There are no physical principles disallowing the existence of extended sources of the electromagnetic field which move faster than light in empty space (Bolotovskii and Ginzburg 1972). Charge-current distributions whose patterns rotate around a fixed axis rigidly, and so have a phase speed that is greater than the speed of light in vacuo outside the light cylinder, emit radiation—different from both Cherenkov and synchrotron radiation—which has the following characteristics:

1. 1. It arises almost exclusively from the vicinity of those points within the source which, at the emission time, approach the observer with the speed of light and with zero acceleration.

2. 2. It consists of the superposition of a (continuous) set of narrow radiation beams whose flux densities fall off with the radial distance R from the source like R–1 (rather than R–2), and so it is intrinsically coherent.

3. 3. It has a power-law spectrum with no exponential cut-off, at high harmonics of the rotation frequency, whose index lies between –1 and –3 if the longitudinal distribution of the source density is pulse-like.

4. 4. It consists of two concurrent elliptically polarized signals whose position angles are predominantly orthogonal and simultaneously vary in direction in the course of each rotation.

I suggest that this radiation, here referred to as Schott radiation (cf. Schott 1912), may in fact be that received from pulsars.

Bunch curvature emission is one of the well known pulsar radio emission mechanisms. The problem of bunch formation is very important for understanding the pulsar radio emission mechanism. In the axisymmetric pulsar magnetospheric bunching arises in the outflow channel as a result of interaction between moving electrons (DP) and captured ones (SP) (Rylov 1988). A numerical simulation was undertaken to determine how strongly the electron flow is bunched. The bunching appears to be very strong. It can be treated as a gas of electron bunches rather than small fluctuations of the electron flow. The thermal radio emission of the electron-bunch gas has a very high brightness temperature and sharp directivity. The power consumed in electron-bunch gas heating is sufficient to explain the pulsar radio emission. The pulsar radio emission mechanism appears to be thermal (the electron bunches move chaotically) and coherent (electrons of the bunch emit coherently) at the same time. For this reason the radio emission mechanism is very stable.

Among the various plasma instabilities which could be responsible for coherent pulsar radio emission, we investigate the two-stream instability, first introduced by Ruderman and Sutherland (1975) in order to account for the physical situation expected in the environment of neutron stars. They describe how, in a polar cap model, pair creation arises and leads to the formation of a very energetic beam of e+ (and/or e−) and of an e−e+ plasma, both with relativistic bulk motion along the bundle of dipolar magnetic field lines. The study of their interaction is limited to the cone of open B lines, a site which provides a natural geometry for the radio emission zone, observed as core and/or conal emission by Lyne and Manchester (1988) and Rankin (1983, 1986, 1990).

We have examined high quality Arecibo data of three pulsars for evidence of strange attractors in pulse-to-pulse intensity fluctuations. Significant structure was found for PSR 0823+26, and seems related to pulse drifting and nulling activity. The low dimensionality of the possible attractor suggest that a simple dynamical model can explain this component of the flux variability.

480 single pulses from PSR 0809+74, recorded at 102.5 MHz with a time resolution of 10μs, have been analyzed by the time delay technique in order to look for the parameters of deterministic chaos in the microstructure of radio pulses. The correlation dimension n was shown to be less than 5 in more than 20% of the analyzed pulses. This means that on such occasions the microstructure of pulsar radio emission with the time scales 10 to 100μs may be determined by the behavior of a nonlinear dynamical system with comparatively small numbers of independent parameters. For example, the observed low dimensional chaos may result from a turbulence process associated with the outflow of plasma—as in versions of the polar cap model where microstructure can be interpreted as reflecting the spatial structure of relativistic plasma outflow in the radio emission region.

However, the correlation-time distribution demonstrates the tendency for microstructure to consist of a random sequence of unresolved micropulses in the majority of cases, which means in the framework of polar cap models the presence of well developed turbulence in the relativistic plasma outflow.

Dynamic autocorrelation functions for 2.56-ms adjusted intervals of single pulses of PSR 0809+74 have been analyzed. The observations were conducted at 102.5 MHz with 10μs resolution. We find that the neighboring 2.56-ms subpulse intervals seem to have independent microstructure parameters, although sometimes the tendency was observed for the occurrence of quasi-periodic structures of close or double periods within the same subpulse. The distribution of a number of well ordered micropulses in quasi-periodic structures was shown to agree with the hypothesis of a random origin of quasi-periodicities. No relation has been found between the microstructure parameters and subpulse longitude.

Our primordial item (~10 years ago) was to use short “unresolved” spikes, which are seen within individual pulses in some pulsars after dispersion removal (Hankins 1971) for precise measurements of dispersion measure (DM). For this purpose we performed three-frequency observations in the 60-102 MHz frequency range for PSR 0950+08, 0809+74 and 1133+16 (Novikov et al. 1983). The result was that “classical” microstructure (with 0.2, 1.2, and 0.55-ms time scales, respectively) is well correlated over this frequency range, but those bursts with a shorter duration (~100μs) do not show any correlation. In particular, no coincidence of the short “unresolved” spikes at two different frequencies was detected. On the basis of these results we reduced the frequency spacing to 1-2 MHz, but the result was the same (Popov et al. 1985).

As is very well known, there are two different kinds of time structure of the individual pulses of pulsars. There are microstructures with time scales less than 1 ms and subpulse structures, the usual scale of which is of the order of 10 ms. Subpulses are a result of the radiation beam rotation and in many cases demonstrate some regularity of arrival phases. In contrast, microstructure looks rather like a short pulse of noise with random phases of micropulse appearance in the radio emission window of the pulsar. Because the appearance phases of micropulses are random in the pulsar period, it is very difficult to establish what the fundamental character of these micropulses is: are they a temporal variation of radio emission intensity or the result of narrow beam rotation?

The properties of pulsar radio pulse microstructure are reviewed, then consideration is given on how, in the frame of the Ruderman-Sutherland pulsar model, an emission mechanism can be devised which explains many of the known characteristics of micropulses and of subpulses.

We have examined approximately 550,000 pulses from PSR 0531+21 and have detected 1814 giant pulses 6σ or more above the background noise level. We hae measured the relative arrival times and the energies of both the main pulse and interpulse, and have found that the giant pulses arrive somewhat earlier than the peaks in the average profile. Approximately 7% of the main pulse and 1% of the interpulse energy is in the form of giant pulses 8σ more above the background level. No giant pulses have been detected in the precursor at the 8σ level. In addition we have found a slight, although apparently real, rise in emission in the region where an interpulse precursor would be if the emission is symmetric.

We have made pulsar microstructure observations of PSR 1133+16 at the Arecibo Observatory. Cross-correlation of linearly polarized microstructure between 111 MHz and 318 MHz results in a value of the dispersion measure of DM = 4.8462 ± 0.0004 pc cm–3 at epoch 1989.9. This value is consistent with the average profile determination of Phillips and Wolszczan (1990) obtained at epoch 1989.2, but not with the microstructure value found by Popov et al. (1987) in early 1980. Time variation of the integrated electron content over this ten-year span is a plausible explanation for the discrepancy.

By using Cross Spectral Density processing between high time resolution radio intensity observations of pulsars made simultaneously at two different receiver frequencies, several high frequency periodicities that are persistent over many pulses were discovered. A possible interpretation in terms of neutron star vibration models is discussed.

A nonlinear plasma model which may account for temporal modulation of pulsar radio pulses is presented. Envelope solitons and envelope nonlinear wave trains can result from the nonlinear interaction of the high-frequency coherent pulsar radiation with the pulsar magnetosphere. Theories of electromagnetic envelope solitons and electromagnetic envelope nonlinear wave trains in electronpositron plasmas are reviewed. The application of this model for observation of pulsar microstructures is discussed.

A new radio luminosity equation for pulsars is presented which is based on the geometry of a polarcap model. Statistical results show that the emission component intensity distribution is Gaussian. There are 100 pulsars with new radio luminosity values. We show that the radio luminosity can be described by one relation L ∝ (BP2)1.07 or L ∝ Q'–1.4, Q' = 2P1.1Ṗ–15–0.4 over the whole range of period and period derivatives, contrary to other recently suggested luminosity laws.

The numerous discussions that took place at the colloquium have reemphasized the primary importance of polarization observations and their interpretation for understanding the magnetospheric structure and the radiation mechanisms of pulsars. I have tried to take them into account in the summary at the end of this written version of my review of polarization phenomena, where I also attempt to address some of the questions raised by both observers and theorists as to what message polarization has for the planning of future observations and for the development of theories to explain pulsar radiation mechanisms. Because I feel it is relevant I shall begin with a little history to put things in perspective.

Observational data show that many pulsars exhibit a moderate amount of circular polarization (CP) as well as linearly polarized emission. Generally, the degree of CP is of the order of several percent, but some of pulsars have higher CP—about 20% (Manchester and Taylor 1977, Taylor and Stinebring 1986) and PSR 1702-19 has 60% (Biggs et al. 1988). No satisfactory explanation for the existence of this CP is now known, and we propose in this paper to outline a possible mechanism for the generation of this CP consistent with our theory of pulsar radio-wave excitation. In Kazbegi et al. (1992; hereafter Paper I) we mentioned that the distribution functions of electrons F− and positrons F+ are shifted with respect to each other due to the presence of the primary electron beam and the quasi-neutrality of the plasma as a whole.

It is shown that circular polarization occurs in the region of cyclotron resonance because the group velocities of right-hand and left-hand polarized waves are different with respect to the direction of the magnetic field. Due to the dependence of the intensity of radio emission on the coordinates across the polar cap, this difference in group velocities leads to noncompensated circular polarization proportional to the derivative of the total intensity as a function of longitude. The indicated dependence corresponds to observations of the so-called core component of pulsar radio emission.

In the popularly accepted empirical model for pulsar emission, bunches of charged particles traveling along open field lines near the magnetic pole emit curvature radiation. Such radiation is linearly polarized. Most of the radiation is assumed to be emitted from a ring shaped region centered on the pole (the hollow cone model). We have calculated the expected average polarization using this model and find them to be in disagreement with observations. The addition of a second ring inside the first with orthogonal polarization solves this problem. This new model explains other observed features of pulsar emission including discontinuities in position-angle profiles and multicomponent profiles. Plasma interactions might account for a second ring.

Polarimetric observations of over 300 pulsars have been carried out between 21 December 1988 and 22 January 1990 at 606, 610, 925, and 1408 MHz using the Lovell Telescope at Jodrell Bank. Many of these pulsars have no previously published polarization profiles and will be published shortly (Gould and Lyne 1990). This large data set along with previously published data from various sources, has been used to test the correlation found by Radhakrishnan and Rankin (1990) between sense reversing circular polarization signatures and the accompanying sense of rotation of the linear polarization position angle.

A brief review of polarization observations at 103, 60, and 40 MHz is given. Our peculiar measurement technique allows us to obtain average polarization profiles as well as statistical distributions of the polarization parameters of individual pulses and subpulses. Some examples are given in this paper.