A number of long-standing puzzles, such as the origin of ultrahigh-energy cosmic rays, could perhaps be solved if we found a mechanism for effectively transferring energy from black holes to particles and, correspondingly, accelerating the latter to (unboundedly, as long as we neglect the back reaction) large velocities. As of today the only such candidate mechanism in the case of the non-extreme Kerr black hole is colliding a particle that freely falls from infinity with a particle whose trajectory is subject to some special requirements to fulfil which it has to be suitably corrected by auxiliary collisions. In the present paper we prove that---at least when the relevant particles move in the equatorial plane and experience a single correcting collision---this mechanism does not work too. The energy of the final collision becomes unboundedly high only when the energies of the incoming particles do.

In a stationary spacetime $S$ consider a pair of free falling particles that collide with the energy $E_{\rm c.m.}$ (as measured in the center-of-mass system). Let the metric of $S$ or/and the trajectories of the particles depend on a parameter $ k$. Then $S$ is said to be a "(super) accelerator" if $E_{\rm c.m.}$ grows unboundedly with $ k$, even though the energies of the particles at infinity remain bounded. The existence of naturally occurring super accelerators would make it possible to observe otherwise inaccessible phenomena. This is why in recent years a lot of spacetimes were tested on being super accelerators. In this paper a wormhole $W$ of an especially simple---and hence, hopefully, realistic---geometry is considered: it is static, spherically symmetric, its matter source is confined to a compact neighbourhood of the throat, and the $tt$-component (in the Schwarzschild coordinates) of its metric has a single minimum. It is shown that such a wormhole is a super accelerator with $ k\equiv \frac 13\ln |g_{tt\ \mathrm{min}}|$. In contrast to the rotating Teo wormhole, considered by Tsukamoto and Bambi, $W$ cannot accelerate the collision products on their way to a distant observer. On the other hand, in contrast to the black hole colliders, $W$ does not need such acceleration to make those products detectable.

I discuss a special class of singularities obtained as a natural 4-dimensional generalization of the conical singularity. Such singularities (called quasiregular) are ruinous for the predictive force of general relativity, so one often assumes (implicitly as a rule) that they can be somehow excluded from the theory. In fact, however, attempts to do so (without forbidding the singularities by fiat) have failed so far. It is advisable therefore to explore the possibility that their existence is not prohibited after all. I argue that quasiregular singularities, if allowed, may appear either in situations where causality is endangered or in the early Universe. In the latter case objects might appear strongly (though not quite) resembling cosmic strings. Those objects would be observable and, moreover, it is not impossible that we already do observe one.

I present a class of objects called gravitational strings (GS) for their similarity to the conventional cosmic strings: even though the former are just singularities in flat spacetime, both varieties are equally "realistic", they may play equally important cosmological r\^ole and their lensing properties are akin. I argue that the enigmatic object CSL-1 is an evidence in favor of the existence of GS.

The paper has been withdrawn by the author, due to a fatal error. A horse stumbles that has four legs.

Suppose, the Universe comes into existence (as classical spacetime) already with an empty spherically symmetric macroscopic wormhole present in it. Classically the wormhole would evolve into a part of the Schwarzschild space and thus would not allow any signal to traverse it. I consider semiclassical corrections to that picture and build a model of an evaporating wormhole. The model is based on the assumption that the vacuum polarization and its backreaction on the geometry of the wormhole are weak. The lack of information about the era preceding the emergence of the wormhole results in appearance of three parameters which -- along with the initial mass -- determine the evolution of the wormhole. For some values of these parameters the wormhole turns out to be long-lived enough to be traversed and to transform into a time machine.

The possibility is discussed of superluminal motion of non-tachyonic (i.e. moving with the instantaneous speed v<c) bodies within general relativity. It is shown that in some occasions quantum field theory apparently prohibits such motion, but not - to all appearance - in the general case.

Recently the structure of the Letelier-Gal'tsov spacetime has become a matter of some controversy. I show that the metric proposed in \cite{letgal} is defined only on a dense subset of the whole manifold. In the case when it can be defined on the remainder by continuity, the resulting spacetime corresponds to a system of parallel cosmic strings at rest w.r.t. each other.