J. Astrophys. Astr. (1984) 5, 187–207 Millisecond Pulsars D. C. Backer Radio Astronomy Laboratory and Astronomy Department, University of California, Berkeley CA 94720, USA (Invited article) Abstract. In 1982 we discovered a pulsar with the phenomenal rotation rate of 642 Hz, 20 times faster than the spin rate of the Crab pulsar. The absence of supernova debris in the vicinity of the pulsar at any wavelength indicates an 5 age of the neutron star greater than 10 yr. The miniscule spindown rate of 1.1 – 19 9 × 10 confirms the old age and indicates a surface magnetic field of 10 G. A second millisecond pulsar was discovered by Boriakoff, Buccheri & Fauci (1983) in a 120-day orbit. These fast pulsars may have been spun-up by mass transfer in a close binary evolutionary stage. Arrival-time observations of the 642-Hz pulsar display remarkably low residuals over the first 14 months. The stability implied by these observations, 3 × 10 – 14, suggests that millisecond pulsars will provide the most accurate basis for terrestrial dynamical time. If so, the pulsar data will lead to improvements in the planetary ephemeris and to new searches for light-year scale gravitational waves. Many new searches for fast pulsars are under way since previous sky surveys excluded pulsars with spins above 60 Hz. Key words: pulsars: 1937 + 21, 1953 + 29—pulsar surveys—time—gravita- tional waves 1. Introduction “If the neutron star hypothesis of the origin of Supernovae can be proved, it will be possible to subject the general theory of relativity to tests which according to the considerations presented in this paper deal with effects which in order of magnitude are large compared with the tests so far available. The general theory of relativity, although profound and exceedingly satisfactory in its epistemological aspects, has so far practically not lent itself to any very obvious and generally impressive applications. This unfortunate discrepancy between the formal beauty of the general theory of relativity and the meagerness of its practical applications make it particularly desirable to search for phenomena which cannot be understood without the help of the general theory of relativity.” F. Zwicky (1939) These prescient comments of Fritz Zwicky are remarkable in the light of recent discoveries of degenerate neutron stars in a heretofore unimagined variety of configurations. Two key developments opened the door to neutron-star investigations. Exploration of the sub-second domain of radio source variability led to the discovery of 188 D. C. Backer a 1.377-s pulsar (Hewish et al. 1968) which was subsequently identified as a neutron star (Gold 1968). The opening of the X-ray spectrum with rocket flights and satellite launches led to the detection of neutron stars accreting mass in close binary systems (Giacconi et al. 1971). Shortly after the pulsar discovery a 33-ms pulsar was discovered coincident with the peculiar 16-mag star noted by Baade (1942) and Minkowski (1942) in the centre of the Crab nebula supernova remnant. Seven years later Hulse & Taylor (1975) discovered a pulsar orbiting a second neutron star. The slow decay of the orbital period of this system has provided the first evidence for gravitational radiation predicted in the general theory of relativity; Zwicky’s dream has been realized. In 1982 the observable range of stellar rotation was extended by a factor of 20 with the discovery of a 1.558-ms pulsar in Vulpecula, not far from the location of the first pulsar (Backer et al. 1982b). Observations subsequent to the discovery soon resolved an apparent discrepancy between the pulsar’s rapid spin, which indicated a youthful object, and the absence of supernova debris at any wavelength, which indicated a minimum age of 105 yr. The pulsar’s spin, Ω = 2π/Ρ, was decaying extremely · 8 slowly, Ω/Ω = P/P· ~ 5 × 10 yr. This new member of the cosmic menagerie placed the pre-discovery speculations of Radhakrishnan (1982) and others concerning anomalous · pulsars in a P–P diagram on firmer ground. Radhakrishnan had identified moderately fast pulsars with low spin-decay rates as neutron stars that had been ‘recycled’ to their present fast spins by angular momentum transfer from an evolving secondary. In the following pages I will recount the path which led to the discovery and then summarize three areas of inquiry which have been stimulated by this discovery: (1) further observations and investigations of neutron star astrophysics; (2) pulse arrival-time measurements and gravitational wave physics; (3) search for new fast pulsars that were excluded in past surveys of the galaxy owing to computational limitations and to a prejudice concerning the period distribution. Scenarios for the origin of the fast pulsars are discussed by van den Heuvel (1984) elsewhere in this issue. 2. Discovery—persistence pays off In 1977–78 Stuart Vogel and I were investigating with VLBI the radio sources in the Cygnus region of the sky to assess the prevalence of interstellar scattering at low galactic latitudes. The source 4C 21.53 came to our attention since it displayed strong interplanetary scintillations (IPS) despite its low galactic latitude (Duffett-Smith & Readhead 1976). Pulsar observations had indicated that at low galactic latitudes interstellar scattering (ISS) would suppress the IPS modulation. The steep intensity spectrum of the object added to its peculiar nature. While identification of 4C 21.53 with a pulsar would have explained its peculiar properties, no known pulsar was within the errors of the 4C 21.53 position. While searching the literature for references to 4C 21.53 in January 1979, I found a source, 1937 + 215, located 30 s west of the 4C 21.53 in several published catalogues. This source could be identified with 4C 21.53 if the 4C position was in error by one lobe; lobe errors in the 4C Catalogue occur with a frequency of about 3 per cent(Backer et al. 1970). The difficulty with connecting 1937 + 215 and the IPS source was that the spectra were very different. Furthermore the large size, 60 arcsec, for 1937 + 215 in the 5-GHz catalogue of Altenhoff et al. (1979) confounded the lobe-error hypothesis since the IPS object was necessarily smaller than 1 arcsec. My initial synthesis of these observations Millisecond pulsars 189 was that the steep-spectrum, IPS object was a pulsar co-located with a faint (1 Jy), extended supernova remnant, 1937 + 215. Curiously no pulsar had been detected in this region in the very sensitive Arecibo survey (Hulse & Taylor 1974). While interstellar scattering could have smeared the pulsation of a pulsar so that it was not detectable in the 430-MHz survey, the IPS observations at 81 MHz discussed above had already indicated that ISS was not severe for this object. A simple calculation demonstrated that ISS could only smear pulsations for periods of order 10 ms or less. The Arecibo survey, and most others, were not sensitive to periods below 60 ms. A report on the hypothesis that 1937 + 215 was a young pulsar-supernova pair similar to the Crab nebula and its pulsar—published prior to the discovery of pulsars (Hewish & Okoye 1965)—was returned from a journal with the referee’s comment as ‘too speculative’. Several new pieces were added to the puzzle early in 1979. Mike Davis at the Arecibo Observatory measured the spectrum of 1937 + 215 between 430 MHz and 2.4 GHz and conducted pulse searches. The Arecibo pulse searches were doomed to failure owing to the imprecision of the position at this time. Rickard & Cronyn (1979) briefly discussed the peculiar properties of 4C 21.53 in a paper which proposed a new class of compact, low-latitude objects which they called ‘scintars’. David Cudaback, Stuart Vogel, John Middleditch and I conducted a pulsar search with the 90-foot antennas at Owens Valley Radio Observatory in 1979 March at 600 MHz. These data had a Nyquist frequency of only 50 Hz and were corrupted by extensive interference. I then learned about the independent investigations by Erickson (1979, personal communication) with the VLA in September 1978 which showed the presence of a second source 30s east of 4C 21.53. The steep spectrum and compactness of the eastern component suggested that it was the IPS object. The pulsar-supernova hypothesis now seemed unwarranted. In 1981 my interest in the western source, 1937 + 215, was rekindled by a report from Erickson (1980) that 34-MHz data from the Clark Lake Radio Observatory indicated IPS sources at the positions of both the eastern and the western components of 4C 21.53. The pulsar-supernova-remnant hypothesis was resurrected. I recalled a discrepancy between the catalogued positions of 1937 + 215 at cm wavelengths and those at metre wavelengths, viz., the 365-MHz position of Douglas et al. (1980) and the 81-MHz position of Slee (1977). I reasoned that these measurements indicated a southward shift of the steep spectrum IPS source from the extended source. In 1982 March, Miller Goss and I obtained a 610-MHz image of 1937 + 215 with the Westerbork Synthesis Radio Telescope which confirmed the suspected offset. This was the first hard evidence for the location of the IPS source in the western component. The intensity, 0.13 Jy, confirmed the steep spectrum seen at decametric wavelengths. A brief VLA measurement in 1982 May further isolated the IPS object from the extended source to the north. When the Westerbork data became available, Val Boriakoff at the Arecibo Observatory conducted a pulse search which was sensitive to periods as short as 4 ms. Again no pulsar was detected. In 1982 August, lively discussions at the Patras IAU meeting about the riddle of the western components of 4C 21.53 gave further investigations high priority.