“…For this reason, if defects do exist in these models, it is crucial to understand their evolution and consequences in both the expanding and collapsing phases. Up to now * Electronic address: ppavelin@fc.up.pt † Electronic address: C.J.A.P.Martins@damtp.cam.ac.uk ‡ Electronic address: cssilva@fc.up.pt § Electronic address: E.P.S.Shellard@damtp.cam.ac.uk all these studies, be they analytic [9,10,11,12,13] or numerical [14,15,16], have only been undertaken for the expanding case, and it is clear that while some results may be expected to carry over to the contracting phase, some others clearly won't. This will have consequences not only for the standard string seeded (or hybrid) structure formation scenario [17], but also for other 'non-standard' scenarios involving defects, such as the production of adiabatic and nearly Gaussian density fluctuations [18,19], or those involving anisotropic or inhomogeneous universes [20,21].…”