Radio, X-Ray, and Ultraviolet Observations
Results published between 1972 and 1977

1972

Between October 23rd and November 10th, 1971, Wade and Hjellming observed the Algol region with the National Radio Astronomy Observatory, Green Bank, West Virginia, interferometer operating at 2,695 MHz (11.1 cm) and 8,085 MHz (3.7 cm), with element spacings of 900, 1,800 and 2,700 m. They reported (Wade and Hjellming, 1972) radio emission from a source whose position closely coincided with that of Algol. They hence concluded that the radio source was closely associated with the multiple star system. The observed radio emission was strongly variable at 2,695 MHz, less so at 8,085 MHz.

Wade and Hjellming noted that the deduced radio spectra were not inconsistent with thermal radiation from an optically thick source. If the emission did arise from an optically thick cloud of ionised gas, at a temperature of 10,000 K, then the flux density of 1.5 × 10 W m^-2 Hz^-1 observed at 8,085 MHz would correspond to an angular diameter of 0.2 arc seconds for the source, or a linear diameter of 5 AU at the distance of Algol (for which Wade and Hjellming quoted a value of 32 pc). Since this diameter is much greater than the separation of the eclipsing pair, Wade and Hjellming suggested that the optical spectra of Algol should then show strong nebular emission lines at all times, with little, if any, variation in the intensity of those lines related to the orbital motions of the embedded stars. However, as Wade and Hjellming noted, this is quite unlike what is actually observed.

Together with Webster, Wade and Hjellming made further observations of Algol with the NRAO interferometer between January 11th and February 16th, 1972 (Hjellming, Wade and Webster, 1972). During this period Algol behaved in a manner spectacularly different from that observed previously: the flux densi ty varied between 0.01 and 0.34 flux units (1 fu = 10 W m^-2 Hz^-1) with a typical time scale of a few hours. From data obtained at 8,085 MHz, Hjellming, Wade and Webster found that an upper limit of 0.5 arc seconds could be placed on the angular diameter of the radio source, and hence obtained a lower limit of about 4 × 10⁴ K for the brightness temperature of the source. The radio emission flared strongly on a number of nights, and Hjellming et al. thought it likely that considerable flaring had occurred during the unobserved times. They found no evidence of correlation between the observed radio emission and the phase of the eclipsing system. The simultaneous variation at 2,695 and 8,085 MHz during strong flares made it unlikely that the changes were caused by interstellar scintillation. Prompted by a private communication from Hjellming and Wade, Huges and Woodsworth made observations of Algol at 10,683 MHz (2.8 cm) with the 46m telescope at Algonquin Radio Observatory during the period of the flare of January 21st/22nd, 1972. They presented the results of their observations and those of Hjellming et al. (Huges and Woodsworth, 1972), and showed that the radio emission could be represented by a “quasi-steady” component and a flaring component which showed an exponential decay with the same time constant for all three frequencies. They suggested a non-thermal origin for the radiation produced by the flare. Hjellming et al. suggested that the radiation might be due to incoherent synchrotron emission from relativistic electrons (in which case it would be difficult to account for the observed behaviour of the radio flares) or to coherent radiation processes due to instabilities of interacting high-velocity plasmas. They also note the report (C. T. Bolton, TAU Circular No. 2388, February 25th, 1972) of changes in the Ca II K lines which may have been associated with the radio flaring observed by Wade and Hjellming.

Hjellming (1972) presented a further model for the radio source. He suggested that the simplest interpretation of virtually all the data was in terms of a variable thermal bremsstrahlung source, which is optically thin at the beginning of a flare, becomes partially optically thick at peak flaring and returns to the optically thin state as the flare ends. In a consideration of this model, Hjellming found that the product T_eD², where T_e is the electron temperature and D the diameter of the source, remained constant during a particular radio flare, but was significantly different for different flares – a striking feature of the Algol radio source. This indicated either that both T_e and D remained constant during a flare, or that T_e was inversely proportional to D² during the evolution of a flare, corresponding to a steady cooling (or heating) of the source as it expands (or contracts). From the observational data obtained at 2,695 and 8,085 MHz and theoretical considerations, Hjellming used a simple argument to give an order of magnitude estimate of the parameters of the thermal source, which suggested that the source was of the same size scale as the Roche lobes of the close pair. He also noted that if the thermal interpretation of the radio data was valid, then Algol should be a transient X-ray source (rivalling Scorpius X-I in strength) during the time when strong radio flaring is observed. However, as radio flaring was infrequent, Hjellming was not surprised that Algol had not been detected as an X-ray source. He further suggested that the then high levels of radio activity were related to mass transfer within the binary system, possibly associated with sudden changes in the eclipsing period (Frieboes-Conde et al., 1970).

When considering the various discussions of the Algol radio source, it should be born in mind that a serious error has crept into the literature. Frieboes-Conde et al. (1970) are frequently cited as giving a value of 0.2 AU for the semi-major axis of the eclipsing system, which, as Sahade and Wood (1970) note, is an obvious impossibility “if Kepler’s third law is to be trusted”. Hjellming was certainly guilty of this. What Frieboes-Conde et al. did was to cite Eggen and Pavel as giving this value for the distance from the centre of mass of Algol ABC to the centre of mass of the system containing the hypothetical fourth component, Algol D. What effect, if any, this has on the validity of the various models suggested for the radio source is uncertain.

Hjellming, Webster and Balick (1972) made observations of Algol with the NRAO interferometer on 28 days between April 21st and July 24th, 1972. The source exhibited sustained flaring activity at the beginning and end of the period, and was remarkably quiescent for at least two weeks in between. Three events which were of particular interest are shown in Fig. 5.

Figure 5a. The radio flux densities of Algol at 2,695 and 8,085 MHz plotted as a function of universal time for the event of April 27th, 1972. (After R. Hjellming et al. [1972].) An example of a flare decrease interrupted by a sudden flux increase (by 60% at 8,085 MHz and 25% at 2,695 MHz). For this event all the change occurred during a ten-minute interval. There was no doubt that the sudden flux change was real, for the jump occurred on all three baselines at both frequencies.

Figure 5b. The radio flux densities of Algol at 2,695 and 8,085 MHz plotted as a function of UT for the event of April 28th, 1972. (After R. Hjellming et al. [1972].) An nearly perfect example of the decay of a “typical” flare.

Figure 5c. The radio flux densities of Algol at 2,695 and 8,085 MHz plotted as a function of UT for the anamalous event of July 11th, 1972. (After R. Hjellming et al. [1972].) An extremely rapid postflare decrease. This was unique among all observations to that date. In th eforty-five minutes ending at about 9 UT, the radio flux at both frequencies dropped from 0.26 fu to less than 0.02 fu.

Apart from the anomalous event of July 11th, these observations were consistent with Hjellming’s thermal-source model. Certainly, the timescales for the cooling (which would be dominated by thermal bremsstrahlung losses) and expansion of the model agreed quite well with the observed decay times of “thermal-like” flares. Hjellming et al. suggested that a likely mechanism for producing the large amounts of energy and mass loss required by the thermal model would involve changes in the structure of one of the stars, proceeding in discontinuous steps and which resul ts inmass ejection. Such “starquakes” could easily drive shock waves, magnetoacoustic waves and suprathermal particles which would energize the plasma in the surrounding Roche lobe, so producing the “typical” radio flares. The observed sudden increases in flux would be explained naturally as energisation by a starquake and the corresponding aftershocks. The non-thermal nature of the event of July 11th was strongly suggested by its rapid decay. The event could be easily understood as synchronous decay of relativistic particles interacting with the stars’ (100 gauss) magnetic fields. Hjellming et al. suggested that such brief ejection events might result from uncommon by-products of starquakes in which there are unusually large fluxes of relativistic electrons.

1973

An alternative model was presented by Jones and Woolf (1973). They suggested that the radio emission was, in fact, entirely non-thermal and arose from the liberation of gravitational energy as mass ejected from Algol B fell onto the surface of Algol A, the radiation being produced by the interaction of plasma waves excited by the infalling matter. The variability of the radio emission was explained as a result of variability in the rate of mass flow. The authors determined a value of 3 × 10^-9 - 3 × 10^-8 m_☉ a^-1 for the rate of mass flow required to explain the observed radio luminosity of the source. Further, they suggested that the stopping of ions in the inflowing matter by the shock front in the atmosphere of Algol A would generate thermal X-rays, the upper limit to the X-ray emission being given by the inflowing kinetic power.

Canizares et al. (1973) presented the results of observations made with the MIT X-ray instrument aboard the OSO-7 (OSO = Orbiting Solar Observatory). The instrument scanned Algol continually between January 31st and February 17th, 1972 (a period in which Hjellming, Wade and Webster had observed radio emission), but found no indication of X-ray emission from the system. Hence they determined a value of 0.2 - 10 × 10^-10 erg cm^-2 s^-1 keV^-1 for the upper limit on X-ray flux. Hjellming’s “thermal” model predicted an X-ray flux of one to four orders of magnitude higher than this value. Canizares et al. thus concluded that such a model was untenable, and that the radio emission was probably due to some complex non-thermal process, such as that suggested by Jones and Woolf.

Clark et al. (1975) made high-resolution observations of Algol using the NRAO 43m antenna and the Haystack, Westford, Massachusetts, 37m antenna as elements of an interferometer with a baseline of twenty million wavelengths, operating at 7,850 MHz (3.8 cm), and the NRAO Mark II data recording and processing system, during a radio flare on May 4th/5th, 1974. The source was found to have an angular diameter of about 4 milli-arc seconds, corresponding to a linear diameter of 0.1 AU. Further, the observations suggested that during the flare the source was expanding, with a mean apparent expansion velocity of 500 - 1,000 km s^-1 (depending on whether a Gaussian or uniform disc brightness distribution was assumed). The comparitively large size of the source seemed to Clark et al. to rule out the model of Jones and Woolf.

1975

On January 15th, 1975, Gibson, Viner and Peterson were observing Algol as part of a routine program with the NRAO interferometer, when the behaviour of the source suggested that it was undergoing a large radio outburst. They alerted several radio and optical observatories. The NRAO VLBI (very long baseline interferometery) group enlarged its previously planned program of observations of the system. In all, nineteen individuals at eight radio and optical observatories contributed observations, which, at that time, comp rised the most substantial body of simultaneous data ever gathered for a non-X-ray radio binary. Gibson et al. (1975) presented the results of flux density, spectral and polarization observations. At 8,085 MHz, two maxima were observed, the first at 2320 UT (S = 0.98 Jy) and the second at 0510 UT (S = 1.05 Jy). (See Fig. 6.)

Figure 6. Simultaneous four-frequency observations of Algol one january 15th/16th, 1975: 6-minute averages. This event was the largest recorded radio outburst to that date for a non-X-ray [sic] radio binary. (After D.M. Gibson et al. [1975].)

Gibson et al. found that the observed behaviour could be explained as two unusually strong, but otherwise “typical”, Algol-type radio flares. The spectral index,

also showed two maxima, which coincided with those in the flux density, of 1.28 and 1.15 respectively. These values are greater than the typical values of a for Algol-type flares. No significant (> 2%) polarization was detected. The results of the VLBI observations were presented by Clark et al. (1975). The observations were made using the NRAO 43m antenna and the Haystack 37m antenna as elements of the twenty million-wavelength-baseline interferometer, operating at 7,850 MHz (3.8 em). During the last three hours of the observations, the Goldstone 64m antenna was also used, adding baselines of 85 and 100 million wavelengths. Clark et ale determined a value of 1.7 � 0.1 milli-arc seconds for the angular diameter of the radio source, corresponding to a linear diameter of 0.05 AU, comparable with the sizes of the individual stars of the close pair. They noted that the observational data were consistent with Algol A, rather than Algol B or the region containing L1 (see Appendix), being the origin of the radiation. They found no evidence to suggest that the source was expanding with a velocity greater than 100 km s^-1. The radio emission was considered to be non-thermal in origin.

Woodsworth and Hughes (1975) presented an extensive mathematical treatment of a new model which describes the radio emission in terms of two components: a low-level background component, and a flaring component. The low level component is associated with an optically thin H II region surrounding the system, which for an assumed electron temperature of 10⁴ K has a minimum radius of 5.8 AU and a maximum electron density of 6.4 × 10⁶ cm^-3. It may be that the H II is associated with the original star formation. Fluctuations in the background component are probably due to fluctuations in the amount of ionizing radiation from the central stellar object, associated with mass exchange or flaring. The flare component itself is produced by synchrotron emission, probably associated with the region between Algol A and Algol B where mass is being exchanged. The radiation is modified by the presence of thermal plasma which is partially optically thin, and, for an electron temperature of 10⁴ - 10⁵ K, has an electron density greater than 10⁸ cm^-3. The observed decays (which are not generally exponential, as had previously been supposed) are explained in terms of either synchrotron losses or adiabatic losses of an expanding source. The rather large (100 gauss) magnetic field required by the first mechanism suggests that the second may be the more likely.

1976 – X-ray obervations

During a five-day observation of the Perseus cluster of galaxies with instruments on board the SAS-3 X-ray observatory in October, 1975, Schnopper et al. (1976) discovered a new X-ray source. The position of the source contained Algol within its 90% error radius of 1.1', and Schnopper et al. hence proposed that Algol was itself the origin of the emission. The measured intensity of the source in the 2 - 6 keV energy range was 3.8 x 10^-11 erg cm^-2 s^-1 keV^-1, corresponding to an X-ray luminosity of 1.6 × 10³¹ erg s^-1for a source located at the distance of Algol. No evidence was found for X-ray emission in the 6 - 11 keV channel. Schnopper et al. noted that models for X-ray emission from binary systems are generally based on mass transfer from the central star to a compact secondary object. Algol has no such component (as far as is known) and so adds another type of stellar system to the class of X-ray binaries. They also commented on the various models for radio emission from Algol which had suggested that the star would be an X-ray source.

Further X-ray observations were made by Harnden et a1. (1977) using a soft X-ray telescope aboard Aerobee 350 rockets launched from White Sands Missile Range on March 15th and December 6th, 1975. (The soft X-ray image obtained from the December observations is shown in Fig. 7.)

Figure 7. A soft X-ray image of a 2° × 2° field containing Algol, obtained froma 25s exposure taken on December 6th, 1975, at a mean energy of 1 keV. Each element is a 4' × 4' square, and □ [open square] = 1-2 counts, ▤ [hatched square] = 3-5 counts, and ■ [solid square] = 9 counts. (After F.R. Harnden et al. [1977].)

These confirmed that Algol is, indeed, an X-ray source. A total of 44 counts was obtained from the source over the 0.15 - 2 keV band, indicating an X-ray luminosity of about 2 x 10³⁰ erg s^-1. This value is one order of magnitude lower than that measured by Schnopper et al. suggesting that the source may be variable. Harnden et a1. discussed models in which the X-ray flux is interpreted as thermal emission produced by the direct accretion of matter from Algol B to Algol A, the mass being transfered either by Roche lobe overflow (see Fig. 8) or in a stellar wind.

Figure 8. Roche lobe overflow model for X-ray emission from Algol. (After F.R. Harnden, Jr., et al. [1977].)

The observational data were insufficient to distinguish uniquely between these two mechanisms. Harnden et al. suggested that variations in the X-ray flux were caused by the interaction of the streaming matter with non-uniformities in the magnetic field of Algol A, and noted that such interactions could trigger the observed radio flares.

1970s – Ultraviolet observations

A few studies based on satellite observations of Algol in the ultraviolet have also been published. Chen and Wood (1975, 1976) obtained observations from the Princeton Experimental Package aboard the Copernicus satellite on September 7th, 8th and 9th, 1973, and January 4th, 1974. In the first paper they presented a study of the Lyman-α line inside and outside of primary eclipse. This showed no emission features and is consistent with what would be expected for a normal B8 star. The second paper discussed variation in the Mg II resonance lines near λ 2,800 Å. Near the centre of primary minimum these appeared to be doubled – a feature discovered much earlier in Mg II at λ 4,481 Å. A possible explanation is the effect of emission from Algol B which becomes detectable only when almost all of Algol A is in eclipse. However, Struve and Sahade had explained the doubling in Mg II at λ 4,481 Å by the blending of the lines of Algol A and Algol C. Hence, the peculiar behaviour of these lines near mid-eclipse warrants further study. The same Copernicus observations were used by Chen et al. (1977) to analyze the light curve of Algol at λ 3,428 Å. The results indicated that the surface temperature of Algol B is higher than 5,000 K. Later ultraviolet observations of Algol (Kondo et al., 1977) have provided evidence of mass flow, and of other peculiarities.

Envoi

In the introduction to a paper published in 1971, the authors (Hill et al., 1971) wrote,

remarkably little is known of the individual components of the system [of Algol]

Despite the work done in the past decade – notably the spectroscopic detection of Algol B by Tomkin and Lambert (1978) – this is still much the case; so there is scope for further observation of the system at optical wavelengths. In addition, the discovery of Algol as both a radio (Wade and Hjellming, 1972) and an X-ray (Schnopper et al., 1976) source calls for further observations at these wavelengths, in order that the appropriate models may be uniquely determined; a single, composite, model for radio and X-ray emission may also be in the offing.

Thus, even after thousands of years of observation. the “demon star” of Algol still merits the attention of astronomers and astrophysicists.