case sensitive/insensitive confusion

I am seeing different ORDER BY results 
on a character column on different machines. 
I have (1)
ResyDBE=# select version();                               version
------------------------------------------------------------------------PostgreSQL 7.4.5 on hppa-hp-hpux10.20, compiled
byGCC gcc (GCC) 3.3.1
 

ResyDBE=# \l ResyDBE      List of databases  Name    | Owner | Encoding
-----------+-------+-----------ResyDBE   | rodos | SQL_ASCII

ResyDBE=# select parameter_name from parameter_define where
parameter_name ilike 'c%' order by 1 limit 8;parameter_name
----------------CAENAMCAENAMCAENAMCBERRYCBERRYCCCTOGCCCTOGCCCTOG

and (2)
ResyDBE=# select version();                                     
version                                       
-------------------------------------------------------------------------------------PostgreSQL 7.4.5 on
i686-pc-linux-gnu,compiled by GCC gcc (GCC) 3.3.1
 
(SuSE Linux)

ResyDBE=# \l      List of databases  Name    | Owner | Encoding  
-----------+-------+-----------ResyDBE   | rodos | SQL_ASCII

ResyDBE=# select parameter_name from parameter_define where
parameter_name ilike 'c%' order by 1 limit 8;parameter_name 
----------------c12a        c12b        c12c        c1a         CAENAM      CAENAM      CAENAM      CAENAM      

It seems to me under hpux the sort is done case sensitive, 
as would one expect on SQL_ASCII encoding, whereas 
under linux a case insensitive sort is done. 
I'd like to see the hpux behaviour under linux too. 
Has anybody seen something like this? 
TIA

Regards, Christoph