Variation of load bug

[GARY FORBIS]

OK, The windows 3.1.50 version loads the file 3.0.3 didn't so I had to
go back to the original file and find the minimum case that wouldn't load
in 3.1.50.  I used unix head, tail and cut commands to extract portions of
the big file that won't load under either of the windows versions.  Again,
I'm using an hp m9250f running the 64 bit version of Vista.  I have 4GB
of memory the Processor is  an Intel Core2 Quad CPU Q6700 @2.66GHz.
The attached file won't load.  It is rows 19 and 20 of the original file.
row 19 and row 20 will load on their own as will the first 250 columns
of both rows and the last 250 columns of both rows.
 
I'm getting tired of trying to black box this.  The script counts on the
space at the beginning of the row but doesn't put it back in.  The leading
space or rather the lack of it doesn't seem to have an effect.  I'm near
my limit of capaibilities without actually looking at the code.  I found
dmlread.cc but haven't found load.
 
I tightened up the minimum that failed to the first 347 columns work but
348 do not:
 
octave-3.1.50.exe:29> load x.19.1.20.348
error: load: failed to load matrix constant
error: load: trouble reading ascii file `x.19.1.20.348'
error: load: reading file x.19.1.20.348
octave-3.1.50.exe:29> load x.19.1.20.347
octave-3.1.50.exe:30>
 
Is load written in C?  If so where is it so I can see what's going on
if I compile it on my computer.
 
Thanks. --gary.

x.19.1.20.500.gz
Description: GNU Zip compressed data