A discussion is made of those observations that place constraints on the pulsar emission mechanism and its location in the pulsar magnetosphere. After reviewing the salient features of pulsar radiation, a number of issues are confronted, including (1) whether coherence is broadband or narrowband and, accordingly, (2) whether the emission frequency is mapped into radius and/or magnetic polar angle; and (3) whether pulse structure (micropulses and subpulses) represents angular or temporal structure. It is proposed that pulse nulling, mode changing, and quantized drift rates for subpulses are related to discrete states of the particle flow in polar cap models.

The difficulties with “coherent” curvature emission by bunches of electrons formed through a streaming instability suggest that it cannot account for the observed radio emission for existing polar-cap models of pulsars. Amongst alternative maser mechanisms, that involving acceleration emission due to the electric fields in the waves generated by the bunching instability can account for the gross properties of the observed emission, provided that the relativistic electrons have γ ≤ 10 in the wave frame.

The light cylinder is the second most popular location for pulsar emission. The fact that it has a sizable following, and yet is not the most popular location, is due to many aspects of light cylinder models. They have several advantages in explaining the pulsar phenomenon, but they have been little developed. This is mainly because they deal with the light cylinder, where our general ignorance of magnetospheric structure is greatest. Although light cylinder models have had considerable success in explaining the optical radiation from the Crab and Vela pulsars, I will restrict this paper to a discussion of radio pulsars.

The three-dimensional structure of a pulsar magnetosphere is hard to calculate. Instead we start with a two-dimensional model (Kahn, 1971). It consists of a vacuum electromagnetic field generated by a line of rotating dipoles, from −∞ to +∞, perpendicular to the rotation axis. This field resembles that in the equatorial plane of a rotating dipole aligned perpendicular to its rotation axis. In both cases the electric field E is perpendicular to the magnetic induction B.

Over ten years ago Radhakrishnan and his coworkers in Australia detected a smooth variation of the electric vector through 90° during the brief pulses from the Vela pulsar. This observation led to the hypothesis that pulsar radiation is beamed along a magnetic pole. While subsequent observations gave ample support to this model, we find that less than half of the pulsars have average profiles which are consistent with the magnetic pole hypothesis. Statistical analyses of individual pulse polarization data now reveal that the average profiles obscure the basic nature of linear polarization in pulsars: two orthogonal modes are typically present, and individually these modes agree with the magnetic pole model. We explain the occurrence of orthogonal modes by a simple geometry for the polarized subpulse beam in relation to the magnetic axis. Theoretical models for subpulse polarization invoke both a combination of emission mechanism effects and propagation effects. We conclude with directions for further research.

Polarization observations of integrated pulse profiles have been made in seven frequency ranges between 170 MHz and 2700 MHz. Forty-nine pulsars have been observed at three or more of these frequencies. The frequency variation of the degree of linear and circular polarization is discussed for these pulsars.

An explanation is offered for the complexity and stability of integrated pulse profiles. We propose that the pulse structure arises from surface irregularities at the polar caps, and we indicate how the surface relief affects the number of charges which are released into the magnetosphere and create the observed radio radiation. The electrons produced in the vacuum break-down in the polar gap carry enough energy to continually modify the surface relief of the polar caps, and we suggest one way in which the surface could be modified in a timescale of a million years.

An investigation of the frequency dependence of the radio emission of pulsars promises to contribute considerably to the understanding of such outstanding questions as:

1. 1. Is the emission mechanism narrowband or broadband?

2. 2. What is the coherence time of the radiation?

3. 3. Does a radius-to-frequency mapping exist?

4. 4. Is pulsed radiation absorbed in the magnetosphere?

The 100-m telescope in Effelsberg is satisfactorily sensitive in the GHz frequency range so that we could observe averaged pulses up to 22.7 GHz (Bartel et al., 1977; Bartel et al., 1978) and single pulses at 8.8 GHz (Bartel et al., 1980a), which are the highest radio frequencies respectively at which pulsars have been detected. Therefore quite naturally the analysis of the frequency dependent pulsar emission became one of the focal points of interest in pulsar research at the MPIfR.

To investigate the frequency dependence of single pulse properties it is necessary to make simultaneous multi-frequency observations. This is because although some of the investigations are of a statistical nature, these should be carried out on the same pulses at the different frequencies. Further, detailed study of pulse shape, time of arrival and spectra can only be carried out on simultaneous data.

Microstructure crosscorrelation of pulses observed simultaneously at various separations in frequency in the range of 196 to 1420 MHz was done for PSR 0950+08 and PSR 1133+16 at the Arecibo Observatory. Microstructure and subpulse components were identified in the crosscorrelations, and their behaviour vs. pulse longitude was studied. Drastic differences in the subpulse and micropulse behaviour are noted, related in particular to component separation in the double peak average pulse profile of PSR 1133+16. Here the subpulses follow the average pulse profile component separation but the microstructure does not. Interpretation and constraints in terms of some current theories are also presented.

The average shape of micropulses in two pulsars is highly symmetric, unlike the subpulse emission which is skewed in the same sense as the average profiles. These conclusions stem from a third-order correlation analysis of the emission from PSR 0950+08 and PSR 2016+28. The symmetric micropulses may be produced either by temporal modulation or angular beaming of the radiation. If due to temporal modulation, this average symmetry implies a distribution of time-scales in the emission process: the difference between the rise time-scale (τr) and the decay time-scale (τd) must be small compared to δτr and/or δτd. Subpulses in PSR 0950+08 are narrower than and have the same sense of asymmetry as the average profile. In PSR 2016+28 the two subpulses typically present in a pulse are both tapered toward the outside edge of the pulse. In both pulsars, the skewness of the subpulses contributes significantly to the skewness of the average profile; a symmetrical distribution of these subpulses within the pulse window could then give rise to an asymmetrical profile.

Microstructure in pulsars occasionally has a quasi-periodic character with an average period of around 0.9 ms. A survey of this little known phenomenon (time characteristics, polarization, frequency behaviour, etc.) together with some new results are presented. Some physical interpretations (e.g. vibration of the neutron star, polar cap periodical sparking, standing wave pattern in the velocity of light cylinder model) are also presented.

The 35-km interferometer at NRAO was used at 2.7 GHz during 1974–1979 to determine pulsar proper motions. Observations from each epoch were reduced to pseudo-position vector offsets which include both unknown, and constant, absolute position errors for the pulsars and their reference sources, as well as the time-varying proper motion of the pulsars. The coordinate system for which the errors in the offset components are independent is chosen. Proper motions resulting from a linear fit to the offsets for five pulsars are given in Table 1.

Two hypotheses exist to explain the radio interpulse of PSR 0950+08 within the context of polar cap emission models:

1. 1) The emission originates at a single magnetic pole as for other two-component pulsars, except that the component spacing is much wider.

2. 2) The interpulse emission originates at the opposite magnetic pole. We discuss here some new observations and their implications for the single and double pole hypotheses.

PSR 0031-07 has a period of 0.94 seconds and the relatively old timing age of 37 million years. Its properties were first established by Huguenin et al. (1970), based on observations at 145 and 400 MHz.

PSR 1822–09 is a nearby galactic disk pulsar with a typical .769 s period. Its large period derivative indicates a young timing age of 250,000 yrs and a strong estimated surface magnetic field of 6.4 × 1012 G. We have observed this pulsar at frequencies around 1700 MHz and 2650 MHz. The mean pulse profile at these frequencies consists of a double peaked main pulse with two dissimilar components and a weak interpulse 185° of pulse longitude after the main pulse second component. The interpulse is detected in both frequency ranges, and the main pulse - interpulse separation does not change between these frequencies and the low frequencies around 327 MHz where the interpulse was discovered by Cady and Ritchings (1977).

The phenomenon of pulse nulling was first found by Backer (1970). The fraction of time, we denote it by n(%), for a pulsar in a null state was given by Ritchings (1976) for 32 pulsars. We have found that the average value of n(%) increases with the period of pulsars. It perhaps implies that the pulsars with pulse nulling could constitute a subtype of pulsars. In this paper, the pulsars with pulse nulling are regarded as a subtype to be investigated.

The structure of the pulsar magnetosphere, the location of the radio emission region and the radio emission mechanism are important theoretical subjects in the research of pulsars. There may be close relations between these subjects. Nevertheless, it should be possible to set up some empirical relations among them which can be considered as a foundation for studying these important subjects. In this paper some of these relations are studied.

Optical identification of radio pulsars has only been achieved for the Crab and Vela Pulsars. The luminosity limits for others could be considerably improved: better astrometry and time-resolved photometry could reach mv = 27. The limits for pulsars with unknown periods, for example in extragalactic nebulae, are usually about mv = 20. The luminosity falls rapidly with increasing period: if it follows a power law the index is minus ten at least.

The optical spectrum of the Crab Pulsar falls toward the infra-red and is flat in the ultra-violet. No self-absorption effect is seen in the infra-red. New observations of the minimum intensity and of the polarisation show a highly linearly polarised component continuing through the whole pulse cycle.