I hope this isn't an overly stupid question. 

I read in the docs that it helped to manage the memory swish-e uses at one
time by creating large indexes by indexing in smaller chunks then splicing
together the individual index files. 

What I tried was to create two index files and put them together like
this:

cat index2 >> index1

But a subsequent search using index1 does not seem to turn up stuff that
should be there after index2 was added. 

Did I misread the suggestion? Is there a better way to join index files? 

Thanks for the help!

--Eric Crump
  eric@serv1.ncte.org