Dear colleagues, By asking our American colleagues if they had encountered this problem we were able to isolate it to our own system. We finally traced it to our setting up our PCs to read Arabic in non-unicode programmes in order to read our records in the older Innovative system Guicat. This is not a problem on the newer Millenium cataloguing system which uses Unicode, where the four Persian adaptations of Arabic letters are also no longer a problem. Peter Colvin Dear Colleagues, Has anybody else started to download Arabic records from the new OCLC Connexion. We are experiencing a rather disastrous problem in which the dot under the letter (which should be coded as {242}) is replaced by a question mark as is the macron (which should be coded as {229}) However the ayn {176} and the hamza {174} download correctly. Please let me know if you are experiencing the same or different problems as I would like to establish if it is a local problem or a problem at the OCLC end. Peter Colvin