On Earth, new ocean crust is continually created at mid-ocean ridges, and subducted again into the Earth's mantle at subduction zones. There is good evidence from the chemical compositions of oceanic basalts that the mantle is heterogeneous as a result of this subduction of oceanic crust, and contains layers of subducted ocean crust that have not been homogenized on a billion year time scale. However, this mantle heterogeneity, consisting of normal mantle rocks (peridotite) and subducted ocean crust ('eclogite'), has never been observed directly in mantle rocks. We set out to test whether traces of billion year old layers of 'eclogite' are preserved and can indeed be observed and analysed in oceanic mantle rocks. A positive result would mean that we have direct evidence that oceanic crust is recycled in the Earth, i.e., that after subduction in can resurface again after a billion year or so. We studied the mantle section of the Jurassic Pindos Ophiolite (Greece), using field analysis and petrology, in combination with Re-Os isotope and oxygen isotope geochemistry. In the first phase of the project, we have confirmed - using Osmium isotopes - that traces of 1-3 billion year old rocks that resemble eclogite are indeed preserved in the form of distinct layers of pyroxenite (a rock primarily consisting of the mineral pyroxene) that are found parallel to high temperature mantle deformation structures. We have proposed a model in which these pyroxenites are formed by an in-situ reaction between mantle rocks and melts derived from the melting of layers of billion-year-old eclogite. In the second phase of the project, we are planning to carry out oxygen isotope analysis on these pyroxenites, to test whether evidence can be found that the ancient eclogite rocks were once altered at the seafloor. Seafloor alteration of ocean crust should leave a distinct oxygen isotope fingerprint, which may be preserved once subducted into the mantle. Finding this signature in the pyroxenite layers in Pindos would be evidence for the process of recycling of ocean crust (formation of ocean crust at mid-ocean ridge, alteration at ocean floor, subduction back into the mantle, storage in the mantle on a billion year time-scale, resurfacing again underneath a mid-ocean ridge). Moreover, we set-out to test whether reactions between mantle rocks and melts derived from the melting of eclogite can indeed produce pyroxenites similar to those observed in Pindos. We identified an ideal test area, the Totalp Massif near Davos, where intermediate stages of this process can be very well observed. Osmium and oxygen isotope analysis of the samples collected in this massif is in progress.