Sorry the mail was sent in the middle of the post. __________________________________________ not actually catalogs.xml: <catalog> <item> <name>music item 1</name> <type>cd</type> </item> ... all other items to be listed in the catalog. </catalog> now if I need something like <sites> it will be totally irrelevant to put it into the catalogs xml. <sites> <site>...</site> <site>...</site> </site> So I have two root nodes, one for sites, one for catalogues: and two seperate xml documents. logically consistent. If I want to enlarge sites xml and say add domains to it. Then it is possible that I am thinking of a larger entity: <web-settings> <sites> <site></site> <site></site> </sites> <response-filters> <filter>...</filter> ... </response-filters> <subdomains> <domain></domain> ... </subdomains> </web-settings> we may expand web-settings even further: <project-settings> <web-settings></web-settings> <environment-settings></environment-settings> <compiler-options></compiler-options> </project-settings> So imho, a root node is something that "perfectly" makes sense. Assigning it a general anonymous name such as root would be at least semantically incorrect. Cheers Volkan.