Based on a combination of the manual at 
http://www.coda.cs.cmu.edu/doc/html/manual-10.html#ss10.6 and the wiki 
walkthrough for dynamically adding a volume replica I think I've made this 
work. Have I done it the right way? I'll post the correct method to the wiki.

On the SCM:
*** Clone the root volume - NOTE: you need to know the volume id _NOT_ the 
name, and despite what it says in the manual you need to supply a name or the 
name will be blank.
# volutil clone <root_vol_id> -n <new_name>
V_BindToServer: binding to host <my_scm>
VolClone: New Volume id = <new_vol_id>
VolClone: New Volume name is <new_name>

*** dump the volume to file
# volutil dump <new_vol_id> <dumpfile_name>

Now copy the dump file to the other server(s)

On the other server(s):
*** restore the volume on the other server(s) - NOTE: you need the -f before 
the filename, unlike in the manual
# volutil restore -f <dumpfile_name> <partition>
V_BindToServer: binding to host <my_non-scm>
.
VolRestore successful, created <another_vol_id>

On the SCM:
Edit the line for the root volume in /vice/db/VRList:
From:
/ 7f000000 1 <root_vol_id> 0 0 0 0 0 0 0 0
To:
/ 7f000000 2 <new_vol_id> <another_vol_id> 0 0 0 0 0 0 0

# bldvldb.sh
# volutil makevrdb /vice/db/VRList

Now restart the client(s)

Read only replication can be checked the same way as for dynamically adding a 
volume replica.