Assume that a client outsources his database to a remote storage-provider (the server), so that for privacy reasons, the client's database is encrypted by his secret key. During a PIR-writing protocol, the client updates one element of the encrypted database without revealing to the semi-honest server which element was updated and, of course, to which value. The best previous PIR-writing protocols had square-root communication complexity. In this paper, we propose two new PIR-writing protocols. The first one can be based on (say) the Damg