In case a mirror world with a copy of our ordinary particle spectrum would exist, the neutron n and its degenerate partner, the mirror neutron n ′ , could potentially mix and undergo nn ′ oscillations. The interaction of an ordinary magnetic field with the ordinary neutron would lift the degeneracy between the mirror partners, diminish the n ′ -amplitude in the n-wavefunction and, thus, suppress its observability. We report an experimental comparison of ultracold neutron storage in a trap with and without superimposed magnetic field. No influence of the magnetic field is found and, assuming negligible mirror magnetic fields, a limit on the oscillation time τ nn ′ > 103 s (95% C.L.) is derived.PACS numbers: 11.30. Er, 11.30.Fs, 14.20.Dh, The concept of a mirror world, as an attempt to restore global parity symmetry, has attracted interest since the 1950's, started by the famous paper of Lee and Yang [1] and significantly expanded in the work of Kobzarev, Okun, and Pomeranchuk [2]. The mirror matter idea was first applied to the Standard Model of particle physics in [3]. More recent overviews can be found in [4,5]. The mirror world could hold a copy of the particle spectrum of our ordinary world. Matter and mirror matter would interact via gravity and present a viable explanation to the dark matter problem [6,7,8,9,10]. Besides gravity, other (new) interactions could show up in minute mixings of neutral matter particles -such as neutrinos, pions, kaons, or positronium (see [11] for e + e − ) -and their degenerate mirror partners making oscillations between them possible. Recently it was pointed out [12] that no direct experimental limits exist on the oscillation time τ nn ′ [13] between ordinary matter neutrons (n) and the speculative mirror neutrons (n ′ ). An indirect limit of the order τ nn ′ ≥ 1 s has been derived in [12] based on the search for neutron -antineutron (nn) oscillations [14]. Fast nn ′ oscillations with τ nn ′ ∼ 1 s, or at least much shorter than the neutron β-decay lifetime, could explain [12,15] the origin of ultra-high energy cosmic rays above the Greisen-Zatsepin-Kuzmin (GZK) cutoff [16,17]. The viability of models and implications have been further discussed in [18].Possible approaches to nn ′ oscillation experiments with sensitivities of several hundred seconds have been discussed in [19]. One approach is to search for nn ′ oscillations by comparing the storage of ultracold neutrons (UCN) in vacuum in a trap in the presence and the absence, respectively, of a magnetic field. The essential idea is that the neutron and mirror neutron states would be degenerate in the absence of a magnetic field and nn ′ transitions could occur. (The absence of mirror magnetic fields at the location of the experiment is assumed throughout this paper [20].) The interaction of the neutron with a magnetic field would lift the degeneracy and suppress the transition into a mirror neutron which, of course, does not interact with the ordinary magnetic field, nor with the trap via the ordinary strong interaction. Thus...