Cells with with out data are described:

Cells that have string information are described:

I want all of the string data but only some of the number data. In some cases normally filled cells
are not filled with number data which upsets my count of right angle brackets ( > ) which I use to
excise garbage.

To complicate matters, the first 8 columns of the table are in one long concatenated line. The other 7
columns are concatenated on a new line (sometimes with a blank line between and sometimes not).

I'm trying to find a more reliable pattern than counting > to determine my position in the table. A
pattern that will not lose my place when occasional additional "garbage" is added or subtracted.

FWIW, I actually don't count. I use a REXX parse template such as:

parse var RawDataLine . '>' . '>' . '>' CallSymbol '<' , /* call option symbol */
. 'td align' , /* "last" */
. 'td align' , /* "change" */
. 'td align' , /* "volume" */
. 'td align' . '' CallBid '<' , /* call bid */
. 'td align' . '' CallAsk '<' , /* call ask */
. 'td align' tCallOpen '' , /* call Open Interest */
. 'td align' . '' StrikePrice '<' , /* strike price */
. /* discard the rest of the line */
/* there may not be an open interest which means there may not be an '' in
the CallOpen column. So, we handle it by putting the whole line in a temp var (tCallOpen above)
and process that var (below) */

if pos('', tCallOpen) = 0 then CallOpenInterest = 0
else parse var tCallOpen . '' CallOpenInterest '.' /* call open interest w/o decimal
digits */

Now that I have laid all of this out for you, Steven, I think my best bet is to forego looking for
some esoteric stable pattern and use the one that must always be there - the