repeat with "anonymous" index

[kessler@xxxxxxxxxxxxx (Billy Kessler)]

I want to use repeat to loop over a large number of filenames
(including some nonexistent ones), bring them in, low-pass
filter the time series, and write them out again. The filenames
look like:

/home/tao2/taodata/alltao/w2s165e.cdf

with similar files at latitudes ranging from 8s to 8n (some missing) 
and the variable I want to filter is called wu_422. (Eventually I'm 
going to want to loop over longitudes as well, but let's just talk
latitudes now). Each file has variables that extend only over time
(and the number of time elements is different for each file).

That is, the data look like:

    1> /home/tao2/taodata/alltao/w2s165e.cdf
 name     title                             I         J         K         L
 WU_422   WIND U (M/S)                     1:1       1:1       1:1       1:4888

I tried:

set mode ignore_error
repeat/j=2:9:1 (use "/home/tao2/taodata/alltao/w`j`s165e.cdf"; let npt=wu_422[l=@ngd]+wu_422[l=@nbd]; save/file="modatp2/w`j`s165e_lp.cdf" wu_422[l=1:`npt`@sbx:121])

Ferret is happy to create those filenames and bring them in, but any attempt
to work with the data results in an error like:

 **ERROR: illegal limits: "NPT" is not in the range J=4
          Axis extremes are J=1

What I really want is an anonymous index that will not be interpreted by
ferret as referring to an index in the data. Any ideas?

Thanks, Billy K