(ed.). Gravitational waves (IOP, 2001)(422s).

296 Elementary introduction to pre-big bang cosmologysecond generation interferometric detectors (LIGO [34], VIRGO [35]), orfor interferometers in space (LISA [36]).• Type II: the large-scale CMB anisotropy ‘seeded’ by the inhomogeneousfluctuations of a massless [37] or massive [38] axion background. Metricfluctuations are indeed too small, on the horizon scale, to be responsible forthe temperature anisotropies detected by COBE [39]; the axion spectrum,in contrast, can be sufficiently flat [40] for that purpose. Such a differentorigin of the anisotropy may lead to non-Gaussianity, or to differences (withrespect to the standard inflationary scenario) in the height and position ofthe first Doppler peak of the spectrum [41]. Such differences could be soonconfirmed, or disproved, by the planned satellite observations (MAP [42],PLANCK [43], ...).• Type III: the production of primordial magnetic fields strong enough to‘seed’ the galactic dynamo, and to explain the origin of the cosmic magneticfields observed on a large (galactic, intergalactic) scale [44]. In the standardinflationary scenario, in fact, the amplification of the vacuum fluctuationsof the electromagnetic field is not efficient enough [45], because of theconformal invariance of the Maxwell equations. In string cosmology, incontrast, the electromagnetic field is also coupled to the dilaton, and thefluctuations are amplified by the accelerated growth of the dilaton during thephase of pre-big bang evolution.Finally, we wish to mention a further important phenomenogical effect,typical of string cosmology (and that we do not know how to classify withinthe tree types defined above, however): dilaton production, i.e. the amplificationof the dilatonic fluctuations of the vacuum, and the formation of a cosmicbackground of relic dilatons [46].The possibility of detecting such a background is strongly dependent onthe value of the dilaton mass, that we do not know, at present. If dilatons aremassless [47], then the amplitude and the spectrum of the relic background shouldbe very similar to those of the graviton background, and the relic dilatons couldbe possibly detected, in the future, by gravitational antennae able to respond toscalar modes, unless their coupling to bulk matter is too small [47], of course.If dilatons are massive, the mass has to be large enough to be compatible withexisting tests of the equivalence principle and of macroscopic gravitational forces.In addition, there is a rich phenomenology of cosmological bounds, which leavesopen only two possible mass windows [46]. Interestingly enough, however, in theallowed light mass sector the dilaton lifetime is longer than the present age of theuniverse, and the dilaton fraction of critical energy density ranges from 0.01–1: inthis context, the dilaton becomes a new, interesting dark matter candidate (see [48]for a detailed discussion of the allowed mass windows, and of the possibilitythat light but non-relativistic dilatons could represent today a significant fractionof dark matter on a cosmological scale). We have no idea, however, of how todetect directly such a massive dilaton background, because the mass is light, but