The problem was caused by the fact that repfactor attribute was not set to "auto" for the newly added indexes in the peers indexes.conf.
repFactor=auto is set automatically for the default indexes when you enable clustering, but new indexes that you wish to replicate have to be configured manually.
See http://docs.splunk.com/Documentation/Splunk/5.0/Indexer/Configurethepeerindexes for full information
Also - customer created indexes don't show up in the clustering dashboard until there is an event in them! This confounded me for hours one day until I decided to send some data to the index, and lo and behold, as soon as there was a hot bucket for the index on one of my indexers, it immediately shows up in the clustering dashboard.
Probably just to do with how the dashboard searches are created. The only way around it is to have a savedsearch csv populating lookup which is then called in the dashboard. This way you can show empty indexes.