A new structural analysis of the clean Cu(100) surface by low energy electron diffraction yields a surprising in-plane lattice contraction of about 1% compared to the bulk lattice parameter, i.e. , a surface reconstruction following the tensile stress in the surface. This sheds new light on the epitaxial growth of other metals on Cu(100). Additionally, we report on a similar contraction determined for the metastable 1 X 1 phase of the (100) surface of platinum indicating that in-plane lattice reconstruction might be a more general feature than believed.PACS numbers: 68.35.Bs, 61.14.Hg, 61.66.Bi, 68. 10.Cr Structure analyses of unreconstructed surfaces of clean crystals usually do not aim at the determination of the surface parallel (or in-plane) lattice parameter a". By symmetry arguments the latter is believed to be the same as in bulk layers; otherwise structural defects must be introduced when the surface is created. On the other hand, each surface exhibits a tensile stress [1 -9] favoring the formation of more densely packed layers. In fact, a few clean surfaces as the stable (100) surfaces of Au, Pt, and Ir and the (111) surfaces of Pt and Au are known to reconstruct in this sense. As these examples are considered rather exceptional, we show in the present Letter that the feature of in-plane reconstruction might be more general (though not universal): Cu(100), when prepared in a conventional way, reconstructs by a 1% contraction of the in-plane lattice parameter. The same holds for the metastable 1 X 1 phase of Pt(100) possibly being a precursor to the full hexagonal reconstruction.Our reinvestigation of Cu(100) was triggered by work on the heteroepitaxial growth of metals on copper.For Fe/Cu(100) a"(Fe) = 2.52 A was found [10 -13] contrasting a"(Cu) = 2.55 A for bulk copper. In fact, nonpseudomorphic growth of Fe was concluded [10,11].For the system Ni/Cu(100) a value a"(Ni) = 2.53 A was determined [14,15]. These results suggest that it might be a reduced in-plane lattice parameter of Cu(100) that determines the lattice of the epitaxial films. We applied low energy electron diffraction (LEED) for the determination of a~. In view of the small contraction we had to accept that -different from the surfaces of Pt, Ir, and Au -no superstructure spots show up when a conventional LEED optics is used. So, detection of the reconstruction can only come by the measurement of accurate data and their careful dynamical analysis.The copper crystal (5% purity,~0 .5 orientation accuracy) was mechanically polished (1 p, m minimum grain size). /n situ cleaning by Ar ion bombardment (600 V, 1 /LA/cm, 800 K) and annealing (1000 K) yielded a surface with impurities like 0, C, and S below the Auger detection limit (~3%) and a low background diffraction pattern. Intensities were taken at 90 K to reduce thermal diffuse scattering. A video based data acquisition system was used [16 -18] whose speed made residual gas adsorption negligible. Measurement of five beams symmetrically inequivalent at normal incidence produced a database...