IBIS question - 100 points in a table

IBIS gurus, another puzzling question:

In IBIS 3.2 spec, the 100 points limit is worded as follows:

| All four columns are required under these keywords. However,
| data is only required in the typical column. If minimum
| and/or maximum current values are not available, the reserved
| word "NA" must be used. "NA" can be used for currents in the
| typical column, but numeric values MUST be specified for the
| first and last voltage points on any I-V table. Each I-V
| table must have at least 2, but not more than 100, voltage
| points.

When doing the best-100 points data-reduction algorithm, the data points that
are included should have the highest density around the areas of inflection in
the curves. However the inflection points for each of the three curves (typ,
min, max) might not be coincident on the x axis.

So the way I read this, if you have reduced the data to the best 100 points for
each curve, the x axis points may not be coincident for y axis data points on
each of the three curves. Therefore you will likely have several rows with data
points in only one of the three columns and NA in the other two.

Example:
| V Ityp Imin Imax
        -5 DATA DATA DATA |The first line must always have data
        -4.9 DATA NA NA |in all three columns.
        ...
        -4.5 NA DATA NA
        ...
        -4.1 NA NA DATA
        ...
        10 DATA DATA DATA |Last line must have data in all 3.

Now the question is in the interpretation of the line "Each I-V table must have
at least 2, but not more than 100, voltage points."

Does this mean that each aggregate table can only have a maximum of 100 rows of
data (100 x axis points)? Or does this mean that each column in the table can
contain a maximum of 100 y axis points?

If the former, then you might end up with less than 33 useable data points in
any one of the three columns. If the latter, then you would have a full 100 data
points in each column, but the aggregate table would almost assuredly contain
more than 100 rows.

Thanks.

-- 
Regards,
Stephen M. Nolan