears that almost all neutron stars have fields of the order of 1010 –1013 G. Again, these are Copyright © Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK 6 Magnetic Fields in Stars E N C Y C LO P E D IA O F A S T R O N O M Y AN D A S T R O P H Y S I C S about the fields that would be expected by magnetic flux retention from the main sequence. However, if this is the origin of magnetic fields in neutron stars, why do almost all neutron stars have large fields, while only a small fraction of white dwarfs have large fields? This question is made still more puzzling by the fact that virtually no magnetic main sequence stars are known in the mass range that is expected to evolve eventually to neutron stars. Bibliography Chanmugam G 1992 Magnetic fields of degenerate stars Ann. Rev. Astron. Astrophys. 30 143–84 Dworetsky M M, Castelli F and Faraggiana R 1993 Peculiar versus Normal Phenomena in A-type and Related Stars (San Francisco, CA: Astronomical Society of the Pacific) Landstreet J D 1992 Magnetic fields at the surfaces of stars Astron. Astrophys. Rev. 4 35–77 North P, Schnell J and Žižňovský J 1998 Proc. 26th Meet. and Workshop of the European Working Group on CP Stars, Contrib. Astron. Obs. Skalnaté Pleso 27 (3) Schmidt G D 1995 White dwarfs as magnetic stars Rev. Mod. Astron. 8 147–62 John D Landstreet Copyright © Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK 7