16GB file problem in Gaussian

 Dear All,
recently I met with the problem that the maximum total RWF file length in Gaussian98 is 16GB. (NB: I'm not referring to the maxiumum file length allowed by your OS; such problem is solved by splitting the RWF into several small files). I met the 16GB problem while trying MP2 frequency, MP4 energy, and QCISD energy calculations. Apparently, Gaussian98 cannot deal with more than 16GB disk space, eg when you set MaxDisk=17GB. Looking at the route section, IOp(8/27) ("maximum amount of dik to use" in the four-index transformation) is negative and equal to -1GB (recall that IOp(8/27) is expressed in Words). Here's a small table of my experiments 
 MaxDisk   : 1 2 ... 15  16  17 ... 31 32 33 ...
 IOp(8/27) : 1 2 ... 15 -16 -15 ... -1  0  1 ...
Perusing G98 manuals and source files was useless. After searching in the net, it seems that the 16GB problem is related to Gaussian98 addressing by 32-bit integers. At http://www.ccl.net/ASC/PET/CCM/skeleton/software/gaussian/gaussian-tips.html 
 I found a brief text claiming this problem can be circumvented, at least
 partially:
"The second limit has a less satisfactory workaround. Scratch data stored in the read-write file can be diverted into a separate file with the option 
 %usescr
while this does not evenly balance the usage it does allow specification of 32 GB of storage and depending on the type of calculation more or less complete utilization of this whole amount." 
 However I did not really understand what it means.
Does anyone knows how to overcome the 16GB limit? Has anyone information about this problem? Any little piece of information might be a precious clue to us. 
 I thank you all in advance for any help.
 Bye
 Alessandro
 --
 ******** NEW TELEPHONE NUMBER AND EMAIL ADDRESS ********
 +------------------------------------------------------+
 |  Dr. Alessandro Ponti                                |
 |                                                      |
 | Consiglio Nazionale delle Ricerche                   |
 | Istituto di Scienze e Tecnologie Molecolari          |
 |                                                      |
 | via Camillo Golgi 19  | Tel   +39 02 5031 4280       |
 | I-20133 Milano        | Fax   +39 02 5031 4300       |
 | ITALY                 | email a.ponti.-at-.istm.cnr.it    |
 |                       | www   www.istm.cnr.it/~ponti |
 +------------------------------------------------------+